Target product profiles for neonatal care devices: systematic development and outcomes with NEST360 and UNICEF

Background Medical devices are critical to providing high-quality, hospital-based newborn care, yet many of these devices are unavailable in low- and middle-income countries (LMIC) and are not designed to be suitable for these settings. Target Product Profiles (TPPs) are often utilised at an early stage in the medical device development process to enable user-defined performance characteristics for a given setting. TPPs can also be applied to assess the profile and match of existing devices for a given context. Methods We developed initial TPPs for 15 newborn product categories for LMIC settings. A Delphi-like process was used to develop the TPPs. Respondents completed an online survey where they scored their level of agreement with each of the proposed performance characteristics for each of the 15 devices. Characteristics with < 75% agreement between respondents were discussed and voted on using Mentimeter™ at an in-person consensus meeting. Findings The TPP online survey was sent to 180 people, of which 103 responded (57%). The majority of respondents were implementers/clinicians (51%, 53/103), with 50% (52/103) from LMIC. Across the 15 TPPs, 403 (60%) of the 668 performance characteristics did not achieve > 75% agreement. Areas of disagreement were voted on by 69 participants at an in-person consensus meeting, with consensus achieved for 648 (97%) performance characteristics. Only 20 (3%) performance characteristics did not achieve consensus, most (15/20) relating to quality management systems. UNICEF published the 15 TPPs in April 2020, accompanied by a report detailing the online survey results and consensus meeting discussion, which has been viewed 7,039 times (as of January 2023). Conclusions These 15 TPPs can inform developers and enable implementers to select neonatal care products for LMIC. Over 2,400 medical devices and diagnostics meeting these TPPs have been installed in 65 hospitals in Nigeria, Tanzania, Kenya, and Malawi through the NEST360 Alliance. Twenty-three medical devices identified and qualified by NEST360 meet nearly all performance characteristics across 11 of the 15 TPPs. Eight of the 23 qualified medical devices are available in the UNICEF Supply Catalogue. Some developers have adjusted their technologies to meet these TPPs. There is potential to adapt the TPP process beyond newborn care. Supplementary Information The online version contains supplementary material available at 10.1186/s12887-023-04342-1.


INTRODUCTION BACKGROUND
Globally, 2.5 million children die in the first month of life and more than half of these deaths are due to conditions that could be prevented or treated with access to simple, affordable interventions [1].
The first 28 days of life -the neonatal period -represent the most vulnerable time for a child's survival. Globally, more children than ever before are being born in facilities and there are well-described, low-cost, evidence-based practices to address neonatal mortality. However, three quarters of neonatal deaths (nearly 2 million) happen in the first week of life when a child is still at or near a health facility [2]. Health interventions are needed that can provide comprehensive neonatal care at facilities to address the major causes of neonatal deaths. Many of these health interventions are known and can be cost-effective. These interventions though may be different from other interventions needed to address broader under-5 deaths [3].
For the first time ever in 2015 the world pledged to end preventable newborn deaths by 2030 (Sustainable Development Goal 3.2) [4]. On current trends, more than 60 countries will miss the Sustainable Development Goal (SDG) target of reducing neonatal mortality to at or below 12 deaths per 1,000 live births by 2030. About half will still not reach the target by 2050. These countries carry about 80 per cent of the burden of neonatal deaths in 2016 [3]. Focused efforts to strengthen the ability of health systems to deliver neonatal care are still needed in sub-Saharan Africa and South East Asia so as to prevent 80 per cent of these deaths [1].
To address neonatal mortality, the World Health Organization (WHO) is working with Ministries of Health and partners to expand quality services for sick and small newborns in the first week of life [5]. Critical to the sustainable implementation of quality facility-based services will be equipping not only people, but facilities with neonatal equipment that is high quality, affordable, robust, and appropriate for comprehensive care delivery in low-resource settings.

v1.2
Globally, the largest contributors to neonatal mortality are preterm birth, intrapartum complications, and infection. Many deaths attributable to these causes are preventable through six categories of care: Most neonatal healthcare technologies that support these pathways of care are designed for high-resource settings and are either unavailable or unsuitable for use in low-resource settings. As a result, providers in low-resource settings lack the tools needed to deliver quality, comprehensive, newborn care.
There is an urgent need for neonatal healthcare technologies that are affordable, rugged, effective, simple to use and maintain, and able to operate from various power supplies.

DEVELOPING TARGET PRODUCT PROFILES
Manufacturers need Target Product Profiles (TPPs) at an early stage in the medical device and diagnostic development process. These TPPs help inform the ideal targets and specifications and align with the needs of end users. TPPs outline the most important performance and operational characteristics as well as pricing. In the TPPs to follow, the term "Minimal" is used to refer to the lowest acceptable output for a characteristic and "Optimal" is used to refer to the ideal target for a characteristic. The Optimal and Minimal characteristics define a range. Products should meet at least all of the Minimal characteristics and preferably as many of the Optimal characteristics as possible. TPPs should also specify the goal to be met (e.g. to initiate treatment), the target population, the level of implementation in the healthcare system and the intended end users.
An initial set of TPPs were developed listing a proposed set of performance and operational characteristics for 16 product categories. The development timeline envisioned in the TPPs was four years, although some commercially available technologies may fit some of the criteria already. For several of the characteristics, only limited evidence was available and further expert advice was sought from additional stakeholders.

DELPHI-LIKE SURVEY PROCESS
To obtain this expert advice and to further develop the TPPs, a Delphi-like process was used to facilitate consensus building among stakeholders. The initial TPPs were sent to a more comprehensive set of stakeholders including clinicians, implementers, representatives from Ministry of Health, advocacy organizations, international agencies, academic and technical researchers and members of industry. In total, 103 stakeholders from 22 countries participated in the TPP development process via survey.
The number of Delphi-like survey respondents is included next to each product category.
• NOTE: Based upon discussion and review throughout the development of these TPPs, the names of the product categories designated above with a * were modified from the time that the original survey was sent. The title of the product category originally included in the Delphi-like survey is included in parenthesis for reference.
Survey respondents were requested to provide a statement on their level of agreement with each of the proposed characteristics for each TPP. Agreement was scored on a Likert scale ranging from 1 to 5 (1=disagree, 2=mostly disagree, 3= neither agree nor disagree, 4=mostly agree, 5=fully agree) with an option to opt out with the selection of "Other -Do not have the expertise to comment". If participants did not agree with the characteristic (i.e., selected 3 or below) they were asked to provide an explanation with comments. Participants who agreed with the statements could also provide comments however were not explicitly asked. In total, over 1,780 comments were reviewed and summarized in this report.

v1.2
For each characteristic in each product category, a percentage agreement was calculated for both the Minimal and Optimal requirements. The percentage agreement was calculated as the ratio of the sum of number of respondents who selected 4 and 5, to the sum of numbers of respondents who gave any score (from 1 to 5 where 5=fully agree, 4=mostly agree, 3=neither agree nor disagree, 2=mostly disagree and 1=disagree). Consensus for the survey characteristics was pre-specified at greater than 50% of respondents providing a score of at least 4 on the Likert scale.
A classic Delphi process requires at least two rounds of survey ahead of an in-person meeting. Initially, two rounds of the survey were planned, but since 50% consensus for most characteristics was reached after the first round survey, a second round survey was not initiated. Survey results are detailed by characteristic in the individual product category sections.
In total, over 180 organizations/individuals were asked to participate in this Delphi-like survey process, of whom 103 (see Appendix A) responded (response rate, 56%). Survey respondents were asked to self-disclose their affiliation. In summary, about half of responders were implementers/clinicians, 15% were from technical agencies/researchers, 5% industry, 5% Ministry of Health Representation, 2% international bodies, 2% advocacy agencies, and the remaining 22% were "Other" which includes distributors, academics, non-profits / NGOs, international bodies and consultants (see summary in Figure 1 below). A breakdown of participation by product category is included in the individual product category sections.  16 TPPs. More specifically, characteristics on which fewer than 75% of the respondents agreed, or on which a distinct subgroup disagreed, were discussed. Consensus Meeting moderators presented the results and comments from characteristics with <75% agreement from the Delphi-like survey, the moderators then solicited additional feedback on each characteristic with <75% agreement from the Consensus Meeting participants, and then a proposed change to the TPP characteristic was discussed amongst Consensus Meeting participants. In some cases, Consensus Meeting participants nearly universally agreed on proposed changes. In other cases, Consensus Meeting participants failed to reach 75% consensus on proposed changes. If consensus was not achieved after two votes on proposed changes, meeting participants agreed to move forward and the disagreement is noted in this report.

Methodology for Mentimeter Voting Results:
Certain proposed changes to TPP characteristics, for which a distinct subgroup disagreed, were anonymously voted on using Mentimeter.com to determine the overall level of agreement and disagreement amongst the Consensus Meeting participants. The Mentimeter Voting Results are presented throughout this report in three distinct categories: v1.2

Page 11
I.
Overall vote -Includes all Consensus Meeting participants who voted on Mentimeter.com. To eliminate the possibility of duplicate votes, all respondents were asked to enter their name (to be viewed only by the report authors) and blank (potentially duplicate votes) were eliminated from the overall vote. II.
Clinicians -Includes all Consensus Meeting participants who voted on Mentimeter.com and who designated themselves as a Clinician on Mentimeter.com. III.
Excluding involvement with product development -Includes all Consensus Meeting participants who voted on Mentimeter.com minus those who indicated on a Declaration of Interest form that they are 'currently or have been involved in the development of a candidate technology or product' specific to the Product Category being voted on.
Of the 133 stakeholders that were invited to the meeting, 69 participants were able to attend. Participants comprised country representatives, stakeholders from technical and funding agencies, researchers, implementers and civil society organizations, and representatives from companies working on newborn care technologies (see Appendix B for the Consensus Meeting Participant List). An overview of the discussion and final consensus achieved is incorporated throughout the sections to follow.
Most characteristics discussed are presented in this report within the individual product categories, however, a few overarching characteristics that applied to all product categories were discussed in unison and are presented together.

OVERARCHING CHARACTERISTICS
The following summarizes the discussion at the Consensus Meeting for the overarching characteristics that appeared in all TPPs.

Target Operator
There was agreement in the room that the Target Operator characteristic for all product categories would be for use in lowand middle-income countries by a wide variety of clinicians, including nurses, clinical officers, and pediatricians.

Optimal: For use in low-and middle-income countries by a wide variety of clinicians, including nurses, clinical officers, and pediatricians Minimal: Same as Optimal
Original Optimal and Minimal: For use in low-and middle-income countries by a wide variety of clinicians, including nurses, clinical officers, and pediatricians.

Target Population
During the discussion, some participants proposed expanding the Target Population characteristic to a wider population beyond neonates. Others, including product developers and clinicians, felt that it was important to specify the patient population by body weight, rather than age (i.e., up to 5 kg) since the gestational age was often difficult to measure. Consensus was achieved in the room for the Optimal characteristic that the product must be useful and effective and validated for neonatal period.

Target Setting
Some participants felt strongly that the Target Setting characteristic should be broadened from "hospitals in low-resource settings" to optimally include "primary care health facilities". Some participants challenged this and noted that personnel in some primary facilities may not have the proper training or resources available. One participant noted that the small and sick babies in most need would likely be transferred to a higher-level referral hospital rather than being treated in the primary health facility. On November 20, a vote was conducted with the results below: Optimal: Health facilities in low-resource settings • Overall Vote: 93% Agree (n = 44) • Clinicians: 94% Agree (n = 32) • Excluding involvement with product development: 94% Agree (n = 32) Minimal: Hospitals in low-resource settings • Overall Vote: 93% Agree (n = 46) • Clinicians: 94% Agree (n = 32) • Excluding involvement with product development: 94% Agree (n = 34) The discussion reconvened on Friday, November 22 as some participants expressed concern about the proposed expansion to expand the optimal Target Setting to encompass all health facilities. The main concern expressed was that certain product categories could be used incorrectly in health facilities without proper staff, training, or infrastructure. Additionally, some participants questioned how verification of technologies would be conducted in health facilities given the wide range staff and infrastructure conditions. Participants also noted that the recently published WHO-UNICEF Technical Specifications and Guidance for Oxygen Therapy Devices [6, p. [10][11][12], provides a general guide for how Oxygen Therapy devices (e.g., oxygen concentrators, flow splitters, CPAP, pulse oximeters, etc.) may be integrated within different levels of the health system. Consensus was achieved that for Glucometer, Flow Splitter, Oxygen Concentrator, Pulse Oximeter, Respiratory Rate Monitor, Suction Pump, and Temperature Monitor the Optimal Target Setting characteristic could include health facilities, however, for the remaining products, both the Optimal and the Minimal characteristic would be for use in hospitals in low-resource settings. Participants voiced universal support that technologies should be manufactured in a quality system even if a specific ISO standard for the device does not yet exist. Given the wide discrepancy in quality management systems, in-country clinicians and procurement agencies and suppliers felt strongly that adherence to international standards, was important. Furthermore, participants noted that compliance with ISO certification is difficult to measure and therefore diminishes the weight of its importance. Product developers explained that requiring the minimum to meet ISO certification could impact the price of the product and potentially limits the approach towards innovation. Some participants emphasized how this characteristic was closely related to the Regulation characteristic requirements. One participant noted that ISO 13485 certification is compulsory for obtaining a CE mark.
On November 20, a vote was conducted with the results below: Minimal: Same as Optimal -ISO 13485:2016 Medical devices -Quality management systems -Requirements for regulatory purposes [80] • Overall Vote: 62% Agree (n = 29) • Clinicians: 56% Agree (n = 18) • Excluding involvement with product development: 55% Agree (n = 20) For Quality Management, it was agreed to proceed with the following requirements for both the Optimal and Minimal characteristic even though 75% consensus was not achieved: ISO 13485:2016 Medical devices -Quality management systems -Requirements for regulatory purposes [80].
On November 22, the group reconvened the discussion on the characteristic Regulation (see below). [80].

