Abstract
Background
Longitudinal data for high-risk infants in the first 2 years of life are limited, particularly in low- and middle-income countries. We aimed to describe survival, diagnostic, neurological and developmental outcomes of infants at risk of developing cerebral palsy (CP) in Bangladesh.
Methods
We conducted a prospective longitudinal cohort study of neonates with risk factors for CP admitted to a tertiary neonatal intensive care unit in Bangladesh. Outcomes at 24 months included CP, developmental delay (DD), no known disability and death. Assessments up to 24 months of age included the Hammersmith Infant Neurological Examination (HINE), Peabody Developmental Motor Scales, second edition (PDMS-2), Gross Motor Function Measure (GMFM-66), Developmental Assessment of Young Children (DAYC-2), Ages and Stages Questionnaire (ASQ-3). Survival probability was analysed using Kaplan-Meier curves and mortality risk using Cox proportional hazards models.
Results
Of 227 infants enrolled, 27% died at the median age of 41 days (range 3–403). Mortality risk was significantly associated with multiple birth (HR 6.66; 95% CI 2.08 to 21.32) and neonatal encephalopathy (grade 3 vs grade 1–2: HR 2.40; 95% CI 1.29 to 4.44). Of 125 infants assessed at 24 months, 29% were diagnosed with CP, 10% with DD and 61% had no known disability. Among infants with CP, median HINE scores were 27.5, 29.5 and 34 at 3, 12 and 24 months respectively, with 57–64% scoring <40 across these time points. Motor scores on PDMS-2 were significantly lower in infants with CP than in those with DD (p≤0.005) and those with no known disability (p<0.001) at 3, 12 and 24 months.
Conclusions
One-third of the infants died, and an additional third of the survivors developed CP, most with non-ambulant CP. These findings underscore the need for context-specific early detection and intervention strategies for high-risk infants in Bangladesh.
Keywords: Cerebral Palsy, Children, Developing Countries, Low and Middle Income Countries, Mortality
WHAT IS ALREADY KNOWN ON THIS TOPIC
Estimates of cerebral palsy prevalence in low- and middle-income countries are likely underestimated due to high early mortality among high-risk infants and historically delayed diagnosis, resulting in limited longitudinal data on early neurological and developmental trajectories.
WHAT THIS STUDY ADDS
In a longitudinal cohort of high-risk neonates in Bangladesh, one-third died before 24 months of age, most within the first 3 months of life and one-third of surviving infants were diagnosed with cerebral palsy, predominantly non-ambulant cerebral palsy and these children showed distinct early neurological and developmental trajectories compared with infants with developmental delay or no disability.
HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY
These findings demonstrate substantial survival bias and underscore the urgent need for context-specific early identification and intervention strategies to address the significant early and long-term support needs of high-risk neonates in Bangladesh and similar underserved settings.
Introduction
Progress towards the Sustainable Development Goals has stagnated since 2019 due to global factors such as the COVID-19 pandemic and geopolitical issues.1 Consequently, improving outcomes for infants with developmental disabilities in low- and middle-income countries (LMICs) has become an increasingly urgent priority. Cerebral palsy (CP) is defined as a group of permanent disorders of movement and posture that cause activity limitation due to a non-progressive disturbance in the developing brain.2 In 2019, approximately 8 million children under 5 years of age had CP globally with the highest rates observed in LMICs, including the African Region and Southeast Asia.3
The prevalence of CP in Bangladesh (3–3.4 per 1000 live births) has previously been reported as an underestimate due to strong survival bias.4 Better understanding survival bias can support the development of culturally appropriate interventions and prevention strategies to reduce mortality and improve outcomes of children with CP. Historical fluctuations in CP prevalence in high-income countries (HICs) are attributable to a range of preventable causes of CP, such as kernicterus, as well as increased survival of very preterm infants with the advent of neonatal intensive care.5 In LMICs, mortality is higher among people with disabilities compared with their peers without disabilities; children with disabilities, in particular, are more likely to die (HR 4.46; 95% CI 3.01 to 6.59) compared to their peers without disabilities.6 Therefore, examining mortality and its early risk factors among infants at risk of developing CP is necessary to understanding survival bias and its implications on prevalence estimates as survival improves.
