Participants

Neonatologists working in neonatal intensive care units (NICU) in the Philippines were contacted through a database provided by the Philippine Pediatric Society. For each hospital, a single neonatologist was identified to respond on their institution’s behalf. In some cases, one clinician was asked to complete the survey more than once, on behalf of multiple hospitals, as they were the only neonatologists working at those institutions. Up to three reminder emails and a single text message were sent to non-responders.

Evaluation instrument

A 34-item SurveyMonkey questionnaire was developed. The survey was written and conducted in English, an official language of the Philippines.24 25 Participation was voluntary, and responses were anonymous. The survey was conducted between January and March 2018.

The survey was structured in three main parts. The first requested information on the characteristics of the hospital including numbers of overall and preterm births. The second focused on availability of resources including number of beds, mechanical ventilators and surfactant. The third requested information about hospital policies and practice relating to resuscitation of EPIs. Respondents were asked how often resuscitation would be provided for infants born at a given gestation.

Questions consisted of Likert scale responses, yes/no, multiple choices and open-ended questions. For questions asking the frequencies of particular occurrences, ‘never’ and ‘rarely’ responses were grouped into a single category, as were ‘often’ and ‘almost always’ in our analysis. Clinicians were also asked about factors influencing decision making.14

Lastly, the survey asked for sociodemographic information from respondents including level of experience, religion and level of religiosity.

Statistical analysis

Data were analysed using the Statistical Package for Social Sciences (SPSS) v25.0.26 Bivariate analyses to assess association between categorical outcomes were performed using Fisher’s exact test. Mann-Whitney U and Kruskal-Wallis tests were used for assessing differences between continuous variables. We analysed separately survival estimates for Level III/IV hospitals, as these would have most experience of caring for EPIs.

We assessed the frequency by which each institution would initiate resuscitation for infants of a given GA. We performed a post hoc analysis using Fisher’s exact test to assess if responses differed between different types of institutions at either 23/24 weeks’ gestation or 27/28 weeks’ gestation.

Participants

Of the 228 hospitals providing neonatal care in the Philippines, we received 103 responses, yielding an overall response rate of 45%. Responses were submitted by 83 different neonatologists (14 entered data for more than one hospital). Most responses were from city hospitals, and approximately two-thirds were from private hospitals (table 1). The majority of respondents (75%) had more than 10 years’ experience of working in NICU.

Births and resource availability

Respondents from higher level hospitals reported a larger number of beds and ventilators (online supplementary appendix 1). Compared with private hospitals, public hospitals reported a significantly higher number of births and preterm births, a larger number of beds but a lower number of ventilators per 100 preterm births.

Most hospitals reported experiences of limitations in available ventilators at least some of the time (table 2). While 32% of private hospitals reported being ‘often’ or ‘always’ at full capacity for ventilators, this was reported by 85% of public hospitals (p<0.001).

Forty-one per cent of hospitals reported that surfactant for respiratory distress syndrome was available only if parents or a charity were able to pay for it. Surfactant was more likely to be ‘always’ available in higher level hospitals (p<0.05) and more often available in city hospitals than provincial/district hospitals (p<0.05).

The most common responses to ventilator shortages were for the family to rent a ventilator, to attempt hand-ventilation, or to transfer the infant to another facility (table 3). Most hospitals reported no specific limitations on which preterm infants receive surfactant. Of the hospitals that did report a limiting factor, most cited the family’s financial capacity.

Costs of care

The median reported costs to parents for different services are shown in online supplementary appendix 2. The costs of daily NICU care, ventilator rental and surfactant administration were significantly higher in private hospitals.

Survival rates

The estimated rate of survival for preterm infants increased with GA (figure 1), though there was wide variation in the estimates. Respondents consistently estimated higher survival rates for preterm infants cared for in private institutions than in public hospitals. City hospitals reported higher estimated survival than district/provincial hospitals at all gestations except at 23–24 weeks’ GA (online supplementary appendix 3).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Estimated survival rates (if actively treated) at different GA categories in Level III and IV hospitals. GA, gestational age.

Initiating and limiting resuscitation for EPIs

A majority of respondents reported using the American Academy of Pediatrics Neonatal Resuscitation Program (NRP) guidelines (online supplementary appendix 4).27 Fourteen hospitals reported using a local guideline developed by their department or institution. When asked about resuscitation at 23–24 weeks’ GA, 14% indicated that at their institution infants would ‘always’ or ‘often’ be resuscitated, while 66% indicated that resuscitation would ‘never’ or ‘rarely’ be initiated (figure 2). There was no significant difference between sectors, region or hospital level in the frequency of resuscitating at 23/24 weeks’ gestation (online supplementary appendix 5A). At 25–26 weeks’ gestation, 41% of hospitals would ‘always’ resuscitate, while 21% would ‘often’, and 23% would ‘sometimes’ resuscitate.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

Frequency of initiating resuscitation for a given gestational age.

At 27–28 weeks’ GA, 84% of respondents indicated that they would always resuscitate (figure 2). The reported reasons for non-resuscitation in this group are listed in online supplementary appendix 5B.

Resuscitation was almost always provided at higher GAs (97% and 99% in the 29–30 and 31–32 weeks’ GA groups, respectively).