Regulation
There was an extensive discussion on regulatory requirements on November 20 which continued November 22. Some participants emphasized how this characteristic was closely related to the Quality Management characteristic requirements. v1.2

Page 14
Product developers emphasized that gaining a CE mark is not indicative of "ability to sell your product" and is often costly and time-consuming which can be restrictive to early stage developers and potentially stifle future innovation. Product developers face many challenges in securing regulatory approval such as the lengthy time cycles, ability to access capital, and access to regulatory experts / consultants. Some participants noted that regulatory approval or CE marking does not necessarily translate to good performance. Participants responded that it is still important for manufacturers to "do the right thing with regard to performance testing" and encouraged more transparency. Additionally, there is a great disconnect in post-market surveillance assistance making it difficult for developers to continue collecting data on follow-up user studies to ensure technologies are delivering as promised. The group determined that there is a great need to further support innovators in these settings.
Another theme emerged highlighting the complexities of regulatory certification and the multitude of options available which can, at times, lead to confusion. The group agreed that there is an opportunity to better harmonize the medical device certification process and acknowledged that the CE mark certification process continues to evolve. In alignment with the theme expressed that regulatory approval does not necessarily translate to good performance, international NGOs explained that they are using their "buying power" to push for greater transparency, especially with public good documents that inform the buyer. However, international agencies and NGOs emphasized the importance of quality control for medical devices and that "we need to hold products in as high-standing as possible". Discussion also ensued on the strength of local regulatory bodies and the importance of local clinical efficacy trials in addition to CE mark to ensure local buy-in.
Some in-country clinicians and international agencies felt that "CE mark only" was enough, while others voiced support for "clearance from at least one stringent regulatory body". In-country clinicians and distributors emphasized that country ministries will often look to "Big 5" for guidance when making purchasing decisions. International NGOs emphasized that manufacturers expanding in international markets will generally secure regulatory approval with broad application. Additionally, international NGOs noted "you will be hard-pressed finding a donor who will procure without approval from a Stringent Regulatory Authority (SRA)".
Optimal: At least one of: CE marking, approved by US FDA or another stringent regulatory body of a founding member of IMDRF (e.g., Japan or Australia or Canada) Minimal: Same as Optimal.

User Manual / Instructions
The group agreed that the primary intent was that a user could read the manual, but there were various opinions on whether the Optimal should include translation into all the relevant UN official languages and at least one national language for the country of intended use. Participants noted that if too many translations were included, this could cause unintended consequences that the manual might become too bulky and burdensome and ultimately not be used. v1.2

Page 15
Optimal: User manual and additional training materials (checklists, videos, guides) in at least one national official language for the country of intended use. Attached to device with labels and markings where possible. Minimal: User manual provided in at least one national official language.
Original Optimal: User manual and additional training materials (checklists, videos, guides) in English and local language. Attached to device with labels and markings where possible. Original Minimal: User manual provided.

Warranty
While most agreed that a 5 year Warranty Period was Optimal, some felt this was not a realistic target. Participants noted that for many medical devices, the WHO-UNICEF Technical Specifications and Guidance for Oxygen Therapy Devices [6, p. 86-145] often require at least 1 year warranty, with 2 to 5 years being the recommendation, depending on the device. One suggestion was to include uptime hours in the warranty language, rather than a period of time. Consensus was achieved on the Optimal characteristic (5 year warranty) via agreement in the room and the Minimal characteristic (1 year warranty) via a vote.
On November 20, a vote was conducted with the results below: Optimal: 5

BROAD THEMES AND CONSIDERATIONS
The following summarizes additional themes that emerged from the Consensus Meeting.

Instrument Pricing
In order to provide a consistent measure of pricing, the ex-works price is included in the TPPs. Participants highlighted that exworks pricing is not a true measure of landed cost and is often vastly understated to what a procurement agent will pay. One participant from an international NGO noted that there is a "minimum 30% mark-up on the ex-works price." The rationale for using the ex-works price is that it is a reliable measure that can be used for consistent comparison across geographies since distributor markups vary by country and geography.

v1.2
Utility Requirements A significant portion of the discussion was devoted to deliberating on how equipment can be designed to work in health facilities with limited electrical infrastructure. Designing the equipment for low-resource conditions often requires back-up batteries which adds to the expense of the technology, as well as the size of the equipment which can pose a challenge in crowded newborn wards. Some participants noted that rather than designing equipment for these facilities with limited electrical infrastructure, to consider whether a broader investment in electrical infrastructure would be a better use of funds. This inherent tradeoff was discussed multiple times when electrical characteristics were discussed.
Additionally, there were a variety of characteristics in the initial survey that related to Utility Requirements (i.e., electricity and power) that varied slightly in title across the TPPs. During the TPP Consensus Meeting, participants agreed that all characteristics relating to Utility Requirements (includes Back-up Battery; Battery Power; Batteries; Voltage; Power Requirement; Maximum Power Consumption; Response During Power Outage; Surge Protection, Electrical Plug) be reviewed and harmonized following the TPP meeting across the product categories. These characteristics have since been reviewed and harmonized into four distinct characteristics (Power Source, Battery, Voltage, and Power Consumption) in the final TPPs.
• Power Source -This defines the desired power source for the device and can be broken down into the following categories: o Mains power -device must be plugged into a mains power source for use o Mains with battery backup -device must be plugged into a mains power source for use, however, in the case of a power failure, the device has a battery backup that can last a specified period of time o Mains with rechargeable battery -device has a rechargeable battery that operates both when the device is charged by a mains power source, or, when the device is plugged in (e.g., a mobile phone) o Battery is disposable and replaceable o No power required (i.e., disposable device) • Battery -This includes the length of time the rechargeable or disposable battery should function • Voltage -This specifies the preferred voltage conversion if the Power Source utilizes Mains Power. Note that for certain technologies (i.e., Bilirubinometer, Glucometer, Hemoglobinometer, pH monitor, and Pulse Oximeter), the Voltage characteristic is included in reference to the rechargeable battery charger requirements. For example, while the Optimal Voltage characteristic is "None" (i.e., no charging is necessary), the Minimal Voltage characteristic should conform to "the voltage and frequency of the purchasing country's local power grid (e.g., 110-120 VAC at 60 Hz or 220-240 VAC at 50 Hz)" to ensure that the charger for the battery is compliant. • Power Consumption -This specifies the maximum Watts of electricity that the device should consume Ideally, all devices should be developed to withstand power surges and voltage spikes.
Note that comments received in the Pre-Meeting survey report highlighted the importance of the correct frequency in electrical plugs. It was noted that a universal adaptor would not safely support the conversion of 60Hz equipment to 50Hz and v1.2 Page 17 that a machine relying on this method could fail in a short period of time (applicable to Oxygen Concentrator, Warming Crib, Radiant Warmer).

RESEARCH QUESTIONS
Throughout the Consensus Meeting discussions, the following research questions were identified: o Syringe Pump: During this discussion, clinicians explained that bulk weight of pumps and footprint of instruments is a challenge and emphasized the importance of stackability and interlocking devices. A research question for product developers was created to further explore how to optimize the stacking of equipment together and the ability to address concerns with the weight of heavy pumps. o Bilirubinometer: One research question that emerged during the pricing discussion was to evaluate the long-term cost effectiveness of a point-of-care Bilirubinometer vs. clinical diagnosis or current standard of care by measuring and evaluating the number of false positives and false negatives based on clinical diagnosis data versus a point-of-care tool.
The proposal was that the outcome of this comparison could be used to justify the value of the point-of-care tool. o Glucometer: The most accessible point-of-care glucometers are designed to be accurate at high glucose ranges for management of adult diabetes; few are intended for use or accurate in the low glucose concentrations seen in hypoglycemic newborns. The group discussed the need to compare and measure the performance of adult glucometers at neonatal-relevant measures vs. neonate specific glucometers. o CPAP: A research question was created to further explore outcomes and effects with and without heated humidification. Some clinicians commented that humidification helps with the avoidance of hypothermia which is becoming increasingly important. These clinicians claimed that it is likely that heated and humidified air is most important for the smallest newborns less than 1-1.25kg. Other clinicians responded that the mortality impact has never been explicitly studied. o CPAP: A research question was created to further explore the impact of reusable accessories. An existing JHPIEGO paper "Infection Prevention and Control -Module 6. Processing Surgical Instruments and Medical Devices" was referenced in providing recommendations on how to develop guidelines on the reprocessing of single-use device [7, p. 77-81]. o Respiratory Rate Monitor: International standards for respiratory rate accuracy do not currently exist. There is therefore a need to define gold standard for respiratory rate accuracy and standardize experimental conditions. Ethical considerations are important in evaluating and validating these standards at upper and lower ranges on neonates. One participant recommended that both SpO2 and respiratory rate accuracy thresholds be based on real clinical data (typical variability). In the Pre-Meeting report survey, one individual commented that given there was not a 'gold standard' measurement for respiratory rate, they specified a reasonable reference standard with human experts and video recordings and specifying an acceptable degree of agreement with that standard, using the 95% Limits of Agreement and the Bland-Altman plot. However, an international NGO responded that using humans as a 'reasonable reference standard' can be troublesome since they can often be inconsistent or incorrect. Furthermore, they noted that "regulators will likely not see [human experts] as a means to validate".  [8,9]. Intravenous (IV) infusions of water, electrolytes and glucose are given to neonates during the first weeks of life to maintain fluid and electrolyte balances and to provide energy for basic metabolic processes [10]. Fluid therapy requires delivery at precise volumes and flow rates, and fluid overload can be lifethreatening [10,11].

SYRINGE PUMP INTRODUCTION: SYRINGE PUMP
Syringe pumps deliver medication and small quantities of fluids continuously through an intravenous line and are a "priority medical device" as described by the World Health Organization. In high-resource hospitals, syringe pumps are used to provide rehydration fluids, breastmilk, dextrose to hypoglycemic infants, and antibiotics to infants with infection. In hospitals where syringe pumps do not exist or are unable to be maintained or operated, these fluids are delivered via a gravity-fed IV drip, slow push by nurses, or using burettes. These are all much less accurate methods of delivery and put infants at significant risk of over/under dosing, medical error, line complications, fluid overload, or hypovolemia. Additionally, since premature babies are likely to need slow introduction to breastmilk over the first week of life, syringe pumps are critical to maintaining normal glucose and hydration until preterm infants can tolerate adequate volumes of breastmilk orally or by nasogastric tube. For these reasons, syringe pumps were listed as a pressing technology for improving newborn care in The Global Action Report on Preterm Birth [12].
The FDA has reported that syringe pumps currently on the market are difficult to use [13]. Moreover, existing syringe pumps are expensive, and require costly, brand-specific consumables, making them unsuitable for use outside of high-resource settings. To be effective in reducing infant mortality on a global scale, pumps must be designed with a simple user interface to avoid setup errors and function accurately with the variety of syringe brand and sizes. In addition to withstanding hot and humid environments, the pump must be easily calibrated and maintained by local technicians. Syringe pumps are often unavailable for infants in need of life-saving IV treatment.

User Instructions
User manual and additional training materials (checklists, videos, guides) in at least one national official language for the country of intended use. Attached to device with labels and markings where possible User manual provided in at least one national official language

Decontamination
Easy to clean with common disinfecting agents 1 There was not 75% voting agreement on the Minimal characteristic. Please refer to the TPP Report discussion for additional detail.

CONSENSUS MEETING SUMMARY: SYRINGE PUMP
To arrive at the final TPP for Syringe Pump (Table 1), we conducted a pre-meeting survey to prioritize the items for discussion at the Consensus Meeting for characteristics that achieved below 75% agreement in the survey results (Table 2). An overview of the discussion at the Consensus Meeting of these characteristics is included below.

JAUNDICE MANAGEMENT
Most neonates, term and preterm, will have elevated levels of unconjugated bilirubin and some amount of jaundice during the first one to two weeks of life due to increased levels of unconjugated bilirubin with transient impaired excretion, which is normal in this age group. This condition is particularly prevalent in preterm babies and, if the levels of unconjugated bilirubin are very high and left untreated, may lead to irreversible neurologic damage known as kernicterus.
Phototherapy treats unconjugated hyperbilirubinemia that exceeds safe levels. These levels are based on day of life and risk factors and typically occur within the first one to two weeks of life.
Treatment with blue light phototherapy is necessary to prevent morbidity and mortality from dangerous levels of neonatal jaundice. The blue light is absorbed by bilirubin, which is then broken down in the blood, allowing the infant to excrete the excess bilirubin before it can accumulate and cause permanent brain damage (kernicterus) or death. Jaundice is preventable and treatable; however, kernicterus is permanent and irreversible, resulting in life-long disability. v1.2

INTRODUCTION: BILIRUBINOMETER
Severe jaundice may not be readily evident to the naked eye until already at dangerously high levels. Additionally, jaundice may not present until several days after birth when an infant has already left the hospital. Thus, early monitoring of bilirubin in at-risk infants is critical in order to prevent severe jaundice, which may result in permanent neurological damage, particularly in premature babies who are at greater risk of death and disability due to jaundice.
All infants should have a laboratory evaluation of serum bilirubin (with result turn around within six hours) both to diagnose jaundice and to guide treatment of infants receiving phototherapy. In low-resource settings though, many facilities do not have the ability to run a blood test, and those that do face many challenges both to run the test and obtain results within a meaningful timeframe.
The ideal solution in a low-resource setting would be a reliable point-of-care test which can test serum bilirubin both before and during phototherapy treatment. Rechargeable battery, >100 tests on a single charge.

Voltage
None.
Model must match the voltage and frequency of the purchasing country's local power grid (e.g., 110-120 VAC at 60 Hz or 220-240 VAC at 50 Hz) 1 There was not 75% voting agreement on the Minimal characteristic. Please refer to the TPP Report discussion for additional detail.

CONSENSUS MEETING SUMMARY: BILIRUBINOMETER
To arrive at the final TPP for Bilirubinometer (Table 3), we conducted a pre-meeting survey to prioritize the items for discussion at the Consensus Meeting for characteristics that achieved below 75% agreement in the survey results (Table 4). An overview of the discussion at the Consensus Meeting of these characteristics is included below.

• Linear Range
o Consensus was achieved in the room (without a Mentimeter vote) for the Minimal characteristic. Clinicians noted that the upper end of the range was more important (since above roughly 25mg/dL will not change behavior) and that 5mg/dL for the lower end of the range was acceptable. From a technical perspective, product developers noted that going above 25mg/dL was relatively easy up to 30mg/dL, especially compared to extending the lower end of the range. Product developers explained that reducing the lower end was more expensive, but 3-4mg/dL detection was reasonable from a manufacturing perspective. o Minimal: 5 -30 mg/dL (85.5 -513 µmol/L) • Accuracy o Consensus was achieved in the room (without a Mentimeter vote) for both the Optimal and Minimal characteristics. Participants noted new proposed CLIA laboratory standards [15]. Clinicians mentioned that central laboratory results take more time in low-resource settings (often a minimum of 24 hours). Since clinicians may rely on quick turnaround point-of-care tests in low-resource settings, clinicians requested better accuracy at higher ends of the range, hence the decision to be more stringent than the proposed CLIA standards for the Optimal characteristic. Clinicians noted that at the high and low ranges though, their behavior for treatment would likely not change. Product developers noted that it is a "big ask" to improve beyond 10% accuracy as a centrifuge and other lab equipment for blood sample testing would be required. o Consensus was achieved in the room (without a Mentimeter vote) that the fewer steps the better and therefore, it was suggested that this characteristic be removed from the TPP since there was variation in measurement of the number of steps.