Longitudinal studies of children with CP are now available from some LMICs, such as the study of gross motor development in Uganda.7 However, longitudinal data are predominantly reported in HICs.8,10 Historically, CP was typically diagnosed after 2 years of age, resulting in limited data on early neurological and developmental trajectories. While early diagnosis is increasingly becoming standard of care globally,11 diagnosis of CP in many LMICs remains delayed. For example, in Bangladesh the average age at the time of CP diagnosis is now 3 years of age, reduced from 5.5 years in 2015.12
Furthermore, children with CP frequently experience comorbidities including epilepsy, non-verbal communication, as well as impairment of cognition, hearing and vision. While early detection of CP enables access to early interventions within the period of optimal neuroplastic potential of the developing brain, early identification of comorbidities and severity can also inform targeted early intervention strategies.13 14 Given the lack of high-quality data on early neurological and developmental trajectories in CP globally, investigation of these trajectories can contribute to advanced understanding of early development of infants at risk of developing CP in Bangladesh and beyond.
Therefore, we aimed to report survival, diagnostic groups as well as neurological and developmental trajectories of high-risk neonates in Bangladesh.
Methods
Study design, setting and participants
This study was conducted as part of a larger prospective longitudinal cohort study, the protocol and primary findings, focused on early detection of CP, have been previously published.15 Only methods relevant to the present study are summarised here.
Eligible neonates admitted to the Neonatal Intensive Care Unit (NICU) of Mymensingh Medical College Hospital (MMCH) in Bangladesh were consecutively recruited between November 2019 and March 2020. MMCH is a postgraduate government medical college and a tertiary referral hospital located in Mymensingh, a major city in north-central Bangladesh.
Neonates were eligible if they met criteria for at least one predefined risk factor for CP which included: (1) <37 weeks gestational age, (2) clinical presentation or history suggestive of neonatal encephalopathy (NE), (3) neonatal sepsis and/or (4) severe jaundice or kernicterus. An eligibility tool was developed in consultation with neonatologists, paediatricians and physiotherapists on the study team, drawing on the best available evidence on risk factors for CP and considering standard clinical investigations available at MMCH (online supplemental Table S1).
Data sources
In the larger study, five in-person assessment time points were planned originally. COVID-19 safety restrictions necessitated adaptations to the planned study design, including the introduction of remote assessments and data collection at additional time points. Finally, 12 and 24 months assessments could be completed safely in-person. All assessment tools and time points are described in detail elsewhere. The subset of assessments and time points analysed in the current study is shown in figure 1.
Figure 1. Assessment tools used at each time point. ASQ-3, Ages and Stages Questionnaires, third edition; CP, cerebral palsy; CPDF, CP Description Form; DAYC-2, Developmental Assessment of Young Children, second edition (including PD Physical Domain and CD Cognitive Domain); GMFCS, Gross Motor Function Classification Scale; GMFM-66, Gross Motor Function Measure-66; HINE, Hammersmith Infant Neurological Examination; MiniMACS, Mini Manual Ability Classification System; PDMS-2, Peabody Developmental Motor Scales, second edition.
Study variables
Gestational age was determined using available fetal ultrasound and/or last menstrual period method. Remote assessments were completed by clinicians via telephone. All ages are reported as corrected age unless otherwise specified. NE was classified using the Sarnat staging system.16 ‘Documented’ denotes characteristics supported by verifiable medical evidence (eg, positive blood culture and/or documented antibiotic treatment for documented infection). Perinatal, parental and socio-economic characteristics data were collected at enrolment by two study clinicians.
Survival and early risk factors for mortality
Infant mortality was confirmed via parent report through verbal autopsy method.
Diagnostic groups
Diagnostic groups at 24 months were CP, developmental delay (DD) and no known disability or developmental delay.
CP was diagnosed according to the consensus definition current at the time of study design and is considered the reference standard. Diagnosis was confirmed at 12 and 24 months by a physician (TK) and physiotherapists (AtV, SA and SI) on the study team. Diagnosis fidelity was achieved through structured training and case discussions.
