The main role of the baby's skin is to provide a barrier which prevents infection, the loss of water from the body, and penetration of irritants and allergens. These functions depend on the maintenance of skin integrity and pH balance. Babies are born with a pH of 6.4 which reduces over three to four days to around 4.9 [1]. A baby's skin has a less developed epidermal barrier than adults and thus is more prone to damage; recent research suggests that the stratum corneum of infants becomes 'adult-like' only after one year of life [2]. The immaturity of babies' skin creates the potential for a number of skin problems, including atopic eczema, infant Candida, cradle cap, baby acne and napkin dermatitis [3]. These problems emphasize the importance of appropriate skin cleansing routines.

The guidelines, 'Routine postnatal care for women and their babies' [4], in the UK, recommend that cleansing agents added to bathwater should be avoided in the early postnatal period. In contrast, The American Association of Women's Health, Obstetrics and Neonatal Nursing (AWHONN) [5] produced clinical guidelines that recommend the use of warm tap water for routine bathing with the option to use mild cleansers that have a neutral pH (5.5-7.0). However, there is a lack of evidence on which to inform practice for the term newborn baby. A survey of maternity units in the North West of England [6] reported that a wide range of products were used by women. Moreover, a systematic review of skin care regimes, in the well term newborn, revealed no prospective trials that met the authors' inclusion criteria [7]. As such, there are no UK evidence-based guidelines about neonatal skin care [8].

The Royal College of Midwives [9] called for further research in this area. A recent European round-table of Dermatologists also acknowledged the dearth of evidence for skin care provision within 6 weeks of birth [10]. The absence of randomized controlled trials comparing different skin cleansing routines is an important issue because of the readiness to use wash products among mothers [11].

Water is the basic component of any cleansing routine. In many countries, despite the lack of strong evidence in one direction or the other, water alone has been considered the least harmful of all alternatives [4]. However, water may not be the optimal skin cleanser for newborns. The buffering capacity of water is being questioned, as it might increase skin pH; after washing with water the skin surface pH may rise from 5.5 to 7.5. This brings the pH to a level that maximizes the activity of the skin proteases and therefore enhances skin barrier breakdown [12]. The other problem with water alone is that it is a poor cleanser as it does not remove fat-soluble substances such as feces and sebum [13]. On the extreme, over-exposure to water leads to higher trans-epidermal water loss (TEWL) and a weakened skin barrier [12]. An appropriately formulated cleansing product may reduce these potential problems but would need to be carefully evaluated.

Prior to the commencement of our study (in 2008), we identified only two small trials that compared baby bathing with a cleansing product to water. Both were available in abstract form only, so our assessment of methodology and interpretation of findings was necessarily limited [14, 15]. Following a small-scale study involving 57 infants, Garcia Bartels' conclusion was that skin barrier development of term newborns was not adversely effected by bathing with a mild detergent cleansing product. Galzote [14], using a different wash product to Bartels [15], found that skin dryness was reported more often in the 'water only' arm. These trials were not large enough to provide definitive guidance.

One concern during skin care is atopic eczema. This is a disease that arises as a result of the interaction of environmental factors (such as harsh soap & detergents) with variants in several genes [16, 17]. Atopic eczema starts as a weakness of the skin barrier [16–22]. This breaks down allowing allergens to penetrate the skin and interact with the immune system. Some of the damage is caused by enzymes in the skin; proteases. Proteases are pH sensitive enzymes with optimal activity at 7.5 to 8.0 [20, 21]. Harsh soap and detergent raise the pH of the skin to within this range thereby increasing the protease activity in the skin and potentially leading to severe skin barrier breakdown. Washing with a detergent which can damage and break down the skin barrier may lead to an atopic flare in susceptible infants. This may be important in bathing practices for newborn babies, but this possibility has not been accounted for in previous work.

As there was limited previous research in this area and the available studies did not report key details of methodology, careful preparation was required for an adequately powered investigation. We therefore conducted a qualitative, exploratory study [11] to gauge support for a trial of bathing practices for term newborn babies, in the UK. The results highlighted the inconsistencies in information provided to parents and in current newborn bathing practices. It also demonstrated that health professionals and parents were likely to support a trial.