Result Units
Optimal: mg/dL and mmol/L 92% n = 12 Minimal: Same as Optimal.

64% n = 11
5 comments as summarized below • Theme: Variation in unit defined in guidelines across countries o "We are used to mmol/L but international guidelines use mg/dL" o "mmol/L because our guidelines are written mmol/L" o "Most tables are labeled with both so I think reporting in one or the other is also fine" o "Easy to change" o "Minimal might be. 'mg/dL or mmol/L' set at factory pre-shipment'"

Bilirubinometer v1Page 45
Optimal Minimal • No test (on the market or in development) will be able to meet the minimal currently. $2.00 is more feasible • Expensive. Around $1 may be okay

INTRODUCTION -PHOTOTHERAPY LIGHT
Treatment with blue light phototherapy is necessary to prevent morbidity and mortality for severe cases of neonatal jaundice. The blue light breaks down bilirubin in the blood, allowing the infant to excrete the excess bilirubin before it can accumulate and cause permanent brain damage (kernicterus) or death.
There is a dose dependent response of neonatal hyperbilirubinemia to phototherapy which depends on: (1) Duration of phototherapy; (2) Degree of irradiance given which is dependent on wavelength and type of light used; (3) the amount of body surface area irradiated; and (4) the distance of light from patient (this will vary and is based on manufacturers recommendation but is typically 10-30cm).
Phototherapy lights can also be paired with an irradiance meter so that clinicians can determine if the infant is receiving a therapeutic dose of light. Typically, optimal spectral irradiance is 25 -30microW/cm2/nm, although higher spectral irradiance of 30-35 microW/cm2/nm may be used in more severe cases. If the dose is too low, clinicians may adjust the placement of the infant, the height or output power of the light, or replace burnt out light elements.
There are many types of phototherapy lights and modalities including LED, spotlights, fluorescent blue lights, halogen lights, and phototherapy blankets. LED lights have been shown to be the safest and most efficacious for administering phototherapy, as they give off the least heat and are associated with the lowest risk of hyperthermia and dehydration; although, this sometimes comes at an increased cost [16][17][18]. User manual and additional training materials (checklists, videos, guides) in at least one national official language for the country of intended use. Attached to device with labels and markings where possible User manual provided in at least one national official language Warranty 5 years 1 year 1 There was not 75% voting agreement on the Minimal characteristic. Please refer to the TPP Report discussion for additional detail.

CONSENSUS MEETING SUMMARY: PHOTOTHERAPY LIGHT
To arrive at the final TPP for Phototherapy Light (Table 5), we conducted a pre-meeting survey to prioritize the items for discussion at the Consensus Meeting for characteristics that achieved below 75% agreement in the survey results (Table 6). An overview of the discussion at the Consensus Meeting of these characteristics is included below.

• Effective Treatment Area
o Consensus was achieved in the room (without a Mentimeter vote) for the Optimal characteristic that the Effective Treatment Area would be expanded to measure >2000 cm 2 and the Minimal would be adjusted to >1300 cm 2 . Clinicians emphasized the importance of expanding the Optimal Effective Treatment Area to be equal to the base size of a basinet or incubator at 2000 cm 2 even though some guidelines (e.g., AAP) specify that effective surface area is1800 cm 2 (60 x 30 cm) [70] [71]. Product developers warned against increasing the size for the purpose of using one machine for multiple babies while clinicians acknowledged that in low-resource settings, this often occurred despite knowing that this wasn't the proper use of the device. Clinicians also noted that increasing the Optimal Effective Treatment Area was necessary to accommodate larger babies ("chubby chaps") and movement ("squiggly wigglies"). o Optimal: >2000 cm 2 o Minimal: >1300 cm 2 • Irradiance Meter o Consensus was achieved in the room (without a Mentimeter vote) for the Minimal characteristic to be adjusted and specify that an irradiance meter is available for use but that it is not required to be bundled with every phototherapy light purchase. The concern expressed was that this would add an additional cost to the price of the phototherapy light. Clinicians and product developers agreed that an irradiance meter could be purchased separately (estimated cost between $100 -$300) or one could be shared across the unit. There was a discussion that broader guidelines/toolkits for procurement officers on the minimal infrastructure requirements should be developed so that hospitals who purchase a Phototherapy Light also ensure that an Irradiance Meter is available. The following characteristics were not discussed at the TPP Consensus Meeting explicitly, however, additional comments were received and incorporated into the discussion: • Battery o Participants commented that ideally, a battery back-up should be available internal to the device. Additionally, ideally the device should not be damaged by cycling of power/voltage spikes in the case of a power surge. The Optimal characteristic for Battery includes "Provides battery backup" in response to this point. Attached to device with labels and markings where possible.

95% n = 22
4 comments as summarized below • Theme: Training materials will likely need to be developed separate from the manufacturer

Warranty
Optimal: >5 years 78% n = 23 Minimal: ≥1 year 82% n = 22 10 comments as summarized below • Theme: 5 years too long • Suggested Ranges: 2 years To honor a 5 year warranty, you will have to have strong incountry representation. All an extended warranty is a degree of assurance of the above, and this will come at a cost. Manufactures of concentrators willing to extend a warranty from 2-5 do so at a cost. What might be more useful is that during any procurement, consideration be given to establishing a SLA with an in-country rep. In this case, you can take care of any major PPM requirements, as well as "swap out" in the event of a break-down, and there is no discussion of warranties and no need for spares and an incountry source for consumables.

Instrument Pricing
Optimal: <$400 ex-works 95% n = 20 Minimal: <$1,000 exworks 75% n = 20 9 comments as summarized below • Theme: Geography is extremely price sensitive and even $400 was viewed as the maximum • Optimal is too low and may impact quality of device provided • Current brands are $2,000 (may not be ex-works) • $400-500 maximum

POINT-OF-CARE DIAGNOSTICS
Access to diagnostic laboratories remains a key challenge in low-resource settings [19]. Point-of-care diagnostic tests can therefore enable health-care workers to provide more rapid and effective care [20]. Simple, rapid, and affordable point-of-care tests which require minimal or no electricity, a laboratory, or highly trained staff, are now available and widely used for several common conditions in low-and middle-income countries (LMICs) [21]. These point-of-care tests offer an unprecedented opportunity to reduce inequalities in health, and to help LMICs achieve the health-related Sustainable Development Goals (SDGs) [4,11]. v1.2

INTRODUCTION -GLUCOMETER
Hypoglycemia is a common metabolic problem in newborns and can result in neurologic complications if left untreated. Small and premature newborns are at increased risk for hypoglycemia. Monitoring blood glucose concentration allows clinicians to intervene with supplemental glucose for at-risk infants. Most common point-of-care glucometers are designed to be accurate at high glucose ranges for management of adult diabetes; few are intended for use or accurate in the low glucose concentrations seen in hypoglycemic newborns. FINAL TPP -GLUCOMETER

Intended Use
Quantitative measurement of blood glucose for diagnosis and management of neonatal

Target Operator
For use in low-and middle-income countries by a wide variety of clinicians, including nurses, clinical officers, and pediatricians

Target Population
Neonates (born at any gestational age and require ongoing care)

Target Setting
Hospitals in low-resource settings, but, may be used in health facilities based on country guidelines Hospitals in low-resource settings Accuracy ± 6% across the whole range ± 0.2 mmol/L at 2.5 mmol/L (± 3.6 mg/dL at 45 mg/dL) ± 8% 2 ± 0.2 mmol/L at 3 mmol/L (± 3.6 mg/dL at 54 mg/dL)

Results Format
Quantitative across whole linear range (should be able to switch between mg and mmol) Rechargeable battery, >100 tests on a single charge.

Voltage
None.
Model must match the voltage and frequency of the purchasing country's local power grid (e.g., 110-120 VAC at 60 Hz or 220-240 VAC at 50 Hz) 1 There was not 75% voting agreement on the Minimal characteristic. Please refer to the TPP Report discussion for additional detail. v1.2 2 Source: https://www.westgard.com/2019-clia-changes.htm CLIA proposed changes define Accuracy as ±8%. Current CLIA standard is ± 6 mg/dL or ± 10% (greater). These changes are proposed as of Feb 2019.

CONSENSUS MEETING SUMMARY: GLUCOMETER
To arrive at the final TPP for Glucometer (Table 7), we conducted a pre-meeting survey to prioritize the items for discussion at the Consensus Meeting for characteristics that achieved below 75% agreement in the survey results (Table 8). An overview of the discussion at the Consensus Meeting of these characteristics is included below.
• Results Format o Consensus was achieved in the room (without a Mentimeter vote) for the Minimal characteristic to be the same as Optimal and to add the ability to change between mmol/L and mg/dL in both settings.  Minimal: Same as Optimal.

82% n = 11
4 comments as summarized below • We need this also for neonatal hyperglycemia • Optimal use would not be restricted to neonates • Minimal use can be restricted to neonates/infants • At a minimum, the device could be semi-quantitative and indicate normal -low -severely low

Target Operator
Optimal: For use in low-and middle-income countries by a wide variety of clinicians, including nurses, clinical officers, and pediatricians.

91% n = 11
3 comments as summarized below Minimal: Same as Optimal.

comments
• Minimal: 20 mmol/L would be too low for hyperglycemia; 40 mmol/L would be better v1.2

Optimal Minimal
• Optimal range (if you're trying to pick a device that could be used outside neo unit) I understand 0-900 mg/dL though seems like anything over 500 mg/dL in peds will generally have the same management (don't know about adults) • Minimal range of 0-300 mg/dL for neonates • Do any actually read to 50 mmol/L? • 0-600 mg/dL may be needed

83% n = 12
Minimal: 5% CV 67% n = 9 5 comments as summarized below • up to 2.5mg/dL seems ok, 2% seems too permissive even for Optimal? • Not sure I fully understand but a precision error of 2% seems large when measuring hypoglycemia where small variants can make a significant difference • convert our units

comments as summarized below
• Needs as small amount of blood as possible • Existing glucometers require very little blood

comments as summarized below
• Should be battery operated

comments as summarized below
• Batteries should be rechargeable with electricity • Minimal: does seem like you would need battery power option? • Simple battery device which does not require electricity will be ideal • Was minimal and Optimal reversed here?

Instrument Pricing
Optimal: <$30 ex-works 82% n = 11 Minimal: <$100 exworks 30% n = 10 6 comments as summarized below • Minimal: $100 seems very high • A device that will cost less than what is available in the market will be ideal, the market price of the current price is around $20. • This seems very high for a glucometer • Good glucometers are available for $30 • Minimal needs to be cheaper than 100$. There are good glucometers for $10-20 on the market v1.2

90% n = 10
Minimal: $1.50 per test exworks 33% n = 9 6 comments as summarized below • Current state-of-the art blood glucose strips (e.g. Freestyle Lite or Bayer Contour) are around $1.00, so $1.50 seems too much • $0.2 may be reasonable • Minimal: current tests cost $1 or 100 KES • The strip cost is more than the machine cost within six months • $1.50 seems high per test • The price of the glucometer itself is not so important as the cost of the strips, which can be prohibitive. Also major barrier to use is the incompatibility of many glucometer strips between different brand machines. Would be hugely beneficial to have generic strips to use on different glucometer machines v1.2

INTRODUCTION: HEMOGLOBINOMETER
Hemoglobin concentration refers to the amount of the oxygen-carrying protein in the blood, and is a diagnostic for anemia (low hemoglobin) or polycythemia (high hemoglobin).

Intended Use
Quantitative determination of hemoglobin in capillary, venous, or arterial whole blood

Target Operator
For use in low-and middle-income countries by a wide variety of clinicians, including nurses, clinical officers, and pediatricians

Target Population
Neonates (born at any gestational age and require ongoing care)

Target Setting
Hospitals in low-resource settings

Regulation
At least one of: CE marking, approved by US FDA or another stringent regulatory body of a founding member of IMDRF (e.g., Japan or Australia or Canada) TECHNICAL CHARACTERISTICS Linear Range 0-25 g/dL 4.5-25 g/dL.

Results Format
Quantitative across whole linear range Quantitative; semi quantitative below 5 or above 25 g/dL v1.2

Number of Steps
No more than 1-3 steps (requiring operator intervention) No more than 4-6 steps (requiring operator intervention)  (Table 9), we conducted a pre-meeting survey to prioritize the items for discussion at the Consensus Meeting for characteristics that achieved below 75% agreement in the survey results (Table 10). An overview of the discussion at the Consensus Meeting of these characteristics is included below.

• Linear Range
o Consensus was achieved in the room (without a Mentimeter vote) for the Optimal and Minimal characteristic. Clinicians noted that 25 g/dL was appropriate for the upper range. For the lower range, clinicians were comfortable with a reading that accurately goes down to 4.5. One participant commented that the lowest reported levels of hemoglobin concentrations measured in blood was 0.6 g/dL [22]. Product developers noted that from a technical perspective, the incremental price to adjust the measurement range is dependent on the type of test. For example, it can be more challenging to get a wider range with a non-invasive test. Furthermore, participants commented that much less expensive tests can go down to 4-5 g/dL while more expensive tests are 0-25 g/dL.
o Consensus was achieved in the room on the Minimal Instrument Pricing. Participants commented that less expensive tests currently exist for $100-$200, however, there is a wide range with more expensive ones at $800-$900 in price. Participants expressed concern over signaling the market with too high of a price and a vote was conducted in the room where the Minimal was agreed at $300. o Minimal: Ex-works Instrument Price of $300 vs. $400  Overall Vote -73% voted "$300" (n = 11)  Clinicians -80% voted "$300" (n = 10)  Excluding involvement with product development -73% voted "$300" (n = 11) DELPHI-LIKE SURVEY: HEMOGLOBINOMETER

Intended Use
Optimal: Quantitative determination of hemoglobin in capillary, venous, or arterial whole blood.