DD was defined as <2 SD below the mean in one or more developmental domains at 24 months. This included motor (Peabody Developmental Motor Scales Total Motor Quotient (PDMS TMQ) score <70,17 cognitive (Developmental Assessment of Young Children, second edition (DAYC-2) cognitive standard score <70,18 communication (Ages and Stages Questionnaires third edition (ASQ-3) communication area score <25,19 problem solving (ASQ-3 problem solving area score <3019 or personal social domain (ASQ-3 personal-social area score <3219).
The diagnostic group with no known disability or developmental delay included all remaining infants not meeting the criteria for CP or DD.
All neurological and developmental trajectories are reported across these three diagnostic groups.
Neurological trajectories across diagnostic groups
Neurological assessment was completed using the Hammersmith Infant Neurological Examination (HINE), a standardised tool recommended for diagnosis and assessment of motor severity of CP.14 20 Where available as part of routine care, cranial ultrasound and/or MRI were extracted from hospital records.
Developmental trajectories across diagnostic groups
A wide range of tools were used for developmental assessment.
Motor development was assessed using PDMS-2, Gross Motor Function Measure-66 (GMFM-66), DAYC-2 Physical Domain and ASQ-3 Motor Domains.
PDMS-217 is a standardised HIC normed test of motor development (0–6 years). HIC norms were used to interpret PDMS-217 because Bangladeshi or LMIC setting norms are not reported, but Indian infants have similar motor development on PDMS-2 TMQ (p=0.054).21 22 PDMS-2 has demonstrated concurrent validity with the GMFM,23 is discriminative17 and responsive to change in infants and toddlers with CP.24
The GMFM-66 is criterion-referenced and gold standard for gross motor assessment in children with CP.25 It is valid, reliable, predictive and responsive in children with CP in LMICs and HICs.25 26
The DAYC-2 Physical Development Domain (PD)18 is recommended for early detection of CP,14 and can be administered remotely or in-person,14 which met other aims associated with the larger study.
ASQ-319 is commonly used and feasible in LMIC settings. It was translated and culturally adapted for this study.
Cognitive development was assessed using the DAYC-2 Cognition Domain (CD). Normative data is not available for DAYC-2 in Bangladesh or other LMICs.18 27 Data on how standardised cognitive assessment norms compare between HIC and LMIC settings is also not available,28 therefore HIC norms were used.
Communication, problem solving and personal social development were assessed using the relevant areas on ASQ-3.
Additional variables for infants with CP
Motor type and topography were assessed using the CP Description Form widely used in CP registers globally.
Motor severity was reported using Gross-Motor Function Classification System–Extended and Revised (GMFCS E&R),29 GMFM-66,25 Mini Manual Ability Classification System (MiniMACS).30
GMFCS E&R is a five-level classification system of gross motor function in CP29 which has been previously used in Bangladesh.4 31 The GMFCS E&R 0–2 years descriptor was used at 24 months.
Motor severity of CP was dichotomised as ‘likely ambulant’ and ‘likely non-ambulant’. Likely non-ambulant CP was defined as the inability to sit independently on item 25 of GMFM-66,25 that is, item score of 0 (does not initiate leaning forward) or 1 (leans forward, does not re-erect) at 24 months.
The MiniMACS is a five-level classification system of manual ability in children aged 1–4 years.30
Bias
Developmental assessment tools normed in HICs have known limitations when used in LMIC contexts.28 To minimise bias, we used the most reliable and valid tools available globally for developmental assessment of infants with CP, and prioritised tools with the best available psychometric evidence in Bangladesh or comparable settings, and interpreted study findings cautiously in relation to local context.
Statistical methods
Clinical and demographic characteristics associated with mortality risk were identified using univariate Cox proportional hazards regression models, and survival probability over time was estimated and compared between diagnostic groups using Kaplan-Meier curves. HINE scores were analysed using previously published optimality scores.20 Developmental trajectories up to 24 months were described using mean and SD for normative data, and median and IQR for ordinal or interval data. Given floor effects in DAYC-2 standard scores and the majority of infants with CP scoring <50 on DAYC-2 PD and CD, raw scores were treated as continuous variables and used for visualisation of these trajectories. Diagnostic group differences were tested using independent t-tests (for normative data) or Kruskal-Wallis tests (for ordinal data), with Bonferroni adjustment applied where appropriate. This accounted for multiple testing and preserved the overall significance level at 5%. All analyses were conducted using R V.4.1.0.
