Neonatal mortality among preterm infants admitted to neonatal intensive care units in India and Pakistan: A prospective study

To explore potential reasons for differences in preterm neonatal mortality in neonatal intensive care units (NICUs) in India and Pakistan.


| I N TRODUC T ION
Pakistan has one of the highest neonatal mortality rates in the world at 42/1000 live births, far surpassing the reported rates in neighbouring countries such as India (23/1000), Bangladesh (17/1000), Nepal (20/1000) and Sri Lanka (5/1000); even Afghanistan has a lower reported neonatal mortality at 37/1000. 1 In 2019, the estimated overall world neonatal mortality rate was just below 18/1000, and some high-income countries have neonatal mortality rates approaching 2/1000. 1 Preterm birth is a major contributor to the high mortality rates in many low-and middle-income countries, even among late preterm infants who generally survive in high-resource areas. 2 The Global Network for Women's and Children's Health Research has maintained a registry of pregnancies and birth outcomes in specific geographical areas of seven countries, including Pakistan, since 2009 in order to understand trends in pregnancy outcomes and factors associated with those outcomes. 3Focusing on a rural area of Pakistan, the neonatal mortality rates were at least twice those found in the other Global Network sites in India, Africa and Central America. 4ne conclusion from data within the Global Network registry was that the medical care for newborns in the area of Pakistan studied was very poor and contributed to the high neonatal mortality, although very low levels of maternal education and poor maternal nutrition also likely contributed. 4e recently completed a study on causes of preterm infant neonatal death in India and Pakistan. 5In assessing the mortality rate of preterm infants admitted to intensive care units (NICUs) in India and Pakistan, we found the mortality rate was substantially higher among preterm infants in Pakistan than in India and many of the deaths were likely preventable. 6To better understand the mortality differences, we evaluated the data that could be interpreted as related to the availability and/or quality of care.Specifically, given a clinical diagnosis, we wanted to determine whether that infant received the tests and treatments that should be standard care for infants with that diagnosis.We also evaluated length of stay for neonates admitted to the NICUs.

| Study design
This study compared data derived from an NICU in a single large (500 beds) children's referral hospital in urban Karachi, Pakistan, with data derived from three smaller NICUs (approximately 30 beds at each unit) within a university system in rural Davangere, India.Women with preterm deliveries (<37 weeks' gestation) were identified and admitted to obstetric units which served these NICUs.
Data were collected for all preterm infants admitted to the respective NICUs during the study period, from July 2018 to February 2020.Trained research staff collected data on each admitted infant at least once daily using standardised forms.All infants were followed until 28 days of age or death, including infants discharged alive prior to 28 days, who were followed until 28 days, usually with home visits or telephone calls with the parents.

| Statistical analysis
First, we examined differences in neonatal mortality across the two sites (India and Pakistan).We compared the percentages of infants enrolled in the study and admitted to the NICUs in both sites who died at less than 28 days of age using chi-square tests.Due to skewness of the distribution, the number of days in an NICU for the two locations was compared using median tests.Differences by site were analysed for all infants and then separately for infants according to birthweight grouping: (1) 500-999 g, ( 2) 1000-1499 g, ( 3) 1500-2499 g and (4) ≥2500 g.
Next, we investigated how often infants in the two countries (India, Pakistan) were assessed using different clinical investigations to identify and diagnose medical conditions.We conducted chi-square tests to compare sites according to the percentages of infants who were assessed with the following clinical interventions: blood group, Rh factor, complete blood count test, bilirubin, random blood sugar, blood culture, chest X-ray, abdominal X-ray, cranial ultrasound and cerebral spinal fluid (CSF) test.
For several clinical diagnoses, standard of care would suggest that specific tests be performed to confirm that clinical diagnosis.Therefore, we examined differences by site in clinical investigations and/or treatments provided for specific clinical diagnoses, including respiratory distress syndrome (RDS), early or late onset sepsis, pneumonia, intraventricular haemorrhage (IVH) and necrotising enterocolitis (NEC).