83% n = 6
2 comments as summarized below • Optimally this would also give WBC and a neutrophil % to risk stratify for sepsis • Is this only measuring Hb? Not that it's bourne out to be great (but it's certainly better than nothing) it seems like we would also want this instrument to get WBC and maybe a neutrophil count? Unless you're on envisioning the utility of this as a rapid diagnostic for anemia NOT generalizable to use in sepsis (where rapid Hb assessment to determine need for transfusion is also important) • Minimal would be capillary whole blood

Target Operator
Optimal: For use in low-and middle-income countries by a wide variety of clinicians, including nurses, clinical officers, and pediatricians.

86% n = 7
Minimal: Same as Optimal 100% n = 6 2 comments as summarized below • With non-invasive it is no longer necessary for trained phlebotomists to take measurements • Optimal would be usable by community health workers as well

80% n = 5
4 comments as summarized below Minimal: Same as Optimal.

67% n = 6
2 comments as summarized below • Why is the range so high? 0-20 or even 2-20 seems more meaningful? • I question anything above 17 as necessary

comments as summarized below
• Range of 3.5g/dL seems high to me?
• Way too strict, propose to update Optimal 7% and Minimal to 15% • CLIA standards are 7%

Results Format
Optimal: Quantitative across whole linear range

100% n = 6
Minimal: Quantitative; semi quantitative below 5 or above 25 g/dL 80% n = 5 2 comments as summarized below • I'd change minimal to "above 20" • For neonates the transfusion threshold would be higher than 5 g/dL so that threshold seems too low in that age group (would be closer to 7.5-8.5, or even higher in the first week of life). Even for older children a higher value around 7 might be more appropriate

Result Units
Optimal: g/dL OR g/L 100% n = 7 Minimal: Same as Optimal.  pH is an important blood gas measurement that assesses the acid-base status of the blood. pH can be assessed on arterial cord blood as well as peripheral arterial, venous, and capillary blood and, when interpreted with other tests and clinical conditions, provide information on the status of the neonate. Although clinically relevant pH values vary by condition, postnatal age (in minutes/hours), and type of blood sample (i.e., venous, arterial, etc.), pH values below 7.4 can indicate acidosis, which can be either metabolic, respiratory, or mixed. In the newborn setting, blood gas analysis is typically employed in an intensive care setting and can be utilized to augment management of invasive and non-invasive positive pressure respiratory support, sepsis, and perinatal asphyxia. To differentiate between the different types of acidosis, it is necessary to measure not only pH but also pCO2, pO2, and base excess.  (Table 11), we conducted a pre-meeting survey to prioritize the items for discussion at the Consensus Meeting for characteristics that achieved below 75% agreement in the survey results (Table 12). Given restrictions on timing, we were not able to discuss any of the characteristics for pH Monitor at the Consensus Meeting. Please note that the number of participants in the pre-meeting survey is low.   [23]. The most important protective interventions for nosocomial infections are frequent hand-washing, exclusive breastfeeding and facility cleanliness [24,25], but widespread implementation of these interventions is challenging in lowresource settings. Infants (and their mothers) who are malnourished or have a chronic illness are at risk of infection because of immunosuppression and a susceptibility to preterm birth [26]. Premature infants have an increased risk of infection, regardless of the mother's antibody status [11,12].

INTRODUCTION: SEPSIS DIAGNOSTIC
Neonatal sepsis is a major cause of newborn mortality and must be identified and treated quickly to ensure survival and minimize morbidity. However, it is not easy to diagnose. Due to the immaturity of a neonatal immune systems, natural history of late deterioration, and high morbidity in the presence of a serious bacterial infection, the standard of care in neonates is to treat while simultaneously screening for sepsis with blood, urine, and spinal fluid cultures and microscopy until studies suggest that infection is unlikely to be present. There are some useful guidelines that help to identify neonates and young infants at risk of sepsis and guide clinical management. However, even when these guidelines are used, many more babies receive antibiotics than those who truly have serious bacterial infections and need antibiotics [27].
Serious bacterial infections can be identified by clinical assessment, biochemically (with biomarkers), or microbiologically. However, limited availability of microbiological diagnostic testing in low and middle income countries (LMIC) is a major barrier to safe antibiotic use and shortening courses of treatment. The currently available diagnostic tests have significant barriers in their use and interpretation [28]. Additionally, there is currently no accepted biomarker for use in low-and middle-income countries [29]. The availability, cost, rapidity of results, sensitivity, specificity, predictive value, and the interpretation of results pose challenges for the widespread use of biomarkers. Small studies have described hundreds of biomarkers associated with severe neonatal infections and biomarkers, alone or in combination, that have been used to identify newborn infections: procalcitonin (PCT), C-reactive protein (CRP), tumor necrosis factor-a (TNF-a), interferon-g (IFN-g), interleukin-6 (IL-6), interleukin (IL-8) [30]. The majority of these studies have evaluated biomarkers in combination with C-reactive protein (CRP), already in widespread clinical use for the diagnosis of infection. As an acute-phase reactant, CRP alone is less useful in the earliest phases of severe neonatal infection because it does not peak until 12 to 24 hours after infection and can also be triggered by a non-infectious insult, such as trauma.

CONSENSUS MEETING SUMMARY: SEPSIS DIAGNOSTIC
It was clear in the time available for group discussion at the Consensus meeting that further analysis and consultation will be required to formulate a TPP, particularly to delineate the practicality and clinical impact of each use case. Give this, a data-based analysis of potential use cases and further survey process are planned.

Developing a Target Product Profile for a Neonatal Sepsis Point-of-Care Test: Next Steps
The process of developing a TPP for a point-of-care test (POCT) for neonatal possible serious bacterial presented distinct challenges. This is largely because point-of-care testing for neonatal infections is not a currently used diagnostic strategy in clinical practice, in both low-, middle-, and high-income settings. There is thus no similar technology routinely used from which the basics of development considerations and implementation measures can be used as learning points for target product profile development specific for wider use across other settings.
To begin the process of developing a TPP, we worked in partnership with Dr. Naomi Spotswood at the Burnet Institute and Dr. David Goldfarb from the University of British Columbia to develop Use Cases for potential Sepsis Diagnostics. In the first stage, six potential use cases were developed to describe the more likely clinical scenarios where a POCT for neonatal possible serious infections might be used. The first four of these were for scenarios to assist healthcare workers to decide if antimicrobials should start, the fifth to decide if antimicrobials should stop, and the sixth to identify infections with antimicrobial resistant pathogens. Initial discussion in the consensus meeting focussed on whether the first four use cases (starting antimicrobials) could be condensed into one use case. While collapsing use cases one to four into a single use case may be simpler conceptually, it was noted that each use case would have different microbiology, immunology and epidemiology, each of which will affect the pre-test probability of infection in the target population. Further, clinical thresholds for starting antimicrobials for the same use case may differ between settings. Overall it was agreed that reducing unnecessary antimicrobial use would be a key attribute of a neonatal sepsis POCT.
Moving forward, relevant questions are below. Each would ideally be estimated for the setting of interest. 1(c) and 2(c) require pre-defined target sensitivity and specificity: • Conduct an analysis of currently available data to provide estimates for the above questions. This will allow clearer evaluation of the potential clinical impact of a POCT for each use case. • Formulate and distribute an extended survey to finalise the TPP for a neonatal sepsis POCT. This is planned to reach beyond the original group: the WHO possible Serious Bacterial Infections Community of Practice group and Medicins Sans Frontiers have been identified as examples of groups to contact given their practical knowledge and experience relevant to this process. • Given the wide relevance of a TPP for a neonatal sepsis point-of-care test, the group will consider publication of the TPP development process and final results in a peer reviewed journal.

USE CASE SURVEY: SEPSIS DIAGNOSTIC
Sepsis (serious infection) in neonates and young infants is devastating for many babies and their families around the world. It is also not easy to diagnose. There are some useful guidelines that help to identify neonates and young infants at risk of sepsis, and guide clinical management. However, even when these guidelines are used, many more babies receive antibiotics than those who truly have serious bacterial infections which need antibiotics [31].
Researchers around the world are trying to develop a point-of-care test for sepsis. This is a test that can be done by any healthcare worker with a quick result. However, a point-of-care test for sepsis could be used in a number of ways, and it is important that researchers know which way (a 'Use Case') will be most helpful to healthcare workers. The following six 'Use Cases' were presented in a survey. The purpose of the survey was to evaluate which of these 'Use Cases' would be of most practical benefit to clinicians who manage neonates with possible serious bacterial infections. The aim is that a test like this would be used in combination with existing guidelines provided by the World Health Organization [27].

Use Case 1. Start Antibiotics -Community Referral:
A test that can be used when a baby first comes to a health facility from the community for assessment, and has one or more signs of possible serious bacterial infection. Examples of these include respiratory rate >60 breaths per minute, being unable to breastfeed, or deep jaundice. The test is to help the healthcare worker decide if they should start antibiotics. If the test is positive, this means that the baby is likely to have a serious bacterial infection. The baby needs antibiotics and supportive care. If a blood culture can be sent, this should be collected before the antibiotics are started. If the test is negative, this means the baby is highly unlikely to have a serious bacterial infection. Instead they need careful observation, and the healthcare worker should consider other reasons for their illness.

Use Case 2. Start Antibiotics -Well Baby with Risk Factors at Birth:
A test that can be used when an otherwise well baby has been born with risk factors for sepsis. Examples of these risk factors are fever in the mother during labour, prolonged v1.2 rupture of the membranes (>18 hours), or foul-smelling amniotic fluid. Other non-maternal risk factors might include preterm labour. The test is to help the healthcare worker decide if they should start antibiotics. If the test is positive, this means that the baby is likely to have a serious bacterial infection. The baby needs antibiotics and supportive care. If a blood culture can be sent, this should be collected before the antibiotics are started. If the test is negative, this means the baby is highly unlikely to have a serious bacterial infection. The baby would stay with mother and receive normal newborn care.

Use Case 3. Start Antibiotics -Unwell at Birth:
A test that can be used when a baby has been born with signs of sepsis with or without maternal risk factors. Signs of sepsis include tachypnea, temperature instability, or tachycardia. The test is to help the healthcare worker decide if they should start antibiotics. If the test is positive, this means the baby is likely to have a serious bacterial infection. The baby needs antibiotics and supportive care. If a blood culture can be sent, the sample should be collected before the antibiotics are started. If the test is negative, this means the baby is highly unlikely to have a serious bacterial infection. If the baby remains unwell, they need careful observation, and the healthcare worker should consider other reasons for their illness.

Use Case 4. Start Antibiotics -Small or Premature Baby who becomes Unwell:
A test that can be used for a baby who is already admitted to a health facility because they are small or premature who becomes unwell and has one or more signs of a possible serious bacterial infection. The test is to help the healthcare worker decide if they should start antibiotics. If the test is positive, this means that the baby is likely to have a serious bacterial infection. The baby needs antibiotics and supportive care. If a blood culture can be sent, this should be collected before the antibiotics are started. If the test is negative, this means the baby is highly unlikely to have a serious bacterial infection. Instead they need careful observation, and the healthcare worker should consider other reasons for their illness.

Use Case 5. Stop Antibiotics:
A test that can be used for a baby who is already admitted to a health facility and who has already received at least one day of antibiotics for a possible serious bacterial infection. The test is to help the healthcare worker decide if the antibiotics can stop. If the test is positive this means that the baby is likely to have a serious bacterial infection. The baby needs to continue their antibiotics. If there are positive blood or cerebrospinal fluid culture results, the antibiotics may need to change to make sure they are the best antibiotic to treat the infection that has been identified. If the test is negative, this means that the baby is highly unlikely to have a serious bacterial infection. A survey with the six use cases was completed by 33 respondents (see Figures 13 and 14). Respondents were asked questions to prioritize and rank the use cases (see Figure 15). Based on the results presented below, use case 1 and 5 received the highest score despite a wide range (see Table 13 and 14).  1 Prioritization takes the average weight assigned to each use case based on the sliding scale. Note that the respondent could assign every use case at the maximum 100 (i.e., no force rank or sum total). 2 Rank takes the average of each assigned rank by use case per submission. 1. "Standards around neonates is that the majority of the time we are treating when infection is highly improbable. Hard to imagine something that can replace cultures" 2. "Use Case 3 + would ideally say which antibiotic to start; Use Case 6 should tell you which bacteria is resistant; all of these tests would depend on sensitivity or specificity" 3. "A) Umbilical cord dx -I believe this is often discarded but several studies have studied biomarkers in cord blood and seen promising results. B) I'm not sure how realistic this would be, but a diagnostic for resistance at the time of diagnosis of sepsis could help guide treatment in one visit. I know the mortality rate of neonatal sepsis is very high, and I wonder if that means that use cases 1-4 should be prioritized over use case 5/6. In my mind, a baby who is unwell (use cases 3/4) will be started on antibiotics anyways so I had those at a lower priority, but of course there are issues of resistance and misdiagnosis there too. I believe a large burden of neonatal mortality occurs soon after birth, which was my justification for putting use case 2 at the highest priority" 4. "If the test is low-cost and simple to use by community health workers, then it could be used during community outreach activities to identify a patient at the community level and refer to the nearest health facility for Abx initiation. This is similar to case #1, but starts from the community level for early identification at community/household level -> referral -> and early/immediate initiation." 5. "A baby who had other problems at admission and becomes unwell after admission (diagnosing hospital-acquired sepsis)" 6. "Treatment response: a use case that enables clinicians to non-clinically monitor response to treatment for diagnosed septic neonates. This use case could herald possible antibiotic resistance and rationalize / prioritize blood culture usage" 7. "Test to guide other intervention (e.g. supporting, referral to a higher level center)" 8. "Surgical patients those who have gone through major operations and some patients with Gastroschisis, open spina bifida. Many of these later develop signs and symptoms of sepsis. Some babies are delivered at home under unsterile procedure. Need proper tests to guide on use of antibiotics" 9. "Hospitalized premature infant with respiratory worsening (increase in ventilation or oxygenation needs)" 10. "If there are signs of infection healthcare workers will start antibiotics -hence less useful. However a more pertinent question would be what antibiotics to start -if the diagnostics could identify the bacteria that would be extremely helpful in all cases" v1.2  1. fluid in the lungs must be absorbed and replaced with air, 2. lungs must expand fully and regular breathing must be established, and 3. pulmonary blood flow is increased.
When these three things do not happen, a baby will have respiratory distress. Respiratory distress syndrome (RDS) is when there is deficiency of surfactant that is needed to prevent alveolar collapse; this is especially common in premature newborns.
Oxygen provision is important in the care of newborn infants because many conditions that affect babies in the first days of life can result in low levels of oxygen in the body. Hypoxemia, or low levels of oxygen in the blood, is a life-threatening condition that results in increased mortality and morbidity. Prematurity and respiratory distress syndrome (surfactant deficiency), pneumonia and other severe infections, asphyxia, and difficulties in the transition from fetal to neonatal life can all result in hypoxemia. Yet, despite its importance in acute severe illnesses, hypoxemia is often not well recognized or managed in settings where resources are limited. It is therefore important for health workers to know the clinical signs that suggest the presence of hypoxemia and how supplemental oxygen can appropriately be used as an essential lifesaving treatment [32]. In high-resource settings, a mother is given steroids before birth if a baby is anticipated to be born preterm to help prevent respiratory distress syndrome (RDS). If RDS still occurs, assisted breathing with continuous positive airway pressure (CPAP) is started. If CPAP is not sufficient, intubation, surfactant and/or ventilation may be needed.
In low-resource settings, many facilities lack the resources to implement CPAP. While many companies make newborn CPAP devices, only a few key players design their devices to work in low-resource settings.
Bubble Continuous Positive Airway Pressure (bCPAP) therapy is a common mode of treatment for RDS in premature neonates and for respiratory illness in young children. bCPAP provides a continuous flow of pressurized air into the patient's nostrils via nasal prongs or a mask; this pressure prevents alveolar collapse during exhalation. In high-income settings, early bCPAP is now preferred over mechanical ventilation as first line therapy for respiratory distress syndrome in preterm infants. bCPAP has been shown to promote production of endogenous surfactant [33] as well as dramatically decrease progression to intubation or death in both high [34][35][36] and low [37,38] income settings.
In low-resource settings, there is a need for CPAP that is designed for patients who weigh between 1 and 10 kg and that includes an oxygen blender which allows users to provide 21-90% oxygen to the patient when an external oxygen source is connected to the CPAP. The CPAP should ideally contain an integrated air-compressor, blender, and patient interface. Although there are short cuts for delivering positive airway pressure to a baby without an appropriate device, these generally rely on pure oxygen sources from oxygen cylinders or concentrators. Procurement officers should consider current evidence, target level of care, provision, and context when choosing between available CPAP devices. The ability of a CPAP device to deliver positive pressure at low fractional inspired oxygen concentrations (FiO2) is a critical feature for preventing retinopathy of prematurity and chronic lung disease associated with oxygen administration [39,40]. Some CPAP units use heated and humidified gas in the circuit, although the exact benefits of humidification in non-invasive ventilation (i.e. CPAP) in terms of survival, complications from therapy and morbidity are not well established. Humidification, while a feature of some CPAP devices, remains a controversial feature of CPAP in low-resource settings, especially for CPAP devices utilizing compressed ambient air rather than gas cylinder sources.