Reporting
The study reporting is in accordance with the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines.32
Results
A total of 227 neonates at risk of developing CP were recruited. By 24 months 61 infants (27%) had died, and a further 30 infants (13%) were lost to follow-up. Of the 125 surviving infants assessed in person at 24 months, 29% were diagnosed with CP, 10% had DD and 61% had no known disability or developmental delay (figure 2). Baseline clinical and demographic characteristics of all study participants are summarised in table 1.
Figure 2. Flow diagram of study participants.
Table 1. Study participant characteristics (n=227).
| Total cohort (n=227) |
|
|---|---|
| Maternal characteristics | |
| Maternal age (years), mean (SD) | 22.4 (4.2) |
| Gravidity, n (%) | |
| 1 | 157 (69) |
| 2–3 | 55 (24) |
| 4+ | 15 (7) |
| Missing | 1 (<1) |
| Parity, n (%) | |
| 1 | 172 (76) |
| 2–3 | 49 (22) |
| 4–5 | 5 (2) |
| Missing | 2 (<1) |
| Consanguinity, n (%) | 38 (17) |
| Birth complications during previous pregnancy/ies, if any (n=70) | 35/70 (50) |
| Infant characteristics | |
| Sex, n (%) | |
| Male | 150 (66) |
| Gestational age, n (%) | |
| Term (≥37 weeks) | 183 (81) |
| Preterm (<37 weeks) | 21 (9) |
| Unknown | 23 (10) |
| Birth weight, n (%) | |
| Low (<2500 g) | 58 (26) |
| Very low (<1500 g) | 4 (2) |
| Extremely low (<1000 g) | 0 (0) |
| Unknown | 4 (2) |
| Birth weight (grams), mean (SD) | 2653 (451) |
| Multiple birth, n (%) | 4 (2) |
| Documented IUGR, n (%) | 5 (2) |
| Documented seizure, n (%) | 159 (71) |
| Documented infection, n (%) | 225 (99) |
| Delivery or birth characteristics | |
| Mode of delivery, n (%) | |
| Normal vaginal delivery | 150 (67) |
| Elective caesarean section | 11 (5) |
| Emergency caesarean section | 43 (19) |
| Instrumental delivery | 0 (0) |
| Unknown | 23 (10) |
| Birth attended by, n (%) | |
| Medical doctor | 109 (45) |
| Dai/untrained birth attendant | 63 (28) |
| Trained birth attendant | 48 (21) |
| Midwife | 7 (3) |
| Birth complications, n (%) | 178 (79) |
| Documented NE, n (%) | |
| NE grade 1–2 | 191 (84) |
| NE grade 3 | 28 (12) |
| Unknown | 6 (4) |
| Required resuscitation, n (%) | 211 (93) |
| Socio-demographic characteristics | |
| Maternal education, n (%) | |
| No formal education | 17 (8) |
| Primary | 60 (27) |
| Secondary | 127 (56) |
| Under/postgraduate | 23 (10) |
| Paternal education, n (%) | |
| No formal education | 47 (2) |
| Primary | 50 (22) |
| Secondary | 108 (48) |
| Under/postgraduate | 22 (10) |
| Monthly household income (BDT ₺), mean (SD) | 12 140 (8735) |
| Household income by urban poverty line, n (%) | |
| Below upper urban poverty line | 196 (86) |
| Above upper urban poverty line | 22 (11) |
| Unknown | 5 (3) |
BDT, Bangladeshi taka; IUGR, intrauterine growth restriction; NE, neonatal encephalopathy.
Survival and early risk factors for mortality
Among the 61 infants who died, date of death was known for 60 infants. The median age at death was 41 days (range 3–403). The risk of death was greatest in the first 3 months of life (figure 3), and was significantly associated with multiple birth (multiple vs singleton: HR 6.66; 95% CI 2.08 to 21.32), severe NE (NE grade 3 vs grade 1–2: HR 2.40; 95% CI 1.29 to 4.440), and a proxy indicator for prematurity (thin transparent skin: HR 1.87; 95% CI 1.03 to 3.40) (online supplemental Table S2).