The most frequently cited reason influencing clinicians’ decision to limit resuscitation was ‘respecting parents’ wishes’ followed by ‘probability of infant death’ and ‘clinician’s morals’ (online supplementary appendix 6).

Most respondents reported also using birth weight as a threshold for initiating resuscitation if the GA were uncertain (online supplementary appendix 7). The median birth weight threshold reported was 500 g (range of 400–800 g).

No significant associations between willingness to offer resuscitation and participant characteristics were observed.

Resource limitations

The number of NICU ventilators per 100 preterm births (a metric we devised for assessing resource capacity) showed striking differences between the private and public sector. The reported differences in availability of neonatal beds, ventilators and surfactant were generally consistent with the intuitive assumption that large, private, urban centres would be better resourced than small, public, rural centres, respectively.

Hospitals reported wide-ranging courses of action taken in the face of resource limitations. When ventilators are at full capacity, the most commonly cited course of action was for the family to rent a ventilator, which may not be possible for poor families. It is possible that in such cases, hospitals resort to other cited responses such as hand ventilation, or attempt to identify a source of charitable funding for the family.28 Approximately one-third (35%) of respondents cited financial capacity of the family as a limiting factor in the provision of surfactant. The reported cost of a single dose of surfactant represents approximately 9% of the average per capita annual income in the Philippines.19 For some families, incurring the costs associated with extreme prematurity may lead to financial catastrophe.

Previous studies in other LMICs have documented the financial capacity of parents as a limiting factor in resuscitation of their preterm infant. Qualitative studies from neonatal units in India have reported that families’ motivations for withdrawing treatment were often based on costs, and that clinicians would only resuscitate EPIs if the family was willing to pay the entire bill themselves.16 29

We found differing estimated survival rates between different hospital types. Rates were significantly lower in public hospitals compared with private hospitals for all GA bands (figure 1). Lower survival rates in public hospitals might relate to the different number and capacity of ventilators reported in our study (tables 2 and 3). Poorer outcomes might also relate to other factors, including staffing levels, experience, rates of nosocomial infection or overcrowding.25 30 It might also conceivably relate to different risk factors in women presenting to public institutions (eg, reduced antenatal care and fetal growth restriction relating to maternal malnutrition).31 32

Resuscitation

Our survey identified that there is no single widely adopted guideline in the Philippines for resuscitation of EPIs (online supplementary appendix 4). The most commonly cited guideline, reported by 54% of respondents, was a US guideline. The NRP 7th Edition identifies the grey zone for resuscitation between 22 and 24 weeks GA, and does not take into consideration constraints on resources.27 These guidelines may not be easily applicable to the Philippine context, and indeed, local practice seemed to differ from the NRP recommendations. In Philippine hospitals, resuscitation was generally considered to be an option from 25 weeks’ gestation.

Thresholds reported in several other LMICs resemble our findings. Resuscitation is generally only considered beyond 25 weeks GA in South Africa, Lebanon and Malaysia.12 14 17 In contrast, studies from elsewhere report thresholds at later GAs: 26 weeks in El Salvador, 28 weeks in India and 31 weeks in Mongolia.13 15 29

As expected, in the absence of national guidelines, we found significant variations between institutions. Particularly at lower GA (23–26 weeks), we found a striking degree of heterogeneity, with some centres reporting that they would never or rarely resuscitate, while others reported usually or always providing resuscitation. Practice appeared more consistent for infants on reaching the 29–30 weeks’ GA band. At 27–28 weeks, some centres cited parents’ inability to pay as a reason to withhold treatment, while others indicated that this was based on anticipated poor outcome.

The divergence of practice seen across institutions could be due to differences in resource capacity. However, we did not observe differences in the reported rate of resuscitation between the different hospital types.

The spectrum of approaches (resuscitation offered for some infants at 23 weeks and withheld for some infants at 28 weeks) may represent a wide grey zone in which resuscitation is provided predominantly according to parental wishes. Approximately 2/3 of respondents reported that ‘parents’ wishes’ and ‘financial cost (for family)’ often or always affected decisions to limit resuscitation (online supplementary appendix 6). This contrasts starkly with an earlier survey of clinicians in six Pacific countries. In that study, only 26% of clinicians in Malaysia and 1% of clinicians in Japan reported ‘financial cost (for family)’ to often or always affect decisions to limit resuscitation.14 Of concern, our study suggested a higher prevalence of litigation fear influencing resuscitation decisions (29%) compared with other Pacific rim countries.

Limitations

We had modest response rates, which may affect the generalisability of our results, though the proportion of private and public hospital respondents in our study was close to the nationwide ratio. We had a higher proportion of level III/IV hospitals than the nationwide distribution (approximately 25%), yet many level I/II hospitals lack newborn care services and were not invited to participate in our study.23

Each hospital’s response was restricted to the views of one neonatologist responding on its behalf. Respondents may have answered some questions according to their own practice, which in the absence of clear guidelines may differ from their colleagues. We relied on participant’s reporting of outcomes (ie, estimated survival rate), yet such recollections or impressions of outcomes may be inaccurate or biased. We were not able to verify the accuracy of such estimates due to a general lack of evidence on outcomes for EPIs in the Philippines.