Voltage
Model must match the voltage and frequency of the purchasing country's local power grid (e.g., 110-120 VAC at 60 Hz or 220-240 VAC at 50 Hz)

User Instructions
User manual and additional training materials (checklists, videos, guides) in at least one national official language for the country of intended use. Attached to device with labels and markings where possible User manual provided in at least one national official language Warranty 5 years 1 year 1 There was not 75% voting agreement on the Minimal characteristic. Please refer to the TPP Report discussion for additional detail. 2 There was not 75% voting agreement on this characteristic. Please refer to the TPP Report discussion for additional detail. 3 There was not 75% voting agreement on this characteristic. Please refer to the TPP Report discussion for additional detail. 4 There was not 75% voting agreement on this characteristic. Please refer to the TPP Report discussion for additional detail.

CONSENSUS MEETING SUMMARY: CPAP
To arrive at the final TPP for CPAP (Table 15), we conducted a pre-meeting Delphi-like survey. Based on the pre-meeting Delphi-like survey results (Table 16), characteristics that achieved below 75% agreement were prioritized for discussion at the Consensus Meeting. An overview of the discussion is included below.  Decreased upper airway mucosal injury  Decreased convective heat losses which may lead to hypothermia and more challenging weight gain in infants  Decreased lung inflammation from aspirated secretions which has unknown impact on morbidity and mortality of very low birthweight infants. o Some potential drawback to heated humidification include:  Iatrogenic infection, especially in settings where clean water may not be readily available and humidifiers, which are typically meant for one time use, are being cleaned and re-used between patients  High financial cost of adding heated humidified gas  High cost of additional consumable required and ongoing maintenance  High human resource costs in terms of repair and preparation of non-invasive ventilation units which may limit not only their use, but availability of this life saving technology within our setting o Clinicians commented that humidification helps with the avoidance of hypothermia which is becoming increasingly important. These clinicians claimed that it is likely that heated and humidified air is most important for the smallest newborns less than 1-1.25kg. Other clinicians responded that the mortality impact has never been explicitly studied. o A research question was created to further explore outcomes and effects with and without heated humidification.  Potential benefits of heating and humidification could include: • Increased comfort and adherence • Decreased upper airway mucosal injury • Decreased convective heat losses which may lead to hypothermia and more challenging weight gain in infants • Decreased lung inflammation from aspirated secretions which has unknown impact on morbidity and mortality of very low birthweight infants.

Optimal Minimal
• Iatrogenic infection, especially in settings where clean water may not be readily available and humidifiers, which are typically meant for one time use, are being cleaned and re-used between patients • High financial cost of adding heated humidified gas • High human resource costs in terms of repair and preparation of non-invasive ventilation units which may limit not only their use, but availability of this life saving technology within our setting It is likely that heated and humidified air is most important for the smallest newborns less than 1-1.25kg although this has never been explicitly studied.

77% n = 39
12 comments as summarized below • A variety of hard and soft copy materials mentioned with particular mentions of difficulty in reading a user manual and preference for videos so people can see vs. read • All claims must be filed with the regulatory dossier, so this is not as straight forward as a simple translation. Appropriate, professional translations are a must and are costly to the manufacturer. Additionally, local language varies greatly across a country and is often-times not even the official language of the country and so this may not be a reasonable ask of manufacturers 19 comments as summarized below • Extremely price-sensitive geography and even $10 was viewed as too expensive by some respondents, especially for countries where patient pays out of pocket for consumables (e.g. Nigeria) • Ex-works not likely a true measure of landed costs • If consumables were reusable, then price point slightly higher than $10 is more realistic • "$10 is too low for effective circuits" v1.2 When using an oxygen concentrator or oxygen with neonates, low flow is critical in order to avoid preventable disability including retinopathy of prematurity (ROP) and chronic lung disease. A significant number of preventable childhood blindness due to ROP in low-and middle-income countries (LMIC) has been documented [42,43]. Importantly, this is observed in children at higher birthweights and gestational ages than children in high-income settings, suggesting an association with rapid expansion of neonatal care, perhaps without adequate attention to the quality of care or harms of oxygen administration. Neonatal units seeking to provide comprehensive care should consider the procurement of splitters and flow meters with precision adjustment at a minimum of 0.1 -0.125 L/min. As health facilities advance, introduction of microcalibrated flow meters with precision finer than 0.1 L/min or oxygen blenders should be considered [44].

Indication
Each flow rate has a visual indicator

Maintenance
No/minimal maintenance 1 There was not 75% voting agreement on the Minimal characteristic. Please refer to the TPP Report discussion for additional detail.

CONSENSUS MEETING SUMMARY: FLOW SPLITTER
To arrive at the final TPP for Flow Splitter (Table 17), we conducted a pre-meeting survey to prioritize the items for discussion at the Consensus Meeting for characteristics that achieved below 75% agreement in the survey results (Table 18). An overview of the discussion at the Consensus Meeting of these characteristics is included below.

• Instrument Pricing
o There was disagreement on the Minimal characteristic for instrument pricing as it was dependent on the number of splitters included in the device. Participants noted that there is a wide range of commercial products available ranging in price from $80 -$600. Accuracy implications remain a key concern for neonatal use. Product developers noted that ISO and CE Mark certification will require that Flow Splitter covers 30-40% accuracy, however, this may increase the price to the $600 mark with 5 ranges included. Therefore, a tradeoff exists in the current market v1.2 whereby a cost reduction would be at the expense of accuracy. One basic work-around discussed at the hospital level was to utilize an oxygen monitor which can cost around $150 but may be used for multiple use-cases. o Minimal: <$600 ex-works  Overall Vote -82% Agree (n = 22)  Clinicians -79% Agree (n = 14)  Excluding involvement with product development -82% Agree (n = 22) DELPHI-LIKE SURVEY: FLOW SPLITTER

94% n = 16
2 comments as summarized below • Recommended for neonatal and low flow oxygen as per interagency oxygen therapy guide • Preference for low-pressure piping and a separate flow meter beside each bed rather than a flow splitter and having the flow meters far from the patients

Target Operator
Optimal: For use in low-and middle-income countries by a wide variety of clinicians, including nurses, clinical officers, and pediatricians.

93% n = 15
4 comments as summarized below  For newborns with breathing difficulties and/or infections, oxygen is vital to survival. Yet, access to oxygen can be scarce in lowresource settings. To meet this need, an oxygen concentrator is a device able to concentrate oxygen from the air for use with a multitude of devices. While use of concentrators is helpful, facilities should always have a power-independent oxygen source, such as a cylinder, available for back up.
Oxygen concentrators typically output oxygen between 85-100% FiO2, with flows between 2-10 LPM with typically one or two outlets. The percent oxygen a patient will receive depends on each mode of delivery (i.e., nasal prongs, nasal catheter, facemask, etc.). Passive humidification is sometimes available but recommended against by the World Health Organization [32]. A flow splitter allows the output of a concentrator to be split between multiple patients while independently monitoring and adjusting the flow rate to each. It is important to consider that high flow oxygen concentrators should be paired with an appropriate flow splitter for the safety of the neonate.

Intended Use
To provide medical oxygen for use in a healthcare setting

Target Operator For use in low-and middle-income countries by a wide variety of clinicians, including nurses, clinical officers, and pediatricians Target Population
Neonates (born at any gestational age and require ongoing care)

Target Setting
Hospitals in low-resource settings, but, may be used in health facilities based on country guidelines Hospitals in low-resource settings

Mains Power Mains Power
Power Consumption <275W at 5 LPM Scales with delivery output -i.e., consumes less power at lower flow rates

Voltage
Model must match the voltage and frequency of the purchasing country's local power grid (e.g., 110-120 VAC at 60 Hz or 220-240 VAC at 50 Hz)

User Instructions
User manual and additional training materials (checklists, videos, guides) in at least one national official language for the country of intended use. Attached to device with labels and markings where possible User manual provided in at least one national official language

User Skill Level
Minimal to none  (Table 19), we first leveraged the extensive work conducted by PATH in the "Design for reliability: Ideal product requirement specifications for oxygen concentrators for children with hypoxemia in low-resource settings" [45]. We conducted a pre-meeting survey to prioritize the items within this existing TPP to discuss at the Consensus Meeting. Specifically, characteristics that achieved below 75% agreement in the survey results (Table 20). An overview of the discussion at the Consensus Meeting of these characteristics is included below.

• Flow Meter
o For the Optimal characteristic, rather than specifying the flow should be split evenly at 0-5 LPM (Liters Per Minute) in each of the two meters, the range should be 0-10 LPM with the ability to split however the user wants across the two outputs. For the Minimal characteristic, clinicians noted that a flow meter that goes to a minimum of 0.25 increments would be beneficial since 0.5 LPM can even be high for neonates. Product developers noted that from a technical perspective, an easy range is 0-10 with 5% resolution, but that there would be inaccuracy at the lower bound and therefore, would recommend 0-1 graduations. It was noted that a flow splitter paired with an oxygen concentrator would suit requirements at low flow rates and therefore, a flow splitter should always be available with an oxygen concentrator. International agencies noted that: "Ideal setup would be to have a concentrator connected to a 5-way flow splitter, with those flowmeters ranging from 0-2 LPM, with increments of 0.25 LPM or less. In other words, if the optimal requirement of 2 flowmeters is to be able to service two neonatal patients at o Consensus was achieved that the sound level characteristic was referring the operating noise level. Product developers noted that from a technical standpoint, CE mark requires that this be under 50 decibels for operating noise [46]. Consensus was achieved in the room (without a Mentimeter vote) that the "lower the decibel level, the better" and that Optimal and Minimal should be the same. The spirit of the conversation emphasized that the noise levels should be as low as possible to protect the babies hearing. o There was disagreement for both the Optimal and Minimal preventive maintenance interval characteristics. The discussion highlighted the importance of cost effectiveness and the risk associated with too frequent maintenance intervals given most hospitals have annual preventive maintenance processes. One idea discussed was creating a device that measures oxygen levels and once it drops below a certain level, would flag that maintenance is required. Product developers noted that manufacturers claim 30,000 hours (roughly 3 years) with regular maintenance, but often the true maintenance frequency may vary based on the wide range of operating conditions (i.e., may require more or less maintenance). One suggestion in the Pre-Meeting survey comments was to "measure oxygen concentration with a calibrated oxygen analyzer" to which another participant clarified that "not all analyzers need to be calibrated (e.g. those with ultrasonic sensors)". o Optimal: Should not need preventive maintenance more than once a year  Overall Vote -83% Agree (n = 23)  Clinicians -79% Agree (n = 14)  Excluding involvement with product development -83% Agree (n = 23) o Minimal: Should not need preventive maintenance more than 4

• Replacement Parts and Consumables
o Given the discussion on Preventive Maintenance Interval highlighted above, participants noted that this characteristic was too detailed and proposed removing from the final TPP as it would be more applicable to a specification. In light of this, further information on the extensive list of replacement parts recommended in the WHO-UNICEF Technical Specifications and Guidance for Oxygen Therapy Devices [6, p. 94] is included below: • User Skill Level o Participants noted that oxygen concentrators were often used by a wide variety of health workers and therefore, the skill level should be "minimal to none" for both the Optimal and Minimal. Consensus was achieved in the room and no vote was taken. Several participants noted that an oxygen concentrator is a medical device whose output is a drug which can be dangerous if not used properly. The following characteristic was not discussed at the TPP Consensus Meeting, however, it was determined that a new characteristic should be added to the TPP with the following justification: The following characteristics were not discussed at the TPP Consensus Meeting explicitly, however, additional comments were received and incorporated into the discussion: • Oxygen Purity o With regard to the Oxygen Purity range, Pre-Meeting survey voting achieved consensus for the Optimal and Minimal characteristic to be (93% ±3%). A theme emerged in the comments though expressing the need to narrow or broaden this range. While pharmacopoeia's guidelines for Oxygen specify 93%, one participant noted that this guideline is "not for individual concentrators". WHO's existing technology specification for concentrators (2015) [47] as well as ISO's 80601-2-69 specified that low oxygen concentration technical alarm condition shall activate before the concentration drops below 82% volume fraction [72]. International agencies commented that the characteristic should note applicability "at all flow settings" since "Some manufacturers will state different purities for different flow ranges, with lower max purity at the highest flow setting (e.g., 95% at 1 LPM, but 90% at 5 LPM)." o Optimal and Minimal: 93% ±3% • Oxygen Monitor o One theme that arose in the Pre-Meeting survey was confusion on why there were three ranges of oxygen concentration in the Optimal characteristic: "Visual and audible status indicator for three ranges of oxygen concentration preferably with color coding for early warning." One participant clarified that this due to the three ranges indicated in pharmacopoeia: 99, 93 and then 'not for individual concentrators' [73].