Figure 3. Probability of survival over time using Kaplan-Meier survival curve. (A) Overall, (B) by NE grade and (C) by birth plurality. NE, neonatal encephalopathy.
All subsequent study findings are restricted to surviving infants assessed in person at 24 months (n=125) and are reported across the three diagnostic groups (CP, DD and no known disability or developmental delay).
Neurological trajectories across diagnostic groups
HINE scores at 3, 12 and 24 months are reported in table 2. Due to COVID-19 restrictions, only 31 infants could be safely assessed in person at 3 months. Almost all infants with CP scored ≤66, and the majority scored <40 on the HINE at 3, 12 and 24 months. Median HINE scores differed significantly between infants with CP and no known disability at all three time points, and between infants with CP and DD at 12 and 24 months. All infants with DD scored ≤66 at 3 months, however at 12 and 24 months the majority scored ≥73, a pattern also observed for infants with no known disability at 24 months. No neuroimaging was completed as part of routine clinical care for any infant.
Table 2. Hammersmith Infant Neurological Examination global score by diagnostic groups at 3, 12 and 24 months.
| Outcomes | Median (IQR) | Global score, n (%) | |||
|---|---|---|---|---|---|
| ≥73 | 67–72 | 40–66 | <40 | ||
| Cerebral palsy | |||||
| 3 months (n=11) | 27.5* (19–49) | 0 | 0 | 4 (36) | 7 (64) |
| 12 months (n=36) | 29.5* (20–48) | 0 | 1 (3) | 13 (36) | 22 (61) |
| 24 months (n=35) | 34* (20–52) | 1 (3) | 1 (3) | 13 (37) | 20 (57) |
| Developmental delay | |||||
| 3 months (n=4) | 55 (40–57) | 0 | 0 | 3 (75) | 1 (25) |
| 12 months (n=11) | 77† (72–78) | 8 (73) | 1 (9) | 2 (18) | 0 |
| 24 months (n=13) | 76† (73–78) | 11 (85) | 2 (15) | 0 | 0 |
| No known disability or developmental delay | |||||
| 3 months (n=16) | 59 (55–61) | 0 | 0 | 15 (94) | 1 (6) |
| 12 months (n=69) | 78 (77–78) | 64 (93) | 5 (7) | 0 | 0 |
| 24 months (n=76) | 78‡ (78–78) | 73 (96) | 3 (4) | 0 | 0 |
CP versus N: p<0.017.
CP versus DD: p<0.017.
N versus DD: p<0.017.
DD, developmental delay; N, no known disability or developmental delay.
Developmental trajectories across diagnostic groups
A summary of developmental trajectories is shown in figure 4 and the comprehensive data informing the summary figure are provided in online supplemental tables 3 and 4.
Figure 4. Neurological and developmental trajectories across diagnostic groups (n=125). ASQ-3 Ages and Stages Questionnaires, third edition (including GM and FM), CD Cognition Domain, DAYC-2, Developmental Assessment of Young Children, second edition (including CD and PD), FM Fine Motor, GM Gross Motor, HINE Hammersmith Infant Neurological Examination, PD Physical Domain, PDMS-2 Peabody Developmental Motor Scales second edition.
Motor development
Motor development assessed using PDMS-2 TMQ was significantly lower for infants with CP compared with both infants with DD and no known disability at 3, 12 and 24 months. At 3 months, PDMS-2 TMQ did not differ significantly between infants with DD and no known disability (p=0.376). By 12 months, infants with DD had significantly lower mean PDMS-2 TMQ scores than those with no known disability (p=0.002). Similar patterns were observed for PDMS-2 Gross Motor Quotient and Fine Motor Quotient scores. TMQ declined between 12 and 24 months for all three diagnostic groups. In-person DAYC-2 PD assessments showed significantly different mean raw scores for infants with CP compared with those with DD and no known disability at both 12 and 24 months. However, no significant difference was observed between infants with DD and those with no known disability at 12 months (p=0.392), or between infants with likely ambulant CP and DD at 24 months (p=0.012). For remote DAYC-2 PD assessments mean raw scores did not differ significantly between infants with DD and no known disability until 18 months (p<0.001) and again at 24 months (p<0.001). In contrast, infants with CP had significantly lower scores than those with DD (p<0.001) and no known disability from 6 months onward (p<0.001).