Therefore, we conducted a pilot randomized controlled trial to compare a skin cleansing agent (specifically formulated for use on newborn skin), to water. We hypothesized that an optimally formulated infant skin-cleansing product improves skin barrier function (measured by TEWL) in newborn babies when compared with bathing with water and cotton wool.

The pilot was designed to address the following uncertainties in the design of the full study: the practicability of using TEWL on newborn babies; the best outcome to use (TEWL, pH or hydrometer), the best locations to use (arm, leg or abdomen), the optimal time point for measurement of the primary outcome and the value of key parameters in the sample size calculation (the magnitude of difference that would be important to detect between the two groups and the precision of our measurements).

Of 225 mothers who were approached to participate, 100 accepted. Figure 1 illustrates study recruitment, participant follow-up and reasons for declining. We did, however, conduct a post-hoc analysis according to assessment location. For all measures, we found no clear evidence of differences in reliability between locations on baby (arm/leg/abdomen) or place of assessment (home versus hospital). Reliability was good during hospital measurements and was maintained when we conducted follow-up assessments in the home. This was crucial to the success of the pilot, as our original plan to conduct all assessments in the hospital was unacceptable to women. The reliability of the tests was good. At all times and body locations the intra-class correlation of repeated measurements was at least 0.92, with an average difference of approximately 0.35, and most differences less than 2.0. Furthermore, the assessing midwife observed that babies being assessed at home were calmer than those in the hospital. Given the sensitivity of the TEWL instrument, it is therefore likely that more accurate readings were recorded as individual assessments were easier to take and position of repeat assessments was easier to locate. Figure 1 shows the number of assessments at home and in the hospital.

An important reason for conducting the pilot was to determine compliance, in terms of the allocated trial arm and adherence to the bathing guidance. Compliance was shown to be an issue. Women's diaries and verbal reports indicated that between 3 and 4 weeks post birth, mothers perceived their baby's skin to be becoming dry. Although we requested that women refrain from using additional products on their babies' skin, this was the time in which they were most likely to introduce products into bathing regimes. As a consequence, there were 53 babies using products at the time when our primary outcome measure was being assessed; this was similar in each treatment group and despite the fact that women remained committed to completing the study. The number using products may in fact be an underestimate as some women may not have revealed the protocol violation. Women appeared to comply with the minimum bathing occasions of three; the median number of bathing occasions per week were 3 (range 2-7) for both groups.

The primary purpose of conducting this pilot trial was to inform a robust definitive trial of water and cotton wool versus a mild wash product for newborn babies. We present one of the largest baseline datasets on newborn skin assessments to date; information which is pivotal to the design of future studies in this field. However, when we set out to design the trial there was little published information on methodology or data from studies on newborns to assist in trial design. Although it is important to report what works in a study (as is usual in reports of a main trial), it is also important to share what does not work. There are many ways to design this type of trial and the field will only advance if the processes of trial design are shared transparently. In doing this, we reveal a number of important process issues, that would not normally be available to readers.

In our qualitative study [11], we asked women to indicate what they thought would be the optimum time to be approached about participating in this trial. Views were mixed; some suggested that the antenatal period was best, while others recommended the postnatal period. In this pilot study we decided to give women information in the antenatal period and re approach them in the postnatal period. The reality was that the majority of women had not absorbed the information prior to giving birth and/or was not ready to make a decision until the baby was born. It was only after giving birth to a healthy baby, and being faced with a real decision, that, for most women, the information was internalized and informed consent could be obtained. Although postnatal consent is appropriate, no woman objected to being given the information in the antenatal period, therefore the approach adopted was acceptable.

We conducted our pilot trial based on a superiority hypothesis. However, although the study was not designed to carry out a hypothesis test, examination of the data suggested that the hypothesis was not plausible. There was no convincing trend for superiority for any measurements on any part of the body. Furthermore, there was no clear evidence of any differences in any direction. The size of any of the small differences observed was deemed of little clinical importance by the Trial Steering Committee and Data Monitoring Committee. Therefore, a trial designed on the principles of non-inferiority appears most appropriate. The trial design therefore should be to generate data concerning the hypothesis that this mild skin cleansing product is not inferior to bathing with water only in its effect on skin barrier function. Moreover, as there was no difference between body parts, it seems reasonable, in future trials, to analyze an average assessment score.