100% n = 29
3 comments as summarized below • Pediatrician / Clinical Officer may decide the settings, but the nurse is the one most likely to use the machine • Separate user for repairing the device / changing the filter

62% n = 29
14 comments as summarized below o "Should be at least 2 flow meters for efficiency" o "I don't think we should encourage the inefficient way concentrators are typically used -moved around the ward and used for one or two children at a time" o "Low-pressure piping system to distribute oxygen from a unified concentrator/low-pressure store/backup cylinder system (automatically choosing the cheapest source available at the time). So we don't really care what the concentrator's flow meter is like, and we see no value in having two flow meters. It is not widely known that a typical concentrator uses the same amount of electricity whether it is running at 0.5 LPM or 10 LPM. There is no efficiency gain in running below full capacity, so we prefer to (i) store 'excess' oxygen for use when the concentrator is off, and (ii) automatically switch the concentrator off when the store is full, to minimize electricity use."

68% n = 28
11 comments as summarized below • A variety of hard and soft copy materials mentioned with particular mentions of difficulty in reading a user manual and preference for videos so people can see vs. read • All claims must be filed with the regulatory dossier, so this is not as straight forward as a simple translation. Appropriate, professional translations are a must and are costly to the manufacturer. Additionally, local language varies greatly across a country and is often-times not even the official language of the country and so this may not be a reasonable ask of manufacturers. o "To honor a 5 year warranty, you will have to have strong in-country representation. An extended warranty is a degree of assurance of the above, and this will come at a cost. Manufactures of concentrators willing to extend a warranty from 2-5 do so at a cost. What might be more useful is that during any procurement, consideration be given to establishing a SLA with an incountry rep. In this case, you can take care of any major PPM requirements, as well as "swap out" in the event of a break-down, and there is no discussion of warranties and no need for spares and an in-country source for consumables."  Pulse oximeters use a non-invasive sensor to measure pulse rate (PR) and blood oxygenation levels (SpO2) (i.e., percentage of oxygenated hemoglobin in arterial blood). While pulse oximeters do report pulse rate, their primary purpose and utility is to detect SpO2 in infants. According to the World Health Organization, pulse oximetry is the most accurate non-invasive method for detecting hypoxemia. It is used to measure the percentage of oxygenated hemoglobin in arterial blood (SpO2). The pulse oximeter consists of a computerized unit and a sensor probe which is attached to the patient's finger, toe, or earlobe. The oximeter displays the SpO2 with an audible signal for each pulse beat, a pulse rate and, in many models, a graphical display of the blood flow past the probe (the plethysmographic or pulse wave). The technology is robust and cost effective. Pulse oximeters can be used to both detect and monitor hypoxemia, make more efficient use of oxygen supplies, and improve patient monitoring [32].

Technical Skill Maintenance
Low SpO2 levels can indicate that an infant is in respiratory distress and monitoring SpO2 is important in the neonatal period as it can indicate the need for immediate, critical care interventions. Additionally, SpO2 monitoring is critical for infants receiving oxygen therapy or continuous positive airway pressure (CPAP) therapy. Low SpO2 levels during oxygen or CPAP therapy can indicate that escalation or additional care is required. On the other hand, if SpO2 remains too high (>95%) for too long (often a side effect of pure oxygen therapy), newborns can suffer from preventable disability including retinopathy of prematurity (ROP), a condition that can cause permanent blindness, and chronic lung disease [39,40]. One other consideration when using a pulse oximeter is that the reading may not be as accurate in specific situations (e.g., when a neonate's peripheries are cold, when the neonate is anemic, etc.).

Target Operator
For use in low-and middle-income countries by a wide variety of clinicians, including nurses, clinical officers, and pediatricians

Target Population
Neonates (born at any gestational age and require ongoing care)

Target Setting
Hospitals in low-resource settings, but, may be used in health facilities based on country guidelines Hospitals in low-resource settings To arrive at the final TPP for Pulse Oximeter (Table 21), we conducted a pre-meeting survey. Based on the pre-meeting survey results ( o There was disagreement on whether the Minimal characteristic should require both an auditory and visual alarm. Clinicians discussed that auditory alarms are better at drawing attention, especially when wards may be short-staffed. Product developers confirmed that an auditory alarm was slightly more expensive than a visual alarm and that having both alarms added roughly $3 to the overall cost. Two concerns with auditory alarms were mentioned (alarm fatigue and noise levels impacting baby), however, clinicians agreed that this was a critical alarm and therefore, the benefits of an auditory alarm to stress the importance outweigh the concerns. Following the Consensus Meeting, one participant commented that "Inability to disable alarms for more than 2 min is a critical safety issue. The ability to configure the default alarm is critical. This will address almost all the discussion we had on this issue." o Minimal: Auditory  Overall Vote -84% Agree (n = 38)  Clinicians -85% Agree (n = 27)  Excluding involvement with product development -86% Agree (n = 35)

• Alarm Limits -Pulse Rate (PR)
o There was disagreement suggesting a wider range for the Minimal characteristic and a discussion of whether the range should be fixed or variable (i.e., users can set the range). Some clinicians felt that the range should be fixed for certain levels of care (e.g., secondary or primary level) while others thought that having a factory setting pre-programed but that could be adjusted would provide flexibility. Some users noted the flexibility would be helpful for trainings and where altitude could present challenges. Clinicians noted that they rarely vary the factory settings (when asked the last time they adjusted the setting, one replied "over four months ago" o There was disagreement on the Minimal characteristic with similar commentary on the concern of alarm fatigue ("it is not helpful if the alarm is sounding permanently on a sick child") and the impact of altitude on the lower range limit. There was a discussion reviewing the Pre-Meeting survey comments for the Minimal characteristic:  Adjustable: "You want to set the alarm according to the environment; e.g., the altitude might impact the levels you want and normal values of oximetry may be lower"  Non-Adjustable: Adjustability of the alarms increase risk of user error and/or use on a different patient population  Partially Adjustable: "Should be closed settings not fully adjustable. For example 1) neonate setting 2) infant setting 3) pediatric setting, etc." o Consensus was achieved in the room (without a Mentimeter vote) that the range should be adjustable for the Minimum, as well as the Optimal, to provide flexibility based on the patient type. o Minimal: Adjustable • Consumables o Agreement was reached in the room on clarification that the consumables in question were to be specified as two neonatal probes (designed for and tested in newborns). Clinicians in the room commented that two neonatal probes should be included in the package when initially purchased. Product developers noted that measuring by a period of time can be challenging since it's often difficult to prove whether the probes have been used improperly. One consideration was changing the measurement to the strength of the probe rather than the length of time. Furthermore, product developers noted that the cabling on the sensor of the probe is the weakest part and that the lifespan will decrease if twisted around improperly. Some users mentioned a preference for reusable probes while others mentioned that disposable probes "fit better" and were therefore preferred. Consensus was achieved in the room (without a Mentimeter vote). o For the Optimal characteristic, many different configurations were noted including: easily movable; not docked, not "pocketable". Specifically, clinicians commented that the device should be "moveable, but not too small that it can be taken away from the unit". The idea of "chaining" the device in the unit to avoid being moved was mentioned. Clinicians noted that for continuous monitoring, they prefer the display screen to be larger so that it is readable from a certain distance. One participant emphasized that often times, there is limited space available in the NICU and there may be limited  Overall Vote -59% voted ">6hr" (n = 32)  Clinicians -62% voted ">6hr" (n = 21)  Excluding involvement with product development -60% voted ">6hr" (n = 30) o Optimal: Rechargeable battery, >24hr on single charge o Minimal: Rechargeable battery, >6hr on single charge • Instrument Pricing o There was disagreement on the Minimal characteristic for ex-works price of the device (inclusive of warranty and two probes for neonatal use). Some participants noted that the ex-works price was misleading given that there are several mark-ups added and that the landed cost may be easier for buyers to understand. Product developers noted that $100 ex-works is not feasible for a continuous measurement device (i.e., not a "finger pulse ox"). o Minimal: <$250 ex-works  Overall Vote -85% Agree (n = 20)  Clinicians -92% Agree (n = 13)  Excluding involvement with product development -85% Agree (n = 20) • Consumable Pricing o There was disagreement on the Minimal characteristic for consumable pricing which, for the basis of the discussion, was assumed to be two neonatal probes per year. Technical developers discussed that the probes were an expensive component and that the current cost per probe is $20-$40 per probe ex-works with an average lifespan of 6 months. o Minimal: <$80 per year ex-works (two probes)  Overall Vote -86% Agree (n = 14)  Clinicians -88% Agree (n = 8)  Excluding involvement with product development -86% Agree (n = 14) • Voltage o As noted in the WHO-UNICEF Technical Specifications and Guidance for Oxygen Therapy Devices [6, p. 68], "In the case of oxygen therapy products, poor power conditions can significantly harm electrically powered oxygen concentrators, as well as pulse oximeters that require power directly from a mains source, or require recharging from a mains source". There was disagreement on the Minimal characteristic and whether a separate TPP was needed for a voltage stabilizer, although it was noted the WHO-UNICEF Technical Specifications and Guidance for Oxygen Therapy Devices [6, p. 133] does provide technical specifications for voltage stabilizers specific to those paired with oxygen therapy products. Agencies noted the importance of considering global ranges in development. From a technical perspective, a message to clinicians was to ensure that facilities install "grounding" (e.g., use of a metal rod). One proposal was to clear safety guidelines for medical device voltage per country. The following characteristics were not discussed at the TPP Consensus Meeting explicitly, however, additional comments were received and incorporated into the discussion: • SPO2 Range o With regard to the SP02 range, Pre-Meeting survey comments highlighted that "Saturation at 0% is not clinically meaningful", "there is no method available for calibrating pulse oximeters below 70%", and that "[readings are] never accurate or clinically useful below 70%". One participant responded that while oxygen therapy ideally would have started before the patient reaches these levels, there may be value and "clinical utility to ensure that the patient IS resaturating".

• Decontamination
o Pre-Meeting survey comments highlighted the need to clarify appropriate disinfection agents. Comments received from an international NGO provided further clarification noting that each country has their own decontamination protocol since the WHO only provides guidance rather than explicit protocol. The guidance provided specifies super-basic mild soap solution, not submerging the device, and wipe-able in the case of contact with bodily fluid, and ability to use scheduled disinfectant [50]. While the process of decontaminating would likely be carried out by an IPC specialist, it is important for the manufacturer to control their Ingress Protection (IP) rating. • The following Product Specific ISO Standards were highlighted in the Pre-Meeting survey responses: o ISO 80601-2-61(current 2017) specific to pulse oximetry, title: Medical electrical equipment --Part 2-61: Particular requirements for basic safety and essential performance of pulse oximeter equipment, and provides guidance on accuracy claims and validation. and ISO13485 • Additional considerations received from participants are as follows: o "We should specify the conditions / context for accuracy testing. In newborns the within subject (breath by breath) variation in SpO2 within a single minute when the SpO2 is below 95% is > 3% RMSD. ISO only requires testing in adults. Currently ISO accuracy is < 4% RMSD. Neonates at low SpO2 will be at least this for a "minimal" requirement." o "Motion, perfusion, skin color and external light interference are key issues that have not been addressed." o "Devices need to be cleanable, waterproof (to a degree-IPX rating), drop and vibration tolerant."

Target Operator
Optimal: For use in lowand middle-income countries by a wide variety of clinicians, including nurses, clinical officers, and pediatricians.

98% n = 42
6 comments as summarized below • Theme: Broaden Users • Add 'nurse assistants' and 'community health workers' • Non licensed providers make up a significant proportion of the healthcare workforce. Pulse oximetry monitoring is simple to learn so it does not exclusively require licensed providers if they are not available (i.e. in lower levels of the healthcare system) • Optimal would be if a lay person could use it

80% n = 41
12 comments as summarized below • Theme: Broaden age range or specify weight range • Typically manufacturers specify a weight range not an age range v1.2

Optimal Minimal
• Pulse oximetry is useful in small hospitals and clinics where newborn care might be a small part of their workload, and any oximeter should be used also for older children • Optimal/minimal would be <28days but also compatible for infants 1-6 kg • Make upper weight higher if aiming to care for older sick infants (upper limit then probably 8-10 kg)

Target Setting
Optimal: Hospitals in low-resource settings

74% n = 43
19 comments as summarize below Minimal: Same as Optimal.

comments as summarized below
• Theme: Accuracy Data at Various Perfusion / Movement Conditions o UNICEF SD/WHO specs will be +/-3% for neonates, and most devices that make claims will not go beyond this because you cannot carry-out a lab desaturation (breathdown) on a neonate to validate otherwise. v1.2

Optimal
Minimal o "SpO2 accuracy (in the range at least 70-100%): within ± 2% under ideal conditions of use, and within ± 3% for all patients and perfusion/movement conditions." o For both minimal and Optimal (whatever the accuracy threshold is chosen to be for each), at least the detection range and motion/no-motion should be specified in order to compare apples to apples o Require as 'Optimal' that proof of accuracy data be available, as we have found that many are unable to provide supporting data showing compliance to ISO

Optimal
Minimal a rare case scenario (for tachyarrhythmias)  160 bpm is too low for an upper limit -suggest using 180 / 200 bpm as upper limit to avoid frequent alarming in the "borderline" babies with HR 160 -180 bpm which may be due to crying or restlessness instead of illness  50-120 bpm  80-180 bpm  I would also want the device to get an alarm at 60 bpm in any resuscitation situation o Non-Adjustable -Adjustability of the alarms increase risk of user error and/or use on a different patient population o Partially adjustable -should be closed settings not fully adjustable. For example 1) neonate setting 2) infant setting 3) pediatric setting, etc.  "In a district hospital, I would want the alarms to be locked; in a tertiary I prefer the alarms to be adjustable."

