Care of neonates in Kilimanjaro region is characterized by poor documentation and a lack of common systems for documentation across facilities; in addition many of our findings suggest substandard care even within the context of available resources. These findings are apparently similar to those in Kenya  and those found among older children in the same region . While the neonatal mortality rates recorded at these facilities are lower than what might be expected given the conditions, the high mortality rate at the zonal referral hospital raises concern since many high risk women and neonates are transferred there.Substantial discrepancies noted between recorded referrals to thezonal referralhospital and records of received neonates at the zonal referral hospital demonstrate some of the challenges faced inhealth systems management, evaluation process and outcome of care.
One of our findings was the lack of a universal system for recording medical assessments and care provided to sick neonates, aside from the antenatal care card, which has very limited space for such evaluations. Neonatal care records were often difficult to locate and information provided within them was scant, with key diagnoses such as prematurity notably lacking in a substantial proportion of cases. Although absence of a record does not imply absence of examination, a poor record of care inevitably damages prospects for good continuity of care between staff and local systematic audits are difficult to perform. This is particularly important in neonatal care where clinical states change rapidly. Given the complexities that are unique to the care of neonates (high likelihood of hypothermia, hypoglycaemia, differences in “cut-off” for empiric antibiotics in the case of fever, need for different types of intravenous fluids), the availability of standard care forms which guide health care workers through the process of a typical neonatal admission could prove useful.
The fact that birth weight was missing from nearly one-fifth of universally filled antenatal cardsindicates poor motivation and/or oversight of clinical staff attending neonates delivered in health facilities. In addition to poor documentation, evidence of suboptimal care was prevalent. Incorrect dosing of antibiotics occurred in a substantial proportion of cases. Improbable diagnoses such as “gastroenteritis” were found on more than one occasion in chart notes, the latter being a highly unlikely diagnosis in neonates, especially in this setting where exclusive breastfeeding is almost universal through the neonatal period of one month; this is concerning since feeding intolerance and vomiting at this age could portend a much more serious diagnosis such as trachea-oesophageal fistula. The notably higher proportion of antenatal cards with HIV test results compared with other test results suggests that training, motivation, and efforts to make tests kits available on the ground can result in improved performance; there have been concerted efforts to promote prevention of maternal to child transmission in Kilimanjaro in recent years with highly successful results .
Missing laboratory tests, such as blood group and Rh factor in mothers and bilirubin in neonates is likely explained by lack of assay availability in many cases. The absence of anything beyond the most basic laboratory facilities at most of the hospitals in this study prevents proper diagnosis and management of key frequently encountered conditions. Since prematurity and infections contributed to most of the hospitalizations, availability of blood culture systems and measurement of full blood picture are critical to appropriate management of these neonates, but most often unavailable [28–30].
However, we showed that even when supplies and diagnostics are available, they are not utilized appropriately, as in the case of glucometers, oxygen and intravenous fluids. The relative lack of expertise in neonatal care almost certainly contributes to these challenges; WISN for nurses tended to be high, but for clinicians was much lower, indicating a need for more specialized training in this area . Similar lack of expertise in managing antenatal care as well as in managing sick children in hospital has previously been documented in Tanzania [14, 32–35]; therefore it may not be surprising that neonatal care faces similar challenges.
While declines in child mortality rates in sub-Saharan Africa are encouraging, these declines are likely to stagnate if quality of neonatal care is not addressed. Achieving MDG 4 remains a challenging task in rural Tanzania, [36, 37] and sub-Saharan Africa . Over the last 5 years declines in infant and child mortality rates have stagnated [39, 40]. Our data demonstrate that improvements will need to cover several areas, but that better care may be delivered even within the confines of currently available staff and equipment if there are improvements in training, motivation and standard operating procedures/guidelines to address key issues in neonatal care.
Our study had several limitations. First, sampling of facilities was not random, though it included the entire group of district and designated district hospitals in the region. Second, the limitations of tools to measure the quality of care are well known; direct observation is likely to bias performance (the Hawthorne effect) and a record of care is not the same as care itself. That having been said, the fact that facility awareness and patient enrolment took place typically on the same day, bias related to altering quality of care because of direct observation is substantially reduced, which is a strength of the study. Third, resource constraints did not allow for a longer period of observation that would have produced larger numbers and more robust findings, however our data comprehensively assessed the most important aspects of care for common conditions found in neonates and the equipment available at each participating facility. Finally, this study did not evaluate the care of healthy neonates with no identified problems and did include neonates born at home or in smaller health care facilities, such as dispensaries. We assumed infants with high-levelcare needs were referred to the zonal referral hospital; the infants in this study reflected a level of mild-to-moderate illness severity for which we expected that their care generally should be able to be managed at a district hospital level.