There was no evidence of consistent differences between those babies with and without a family history of atopic eczema. If those with a family history of atopic eczema had shown more evidence of barrier dysfunction, this may have led us to design a trial based on this population only. Such a trial would be attractive because if a trial recruited a group with a propensity to disease and found no difference between treatments it would be unlikely that we would find a difference in a 'healthy' population. Given our results, however, it appears appropriate to include both groups, with stratification for family history. On a practical level this is more feasible, as those with a family history of atopic eczema were particularly difficult to recruit.

The importance of assessing compliance was highlighted in this pilot. Although women in our qualitative study [11] and at recruitment for this pilot study told us that they were happy to conform to protocol, compliance was an issue, making the findings difficult to interpret. This is one possible explanation for not observing a treatment difference. Some women, particularly those having their first baby, may not have been able to anticipate the difficulties of daily routines with a newborn baby. Given that parents introduced products around 3-4 weeks, and it is impossible to enforce or ensure compliance, it is more appropriate to have a primary endpoint prior to this.

This study provides an important exemplar of the importance of conducting a pilot study, particularly when there is a dearth of prior knowledge. The findings have indicated that the research processes, trial management and chosen primary outcome (i.e. TEWL) were appropriate. The feasibility of the main trial was also established. The trial management group and the independent Data Monitoring Committee have reviewed the process and data. A small number of important amendments to the trial have been made as a result of the findings of the pilot study. Our experience illustrates some contrasts between our qualitative study about the issues involved in a potential trial and what actually happened. For example the timing of information-giving was refined from the suggestions arising from the qualitative study. This provides an instructive example of the need to develop a large trial in several stages.

Three relevant RCT's, two by the same authors, were published after the completion of our pilot trial. Bartels study [26] tested the hypothesis that neither twice-weekly washing nor bathing would harm the natural adaptation of the skin barrier with respect to long-term effects on skin function in healthy newborns. The bathed group showed statistically significant lower TEWL on the buttock and higher hydration on abdomen and forehead compared to the wash group at day 28. The authors claim that both skin care regimes do not harm the adaptation of the skin barrier in healthy newborns in the first 24 hours of life. The second study [27], aimed to test the hypothesis that twice-weekly bathing with a commercially available baby wash gel and additional baby cream would not harm the natural adaptation of skin barrier in healthy newborns. At 8 weeks, the group using clear water and topical cream had lower TEWL measurements on their fronts, abdomen and upper legs as well as higher stratum corneum hydration on their fronts and abdomen compared with bathing with water only. The group bathing in wash gel had lower pH on all sites compared with bathing in water only at week 8. No differences in sebum levels, microbiological colonization and skin scores were found. The authors conclude that skin adaption as a barrier function was not harmed by tested skincare regimens in full term healthy infants. The final RCT [28] was a three armed trial, conducted in the Philippines, which compared a Johnson's^® baby top- to-toe ™ wash (Johnson & Johnson Consumer Companies, Inc.) with Sebamed^® baby liquid cleanser and water alone; 60 babies were randomized in each arm. Assessment measures were similar to those reported in the Garcia Bartels studies, with the addition of skin oxyhemoglobin, deoxyhemoglobin and parental satisfaction measures. The authors conclude that all three regimes are 'safe for use in infants with normal skin.' However, although these three studies provide novel information relating to term baby bathing none report a priori primary outcomes or sample sizes, which make it difficult to assess the extent to which the results could have arisen by chance. Furthermore, it is not clear how the results of either trial relate to clinically important safety outcomes since the investigators do not state what they mean by "harm". None of these published studies use home measurements. In our experience the families who complete a trial using hospital assessments are a subset of families who were committed to the study. This subset may not be representative. Such a committed group may be particularly concerned about skin and skin care. This concern may mean that their skin care practices at home may be different from other families. We have described and validated an approach that reduces the potential for selection bias.

Our study adds to existing literature by providing valuable baseline data and important information on trial processes. Our study observations were consistent with previously published papers but we believe that the way forward is to test the hypothesis in a properly designed and adequately powered non-inferiority trial.