| R E SU LTS
Overall, 4030 preterm infants were screened for eligibility (Figure 1).Of those infants, 3471 were eligible and enrolled in the study.Among infants enrolled, 1673 (48%) had at least one NICU admission.
Table 1 presents the characteristics of the mothers of the infants in the study.Among mothers of all preterm live births, most (81%) mothers were between 20 and 30 years of age.Two-thirds of these mothers had 5-12 years of formal education and nearly all (95%) listed their occupation as homemaker.Thirty-nine per cent of all mothers were gravida 0 and 50% were gravida 1-3.
Next, we assessed characteristics of study infants (Table 2).Six per cent of the infants were born at less than 28 weeks gestation, 48% were female, and over half (57%) had birthweights 1500-2499 g.
Of the infants admitted to an NICU, 261 (24%) in India and 340 (58%) in Pakistan died before 28 days of age (P < 0.001; Figure 2).Neonatal mortality was significantly higher in the Pakistan than in the India site for each of the three smallest birthweight categories: 500-999 g (P = 0.004), 1000-1499 g (P < 0.001) and 1500-2499 g (P < 0.001).There was no statistically significant difference in the percentages of neonatal deaths for those born at ≥2500 g (P = 0.203).
Overall, the length of stays in the NICU was statistically significantly shorter in Pakistan than in India (P < 0.001).The median (interquartile range [IQR]) length of NICU   The lengths of stay for each of the birthweight categories ≥1000 g were significantly shorter in Pakistan than in India (P < 0.001); however, there was no statistically significant difference in median length of NICU stay for infants at the lowest birthweights (500-999 g; P = 0.086).The percentage of clinical investigations performed by site is shown in Table 3.For every test except for abdominal X-ray, the overall percentage of infants who received that test was significantly lower in Pakistan (P < 0.01).For example, complete blood counts were performed on 93.2% of infants in India but only 35.2% in Pakistan.Blood cultures were performed for 51.1% of the study babies in India but only 15.4% in Pakistan, and chest X-rays were performed for 65.7% of the infants in India but only 9.7% in Pakistan.
Next, we examined clinical investigations by diagnosis and birthweight (Table 4).For every diagnosis considered, a higher proportion of Indian infants received the appropriate test.Specifically, the diagnosis of RDS should require a chest X-ray.In India, 90.8% of the infants with this diagnosis received a chest X-ray, whereas in Pakistan, only 10.1% with a clinical diagnosis of RDS received a chest X-ray.Neonatal sepsis should be evaluated by a white blood cell (WBC) count and a blood culture.In India, 99.4% of infants clinically diagnosed with sepsis had a WBC and 77.3% had a blood culture compared with 52.1% and 23.3% of infants clinically diagnosed with sepsis in Pakistan.For infants with a clinical diagnosis of IVH, all infants in India received a cranial ultrasound compared with only one of two infants with a clinical diagnosis of IVH in Pakistan.Similarly, for pneumonia, 90.0% of infants in India with that clinical diagnosis received a chest X-ray and 54.8% of infants with a diagnosis of NEC received an abdominal X-ray; these conditions were almost never tested for or diagnosed in the Pakistan NICU.
Similarly, we examined treatments used in infants who had each diagnosis (Table 4).With a diagnosis of RDS, the standard treatment would normally include continuous positive airway pressure (CPAP) and oxygen.These treatments were used more frequently than in India than in Pakistan.However, because pneumonia was rarely diagnosed among infants admitted to the NICU in Pakistan, only two infants received antibiotics for this condition, whereas antibiotics given were given to all newborns in India diagnosed with pneumonia.While most infants with a diagnosis of sepsis were treated with antibiotics in Pakistan, all infants diagnosed with this condition were treated in India.Because many fewer infants in Pakistan were diagnosed with sepsis, many fewer infants actually received antibiotics for this condition.

| Main findings
The PURPOSE data confirm prior observations related to the reduced availability and possible poor quality of medical care for preterm neonates in Pakistan and the relation of this to the very high neonatal mortality rates in that country.In comparison with our prior study that was in a rural area, the PURPOSE population was urban and the nursery studied was in a large children's hospital. 4][9] For preterm infants of similar birthweight, gestational age and gender admitted to NICUs in India and Pakistan, the neonatal mortality rate was twice as high in Pakistan.This discrepancy was consistent across all preterm birthweight groups.The difference in care utilisation that was most apparent was the difference in mean length of stay, 2 days in Pakistan compared to 10 days in India.Also apparent were the different rates of testing, both overall and for specific conditions.Tests generally considered the standard of care for preterm infants and for infants diagnosed with specific conditions, were performed far less commonly in Pakistan than in India.