59% n = 39
18 comments as summarized below 9 comments as summarized below • Theme: 'Minimal' is too subjective; need something more specific • Users need to be trained on the significance of monitoring • Most of training is not on the device but the application of the sensor and the interpretation of information • More specificity required, both with respect to minimum user qualifications and time -e.g., "A health care worker with at minimum a nursing degree can be trained in a 2-day workshop" or "A community health worker can be trained in a 1week course", etc... • Ideally should not require training or training built into device or easily accessible via phone

User Manual
Optimal: User manual and additional training materials (checklists, videos, guides) in English and local language. Attached to device with labels and markings where possible.

85% n = 40
15 comments as summarized below • Focus on limits of the pulse oximeter • One manual per ward versus one per device • Manual should be easily found online • Not necessarily the responsibility of the manufacturer • All claims must be filed with the regulatory dossier, so this is not as straight forward as a simple translation. Appropriate, professional translations are a must and are costly to the manufacturer. Additionally, local language varies greatly across a country and is often-times not even the official language of the country and so this may not be a reasonable ask of manufacturers

comments as summarized below
• Applicable to the battery charger and charging station • The requirements for power input voltage/frequency and plug type of the equipment must be chosen according to the local electrical supply. Source: https://www.220electronics.com/media/images/world-voltage-map.gif • Voltage can always be corrected with step-up / stepdown transformers; however, these come at an added cost. So whether the cost be borne by the purchaser (Caribbean, Central-or South-American countries w/ 120V) or the manufacturer who makes devices that can work across all contexts • Frequency needs to be appropriate for frequency rating of specific country, as this is something that cannot be corrected and though 50 Hz can be used in a 60 Hz system, it is hard on the device and it will be compromised v1.2

Optimal Minimal
Voltage stabilizers and surge suppressors are important to consider

Battery Powered
Optimal: >24hr on single charge

93% n = 40
Minimal: None 36% n = 36 23 comments as summarized below 13 comments as summarized below • Theme: 5 years may be unrealistic • UNICEF-WHO spec is 2 years recommended, at least 1 year mandatory • Optimal should by 2 years • To honor a 5 year warranty, you will have to have strong in-country representation • "Any manufacturer that I have ever spoken to was more than willing to extend a warranty (to 2, maybe 3), but at a cost" • "What might be more useful is that during any procurement, consideration be given to establishing a SLA with an in-country rep. In this case, you can "swap out" in the event of a break-down, and there is no discussion of warranties" Instrument Pricing Optimal: <$150 exworks

comments as summarized below
• Theme: Extremely price-sensitive geography and even $250 was viewed as too expensive by some respondents • Optimal price was viewed as potentially overly ambitious for bedside rather than handheld type • This device needs to be better than devices sold in high-income countries so may be tough to hit target price • Cheaper options available • Would need to understand quality of the device before paying this much • I think you could safely set "Optimal" to <$100, and "Minimal" to <$175 for ex-works, including 1 probe (min) and 1 year warranty on unit v1.2

79% n = 33
Minimal: <$100 per year ex-works 47% n = 34 16 comments as summarized below • Theme: Extremely price-sensitive geography and $100 was viewed as too expensive by some respondents o "Generic probes cost much less than that, and last more than a year" o Too costly if above $50 / year • Theme: Provide more specificity for quantity and type of consumable o Differentiate between a consumable (disposable probe) and spare (reusable probe). I am assuming that this question is about reusable probes. o I think you could safely set "Optimal" to <$40, and "Minimal" to <$80 for ex-works, probes have 6 mo. warranty for 2 disposable probe and 2 reusable probe v1.2 Respiratory rate is a critical vital sign. The causes are many but are commonly due to respiratory pathology. Increased respiratory rate (> 60bpm) in newborns can indicate respiratory distress syndrome (RDS), but as with infants and children, a high respiratory rate can also indicate pneumonia, which is the primary infectious cause of childhood death worldwide. A low respiratory rate or gaps in breathing in infants is likewise indicative of potentially severe health concerns. Apnea of prematurity is a condition in which newborns temporarily stop breathing. Many apneas resolve without intervention, but frequent apnea (often paired with bradycardia and low SpO2) can indicate an underlying condition such as sepsis, hypoglycemia, or anemia. Apnea of prematurity (AOP), a condition in which newborns temporarily stop breathing due to neurologic immaturity, affects nearly 50% of infants born earlier than 32 weeks gestational age and nearly 100% of those born at fewer than 28 weeks, and may last for several weeks [51]. AOP can be associated with dangerous decreases in heart rate and oxygenation, which, left unchecked, could lead to respiratory arrest, increased morbidity, or death.
In high-resource settings, respiratory rate is monitored using impedance pneumography, which requires expensive patient monitors and delicate electronic sensors. Alternatively in high-resource settings, AOP is monitored by using low nursing ratios (1:2) in conjunction with continuous heart rate and pulse oximetry monitoring. In this setting, a nurse or caregiver would provide a manual intervention in the event of an AOP event causing a low heart rate or oxygen saturation, in order to reestablish normal breathing. In low-resource settings, a nurse, normally faced with high nurse to patient ratios, must rely on limited continuous monitoring capability of heart rate and saturation with most infants only receiving intermittent manual monitoring. Additionally, they should observe the number of breaths a child takes in one minute, a procedure that is both timeconsuming and inadequate for monitoring infants at risk of AOP.
FINAL TPP: RESPIRATORY RATE / APNEA MONITOR

Intended Use
To provide continuous monitoring of respiratory rate

Target Operator
For use in low-and middle-income countries by a wide variety of clinicians, including nurses, clinical officers, and pediatricians

Target Population
Neonates (born at any gestational age and require ongoing care)

Target Setting
Hospitals in low-resource settings, but, may be used in health facilities based on country guidelines Hospitals in low-resource settings

Regulation
At least one of: CE marking, approved by US FDA or another stringent regulatory body of a founding member of IMDRF (e.g., Japan or Australia or Canada) TECHNICAL CHARACTERISTICS

Apnea Detection
Detect periods of central apnea exceeding 20s duration (at 0)

Alarm
Visual and auditory An alarm (visual or auditory)

Patient Interface
Interface is biocompatible and reusable Interface is biocompatible

Respiratory Rate Alarm Limits
Automatically adjust based on patient age 30-60 bpm

Voltage
Model must match the voltage and frequency of the purchasing country's local power grid (e.g., 110-120 VAC at 60 Hz or 220-240 VAC at 50 Hz) TRAINING AND MAINTENANCE

User Instructions
User manual and additional training materials (checklists, videos, guides) in at least one national official language for the country of intended use. Attached to device with labels and markings where possible User manual provided in at least one national official language

years 1 year
Decontamination Easy to clean with common disinfecting agents 1 There was not 75% voting agreement on the Minimal characteristic. Please refer to the TPP Report discussion for additional detail.

CONSENSUS MEETING SUMMARY: RESPIRATORY RATE / APNEA MONITOR
To arrive at the final TPP for Respiratory Rate/Apnea Monitor (Table 23), we conducted a pre-meeting survey to prioritize the items for discussion at the Consensus Meeting for characteristics that achieved below 75% agreement in the survey results (Table 24). An overview of the discussion at the Consensus Meeting of these characteristics is included below.

• Apnea Detection
o Consensus was achieved in the room (without a Mentimeter vote) for both the Optimal and Minimal characteristics. Clinicians confirmed that they definitely wanted the monitor to alarm for apnea and that, additionally, it would be helpful to have the ability to adjust the interval detection frequency based on the baby. Product developers noted that this technology was not fully mature yet and challenging to improve. They explained that from a technical perspective, the rate was retrospective and therefore more complex to technically calculate the average over a historical period of time and produce a read out based on the determined algorithm. One clinician suggested that the algorithm be built so that when a period of apnea was detected, a side countdown begins and if it hits 20 seconds, an alarm would sound. Both clinicians and technical developers agreed on the importance of two separate counters: one for historical averages of respiratory rate and a second for when a baby experiences apnea, upon which a prompt warning alarm would sound. One international NGO participant mentioned an interest in better understanding 'normal' apnea patterns/trends in newborns prior to agreeing on alarm levels since desaturation could happen quite quickly. o Consensus was achieved in the room (without a Mentimeter vote) for the Minimal characteristic. Product developers noted that it can be challenging to conduct validation on accuracy for ±2 bpm since a gold standard does not currently exist to measure respiratory rate accuracy. A research question was developed emphasizing the need for an improved way to measure accuracy since international standards for respiratory rate accuracy do not currently exist. There is therefore a need to define gold standard for respiratory rate accuracy and standardize experimental conditions. Ethical considerations are important in evaluating and validating these standards at upper and lower ranges on neonates. One participant recommended that both SpO2 and respiratory rate accuracy thresholds be based on real clinical data (typical variability). In the Pre-Meeting report survey, one individual commented that given there was not a 'gold standard' measurement for respiratory rate, they specified a reasonable reference standard with human experts and video recordings and specifying an acceptable degree of agreement with that standard, using the 95% Limits of Agreement and the Bland-Altman plot. However, an international NGO responded that using humans as a 'reasonable reference standard' can be troublesome since they can often be inconsistent or incorrect. Furthermore, they noted that "regulators will likely not see [human experts] as a means to validate". o Clinicians noted that the intention is to leave the device on for 24 hours, hence the time period. Discussion in the room encouraged product developers to be creative (e.g., device could plug into wall, connect with other devices, etc.). Clinicians noted a preference to avoid wired connections to mains and emphasized that "there are already too many wires". There was agreement in the room that if the device was not connected to a mains power source, constant power for 24 hours would be required, however, if it was connected to a mains power source, then 12 hours back-up for power shedding would be sufficient for the Optimal characteristic. For the Minimal characteristic, if the device was not connected to a mains power source, constant power for 24 hours would be required, however, if the device was connected to a mains power source, then at least 6 hours of back-up for power shedding should be required. Product developers noted that the battery was more complex than a "watch battery" since certification was required for each part and supplier used in development. o A research question was established to review existing literature on power cuts to determine how long power supply should last. One meeting participant subsequently sent the following recommendations providing data on power cuts to share with the broader group in this report: 1) Limited electricity access in health facilities of sub-Saharan Africa: a systematic review of data on electricity access, sources, and reliability [66] 2) Oxygen insecurity and mortality in resource-constrained healthcare facilities in rural Kenya [67]   • "I agree that this is the population that should be able to apply and trouble shoot a respiratory monitor -but it's not realistic in my opinion that the nurse:patient ratio will be such that they can respond to all the alarms."

73% n = 11
6 comments as summarized below • Theme: Broaden age range or specify weight range

Target Setting
Optimal: Hospitals in lowresource settings

Apnea Alarm Limits
Optimal: Adjustable 82% n = 11 Minimal: None 70% n = 10 6 comments as summarized below • "If the system has a built in apnea alert for pauses > 20 seconds, then there shouldn't be room to adjust it, possibly to silence the alarm but not to change the limits" • "What does it mean to have an "adjustable" apnea alarm? Like it only alarms if it's associated with a decrease in heart rate as well? Or do you mean that you can adjust the length of the apnea period for which it alarms? That also wouldn't really make sense to me as it seems like this would be a parameter internally set to optimize sensitivity/specificity of alarms" • "What about alarms for battery, error, etc."

Consumables
Optimal: >12 months before required

82% n = 11
5 comments as summarized below • Electronic copy is highly preferred • All claims must be filed with the regulatory dossier, so this is not as straight forward as a simple translation. Appropriate, professional translations are a must and are costly to the manufacturer. Additionally, local language varies greatly across a country and is often-times not even the official language of the country (take India, for example) and so this is simply not a reasonable ask of manufacturers. "User language preference prioritized, English is mandatory." Also, any manufacturer should be encouraged to use pictograms to support user manuals

80% n = 10
4 comments as summarized below • Theme: Even low-cost consumables become a financial burden, and single-use items should be avoided wherever possible

Respiratory Rate Alarm Limits
Optimal: Automatically adjust based on patient age

comments as summarized below
• Theme: 5 years too long • Suggested Ranges: o 2 years • To honor a 5 year warranty, you will have to have strong in-country representation. All an extended warranty is a degree of assurance of the above, and this will come at a cost. Manufactures of concentrators willing to extend a warranty from 2-5 do so at a cost. What might be more useful is that during any procurement, consideration be given to establishing a SLA with an in-country rep. In this case, you can take care of any major PPM requirements, as well as "swap out" in the event of a break-down, and there is no discussion of warranties and no need for spares and an in-country source for consumables.

Intended Use
Aspiration and removal of secretions, bodily fluids and foreign objects from a patient's airway or respiratory support system in the nasal, pharyngeal and tracheal areas

Target Operator
For use in low-and middle-income countries by a wide variety of clinicians, including nurses, clinical officers, and pediatricians

Target Population
Neonates (born at any gestational age and require ongoing care)

Target Setting
Hospitals in low-resource settings, but, may be used in health facilities based on country guidelines Hospitals in low-resource settings Bottle Capacity 1 L

Noise Level
As low as possible

Cleaning
Collection vessel easy to clean reusable

Maintenance
No maintenance or lubrication

Operation Mode
Adjustable to neonatal setting (60-100 mm Hg)

Voltage
Model must match the voltage and frequency of the purchasing country's local power grid (e.g., 110-120 VAC at 60 Hz or 220-240 VAC at 50 Hz)

User Instructions
User manual and additional training materials (checklists, videos, guides) in at least one national official language for the country of intended use. Attached to device with labels and markings where possible User manual provided in at least one national official language Warranty 5 years 1 year 1 There was not 75% voting agreement on the Minimal characteristic. Please refer to the TPP Report discussion for additional detail.

CONSENSUS MEETING SUMMARY: SUCTION PUMP
To arrive at the final TPP for Suction Pump (Table 25), we conducted a pre-meeting survey to prioritize the items for discussion at the Consensus Meeting for characteristics that achieved below 75% agreement in the survey results (Table 26). An overview of the discussion at the Consensus Meeting of these characteristics is included below.  o Consensus was achieved that the sound level characteristic was referring to the operating noise level. Some product developers noted that from a technical standpoint, CE mark requires that this be under 50 decibels for operating noise, however, another participant confirmed that this was simply the minimum end of the range required and that "in operating rooms, the background noise can vary from 50 dBA to 85 dBA". Ultimately, consensus was achieved in the room (without a Mentimeter vote) for both the Optimal and Minimal characteristic to be the same and specify that the "lower the decibel level, the better". The spirit of the conversation emphasized that the noise levels should be as low as possible to protect the babies hearing.

Optimal Minimal
Cleaning Optimal: Collection vessel easy to clean reusable.

100% n = 12
2 comments as summarized below • Collection and patient interface easy to clean and reusable

Maintenance
Optimal: No maintenance or lubrication.

82% n = 11
4 comments as summarized below • Clarity on what is meant by lubrication -nasal saline prior to suctioning? • Maintenance should be required but minimal and/or easy

90% n = 10
4 comments as summarized below • Optimal: 60-120 mm Hg • Optimal: recommend adding -continuous adjustment within the full range • Need clarity as to why this is linked to the pressure only. I would maybe think of battery or mains operation mode or electrical or manual operation mode, or adult, pediatric or neonatal operation mode

User Manual
Optimal: User manual and additional training materials (checklists, videos, guides) in English and local language. Attached to device with labels and markings where possible.

THERMAL MANAGEMENT
In general, newborns require a warmer environment than adults and the smaller the newborn, the higher the temperature needs to be. A newborn's ability to stay warm can be easily compromised by the temperature of its surroundings since newborn infants regulate body temperature much less efficiently than adults and lose heat more easily. Low birth weight and premature babies often face even greater risk [74].
As many as 85% of infants born in hospitals in low-resource settings become cold (defined as <36.5°C) [52]. Mortality rates increase with each degree Celsius of temperature lost. While the risks of being too cold are well recognized, hypothermia remains a largely invisible problem in overcrowded newborn units in low-resource settings. Hypothermia in newborns requires rapid diagnosis, which is often difficult in crowded and understaffed wards. Hypothermia not only increases the chances of acidosis, sepsis and RDS, but may indicate the presence of system illness such as infection or hypoglycemia.
Hypothermia can be treated using Kangaroo Mother Care (KMC), blankets/hats, warming cribs, warming mattresses, and radiant warmers. While hypothermia can be treated using KMC, infants and their caregivers may not be eligible for reasons such as, but not limited to: mother is recovering from surgery or the infant is in need of intensive care.
Attempts to warm a cold baby without monitoring temperatures carefully can result in hyperthermia. Rapid swings in temperature -known as thermal shock -can lead to negative outcomes, including death. Additionally, unrecognized fever in infants may lead to delays in treating neonatal sepsis and resulting in increased morbidity.
In high-resource settings, these negative outcomes are prevented by using incubators which continuously monitor and adjust temperature, or, with intermittent monitoring (every 3-4 hours) for infants who are in open cribs. However, incubators cost thousands of dollars and often require delicate sensors and expensive consumables. Existing temperature monitoring devices that are affordable in lower resource settings do not have the features necessary for the accurate detection of hypothermia or are not designed for a clinical setting.
In addition to the risks of hypothermia, pre-term infants and children are at high risk of infection, which can cause hyperthermia. A diagnosis of fever is not conclusive for any of these conditions, but it is a critical early sign of potentially severe illness. In combination with a respiratory rate monitor and pulse oximeter, continuous temperature monitoring can provide guidance to v1.2

Page 185
clinicians on what type of treatment to pursue; once treatment has begun, it can indicate whether treatment is working or needs to be increased.

RADIANT WARMERS INTRODUCTION: RADIANT WARMER
Hypothermia can be prevented using radiant warmers that carefully control heat based on manual settings or the infant's own temperature. Radiant warmers provide heat using an overhead heating source and are preferred for infants who may require greater access or closer short-term monitoring. Radiant warmers are preferred, in the short term, to warming cribs/incubators for infants who are unstable and may require significant intervention (such as resuscitation or invasive procedures).

Voltage
Model must match the voltage and frequency of the purchasing country's local power grid (e.g., 110-120 VAC at 60 Hz or 220-240 VAC at 50 Hz) TRAINING AND MAINTENANCE

User Instructions
User manual and additional training materials (checklists, videos, guides) in at least one national official language for the country of intended use. Attached to device with labels and markings where possible User manual provided in at least one national official language

years 1 year
Decontamination Easy to clean with common disinfecting agents 1 There was not 75% voting agreement on the Minimal characteristic. Please refer to the TPP Report discussion for additional detail.

CONSENSUS MEETING SUMMARY: RADIANT WARMER
To arrive at the final TPP for Radiant Warmer (Table 27), we conducted a pre-meeting survey to prioritize the items for discussion at the Consensus Meeting for characteristics that achieved below 75% agreement in the survey results (Table 28). An overview of the discussion at the Consensus Meeting of these characteristics is included below.

comments as summarized below
• It should be specified the consumable presentation: box/piece/set • Should be as cheap as possible -temperature probes easily break and will be used heavily • We are talking about low resource setting, and high prices for the equipment will not be feasible for this countries v1.2 Given that temperatures less than 36.5°C have been shown to be an independent risk factor for death in neonates [53], early recognition and treatment of hypothermia is critical. In overcrowded and understaffed hospital wards, where nursing to patient ratios are often in excess of 1:10 and most infants are not in incubators which continuously record temperature, obtaining temperature readings even 3-4 times per day can be challenging.
In high-resource settings, low nursing to patient ratios and availability of incubators, which continuously monitor temperatures, allows for close monitoring. In settings with high nurse to patient ratios, where incubators are limited, KMC is the preferential warming option. However, some infants require closer monitoring of temperature in open cribs and the ability to continuously monitor temperature and notify staff when an intervention is needed could greatly reduce hypothermia and increase recognition of neonatal fever associated morbidity and mortality. User manual and additional training materials (checklists, videos, guides) in at least one national official language for the country of intended use. Attached to device with labels and markings where possible User manual provided in at least one national official language

years 1 year
Decontamination Easy to clean with common disinfecting agents Same as Optimal 1 There was not 75% voting agreement on the Minimal characteristic. Please refer to the TPP Report discussion for additional detail.

CONSENSUS MEETING SUMMARY: TEMPERATURE MONITOR (CONTINUOUS)
To arrive at the final TPP for Temperature Monitor (Table 29), we conducted a pre-meeting survey to prioritize the items for discussion at the Consensus Meeting for characteristics that achieved below 75% agreement in the survey results (Table 30). An overview of the discussion at the Consensus Meeting of these characteristics is included below.  Overall Vote -74% voted "36.5°C" (n = 19)  Clinicians -81% voted "36.5°C" (n = 16)  Excluding involvement with product development -75% voted "36.5°C" (n = 16) • Battery (previously titled 'Battery Power') o There was agreement in the room that all characteristics relating to Utility Requirements (e.g., Back-up Battery; Battery Power; Batteries; Voltage; Power Requirement; Maximum Power Consumption; Response During Power Outage; Surge Protection, Electrical Plug) be reviewed and harmonized following the TPP meeting. o There was a discussion in the room emphasizing the importance of reliable power supply for minimum of 24 hours. Clinicians noted that the intention is to leave the device on for 24 hours, hence the time period. Discussion in the room encouraged product developers to be creative (e.g., device could plug into wall, connect with other devices, etc.). Clinicians noted a preference to avoid wired connections to mains and emphasized that "there are already too many wires". There was agreement in the room that if the device was not connected to a mains power source, constant power for 24 hours would be required, however, if it was connected to a mains power source, then 12 hours back-up for power shedding would be sufficient for the Optimal characteristic. For the Minimal characteristic, if the device was not connected to a mains power source, constant power for 24 hours would be required, however, if the device was connected to a mains power source, then at least 6 hours of back-up for power shedding should be required.

INTRODUCTION: CONDUCTIVE WARMER
Since low birth weight or sick newborns are most vulnerable to hypothermia, the World Health Organization has outlined various methods that can be used to keep high-risk babies warm including kangaroo-mother care, "warm rooms", heated mattresses, radiant warmers, and incubators. These methods vary in their response to addressing the four different ways in which newborns lose heat: radiation, convection, evaporation, and conduction [74].
Conductive warmers provide conductive heating either below or around the patient while also allowing health care workers with visibility and access to the baby. Given the high cost of some warming devices (e.g., incubator, radiant warmer), a need exists in low-resource settings for a technology that is both affordable and easy to use, and that can accurately detect hypothermia while keeping the newborn warm. The advantages of using warming devices include the fact that extra warmth can be given locally instead of having to warm the whole room; temperature control is easier; and newborns can be fully observed and visible. The World Health Organization explains that different devices serve different purposes and advises that incubators are the proper choice for the care of very small newborns during the first few days or weeks. When these babies no longer have acute problems, they can be cared for safely on heated water-filled mattresses. Radiant heaters are best used for resuscitation and interventions where a number of people are involved [74].
Negative outcomes associated with hypothermia can be prevented using warming cribs that carefully control heat. Conductive warmers may be called warming cribs however are distinct from incubators. The intent in the development of this TPP was to provide developers with the opportunity to be innovative in the design process rather than be constrained by existing technologies or preconceived notions that a "crib" must be enclosed.
A need for the creation of a separate TPP for an incubator was identified at the Consensus Meeting. Incubators are the conventional method for maintaining normothermia in preterm and low birthweight neonates. Risks associated with incubator care include hypothermia [54,55]; hyperthermia [56]; nosocomial infections, related to lack of effective cleaning standards [57][58][59]; and cross-infection from other neonates when incubators are shared, a common practice in low-resource facilities. Failure of incubators to properly regulate temperature may be related to malfunction (e.g., over-or under-heating) [56,[59][60][61][62], loss of electrical supply [63], ignorance of how to regulate set-points [56], as well as environmental factors [60]. In low-and middleincome countries, where there may be few nurses and doctors available, neonates in incubators may not receive adequate v1.2 monitoring and serious events (e.g., apnea) may not be detected in time. Due to high purchase cost and poor routine maintenance practices, hospitals in such settings commonly face shortages of functional incubators [52,[63][64][65].

Intended Use
Treatment and prevention of hypothermia in neonates requiring thermal care

Target Operator
For use in low-and middle-income countries by a wide variety of clinicians, including nurses, clinical officers, and pediatricians

Target Population
Neonates (born at any gestational age and require ongoing care)

Target Setting
Hospitals in low-resource settings

SAFETY AND STANDARDS
Quality Management 6 ISO 13485:2016 Medical devices -Quality management systems --Requirements for regulatory purposes

Regulation
At least one of: CE marking, approved by US FDA or another stringent regulatory body of a founding member of IMDRF (e.g., Japan or Australia or Canada)

Form Factor
Enclosed or not enclosed (no preference)

Voltage
Model must match the voltage and frequency of the purchasing country's local power grid (e.g., 110-120 VAC at 60 Hz or 220-240 VAC at 50 Hz)

User Manual
User manual and additional training materials (checklists, videos, guides) in at least one national official language for the country of intended use. Attached to device with labels and markings where possible User manual provided in at least one national official language

Conductive Warmer Page 214
To arrive at the final TPP for Conductive Warmer (Table 31), a smaller group convened at the TPP Consensus meeting to determine which characteristics should be included in a brand new TPP for a Conductive Warmer. A need for a new TPP arose when it was determined that there separate TPPs were required based on the method of heating. Three methods of heating were outlined: 1) Radiant Heat (e.g., Radiant Warmer / resuscitaire) 2) Conductive Heat (e.g., Conductive Warmer) 3) Convective Heat (e.g., Incubator) The smaller group discussion focused on the Conductive Warmer TPP as standards for incubators in high-resource settings currently exist. It was noted that there is a potential need for adjustment of these incubator standards for low-resource settings. Note that a pre-meeting survey for a Warming Crib was conducted and survey results are included in Table 32.  Minimal: Same as Optimal.

89% n = 9
2 comments • Assuming that this refers to the temperature of the baby e.g. through skin temperature probe, it is difficult to comment on what this number should be as we do not know deviance between bench testing (as above) and real-world. We can only really design to meet a bench-testing level • +-0.1

comments
• In my mind, a warming crib should not lose heat but stay at a constant temperature • Is this the retention of heat within the baby or the device (mattress/air etc.)

Optimal Minimal
• Atom, a recognized Japanese incubator manufacturer, state that the CO2 level for their V-2100 incubator is 0.4% following this test. Quote "CO2 concentration when stability has been achieved after administering air mixed with 4% CO2 to a point 10cm above the center of the mattress at 750mL/min doesn't exceed 0.4%."

Maximum Rate of Change in Infant's Temperature
Optimal: 0.5°C/hour 75% n = 8 Minimal: Same as Optimal.

86% n = 7
3 comments as summarized below • 1 degree per hour should be better • This is assuming closed-loop control with sensor. That was not mentioned above so may be confusing • In the case of incubators, there are specific standards to follow and we are not in a position to comment on how quickly a baby will warm up or lose heat as this will depend on their clinical state

comments as summarized below
• Should specify that this is referencing "Maximum Air Temperature" • Need to clarify what temperature this is. Is it baby, air, pad? • Maximum temperature should be 37.5°C • This may be specific to incubators (most of which actually go as high as 39°C) but feedback from all users indicate that they are never set above 36.5°C and rarely higher than 36°C (note, I refer to the temperature of the air, not the baby)

Patient Accessibility and Visibility
Optimal: Patient is visible and accessible to healthcare worker.

90% n = 10
2 comments as summarized below • Define visible and accessible

Patient Size
Optimal: Should fit a single infant <10kg

60% n = 10
6 comments as summarized below • Theme: Should the warming crib fit more than one baby or not • This will be most critical in septic and new, preterm infants. So you need a lower limit (1kg) for which the warming crib also works • 10 kg seems large for a neonate • Should correspond to babies <28 days -6kg max, 8 with contingency

Temperature Control
Optimal: Based on infant's temperature and includes fail-safe mode

78% n = 9
2 comments as summarized below • For the incubator, temp control is based on air temperature. User research early on identified risks with patient temp control e.g. probes not properly attached. Agree fail-safe mode required -if temp runs higher than set temp • Should this also include a manual mode with simple settings? How does this spec limit developers to address the risks of multiple babies in one device?

User Manual
Optimal: User manual and additional training materials (checklists, videos, guides) in English and local language. Attached to device with labels and markings where possible.