| Strengths and limitations
This study had a number of strengths and weaknesses.Among the strengths were the large number of preterm infants studied, its prospective nature and that it was performed in two locations with similar data collection instruments and research staff training.Among the weaknesses were that as it was performed in one location in each country, generalisation of the results to other areas of each country cannot be done.However, the high preterm neonatal mortality in Pakistan seen in this study is compatible with our previous observations of high neonatal mortality in another area of Pakistan and the published national data from the World Health Organization. 4,10

| Interpretation
The lack of clinical testing in Pakistan appears to have important implications for diagnosing or failing to diagnose specific conditions.In Pakistan, there were virtually no infants diagnosed with pneumonia, IVH or NEC, and a much smaller percentage of neonates was diagnosed with sepsis than in India, likely because the appropriate tests were not done.Nearly three-fourths of the infants in Pakistan were diagnosed with RDS, mostly without an X-ray.We hypothesise that in the absence of appropriate testing for sepsis and pneumonia, IVH and NEC, the fallback clinical diagnosis was often RDS.The hospital in Pakistan receives 600 infants daily in its emergency room, and NICU care is in high demand.Because this hospital caters for a poor population and receives referrals from over the whole of Sindh province, especially from rural areas, it often lacks important staff, equipment and financial resources to provide care for the patients referred.This may explain the reason for the short lengths of stay in Pakistan.While discussing the issue of the low number of diagnostic tests in Pakistan, the doctors at this hospital often mentioned that X-ray machines were often nonfunctional.They also noted that many doctors did not perform cranial ultrasounds despite the provision of an ultrasound machine by the PURPOSE project along with training on its use.They mentioned that because treatment for IVH was not generally available, the results of the test would not have altered clinical care.In general, the doctors felt that because of limited resources, the attending doctor used his or her best judgement to reach the diagnosis without ordering tests.They believed the biggest drawback to providing appropriate care was the lack of resources including staff, beds and equipment.

| Conclusions
This study found a high mortality rate among preterm infants admitted to NICUs in India and Pakistan.The rates were especially high in the Pakistan site, which also had lower availability of care than in the India site.Given the

F U N DI NG I N FOR M AT ION
This study was funded by a grant from the Bill & Melinda Gates Foundation.

C ON F L IC T OF I N T E R E S T S TAT E M E N T
None declared.

DATA AVA I L A BI L I T Y S TAT E M E N T
De-identified data are available upon reasonable request from the authors.

E T H IC S A PPROVA L
The study was approved by the Ethics Review Committees at participating institutions (JJM Medical College, Davangere, India; KLE Academy of Higher Education and Research,

F
I G U R E 1 Selection of study population.Total neonatal intensive care units admissions (n = 1727).T A B L E 1 Characteristics of the mothers of all preterm live-born study infants and of infants admitted to neonatal intensive care units (NICUs) in India and Pakistan.
stay was 2.0 (1.0-5.0)days in Pakistan compared with 10.0 (5.0-17.0)days in India.Distributions of the number of days in the NICU by birthweight group are shown in Figure 3.

F I G U R E 2
Percentage of neonatal deaths <28 days by birthweight group for infants admitted to neonatal intensive care units in India and Pakistan.

F I G U R E 3
Neonatal intensive care unit (NICU) days by birthweight group for admitted preterm infants in India and Pakistan.

All preterm live births enrolled Preterm live births admitted to NICU
T A B L E 2 Characteristics of all preterm live-born study infants and of infants admitted to neonatal intensive care units (NICUs) in India and Pakistan.Abbreviation: SD, standard deviation.
Clinical investigations for live-born study infants admitted to neonatal intensive care units (NICUs) in India and Pakistan.
T A B L E 3 Investigations and treatments by clinical diagnosis for infants admitted to neonatal intensive care units (NICUs) in India and Pakistan.SS and RLG conceived of the analyses and wrote the first draft.SS, SST, SSG, SD, GG and EMM developed the protocol.SST, SSG, SD, GG, NGN, VBK, LGCP, SES, SY, MSS, SR, MK, MS and MH implemented the study.SS, KH, SST, SSG, SD, GG, VBK, SY, MS, EMM and RLG monitored study progress.KH, CMB and EMM performed study analyses and interpretation.All authors reviewed and approve the final article.
T A B L E 4Abbreviations: CPAP, continuous positive airway pressure; GA, gestational age.majority of the preterm infants are late preterm, availability and quality of care should reduce mortality among this group which contributes to overall high neonatal mortality in LMICs.AU T HOR C ON T R I BU T ION S