Cognitive development
Infants with CP had significantly lower cognition on DAYC-2 CD mean raw scores than those with DD and no known disability (p<0.001) from 6 months for both in-person and remote assessments. In contrast, cognitive trajectories for infants with DD and no known disability did not differ significantly until 24 months (p<0.001).
Communication, problem solving and personal and social development
Total scores for infants with CP on ASQ-3 communication were significantly lower than those with DD and no known disability at 12 and 24 months for in-person assessments (p<0.001). Problem solving and personal and social development trajectories, also assessed using ASQ-3, followed a similar pattern to ASQ-3 communication trajectories.
Cerebral palsy: likely ambulant versus non-ambulant
Comparison of the neurological and developmental trajectories of infants with ambulant CP to those with non-ambulant CP is summarised in figure 5.
Figure 5. Neurological and developmental trajectories of infants with cerebral palsy (n=36). ASQ-3 Ages and Stages Questionnaires, third edition, CD Cognitive Domain, CP-A Cerebral Palsy - Ambulant, CP-NA, Cerebral Palsy - Nonambulant, DAYC-2 Developmental Assessment of Young Children, second edition (including PD and CD), GMFM-66 Gross Motor Function Measure-66, HINE, Hammersmith Infant Neurological Examination, PD Physical Domain, PDMS-2 Peabody Developmental Motor Scales, second edition.
Motor type, topography and severity
Infants had predominantly spastic motor type (72.2%) with a further 19.4% who additionally had dystonia; smaller proportion of infants were classified as dyskinetic (2.8%), ataxic (2.8%) and hypotonic (2.8%). Of the 33 infants with spastic CP, n=2 (6.1%) had unilateral CP and n=31 (93.9%) bilateral CP including diplegia (n=4), triplegia (n=4) and quadriplegia (n=23).
Gross motor severity was classified as GMFCS E&R Level I (5.5%), Level II (11.0%), Level III (19.4%), Level IV (13.9%) and Level V (50.0%). Manual ability was classified as MiniMACS Level I (2.8%), Level II (16.7%), Level III (13.9%), Level IV (16.7%) and Level V (50.0%). Overall, 81% of infants with CP were likely non-ambulant.
Motor trajectories
Trajectories using mean PDMS-2 TMQ scores at 3 months did not differ significantly between those with likely ambulant CP and non-ambulant CP, mean 76.0 (SD 4.4) vs 75.6 (SD 4.7), p=0.907. By 12 and 24 months the trajectories diverged, with higher average scores observed among likely ambulant infants compared with those who were likely non-ambulant; 12 months: 70.3 (SD 12.1) versus 48.7 (SD 4.8); 24 months: 56.0 (SD 13.3) versus 40.8 (SD 2.1), p<0.001. Mean PDMS-2 TMQ scores did not differ among infants with likely ambulant CP and DD at 3 months (p=0.081), but the trajectories diverged by 12 months (p<0.001) and 24 months (p=0.008).
At 3 months, GMFM-66 scores did not differ significantly between infants with likely ambulant CP (median=13.9; IQR 13.1–17.0) and likely non-ambulant CP (median=12.4; IQR 11.9–12.9) (p=0.216). By 12 months likely ambulant infants with CP had significantly higher scores (median=35.9; IQR 32.0–38.9) than those who were likely non-ambulant (median=15.0; IQR 9.8–18.3) (p<0.001), a difference that remained significant at 24 months (median=40.0, IQR 36.8–54.9 vs median=17.6, IQR 11.9–26.7; p<0.001). At 24 months, GMFM-66 data were available for 86% of likely ambulant infants and 45% of likely non-ambulant infants with CP, compared with 100% at 12 months.
DAYC-2 PD mean raw scores did not differ between likely ambulant and non-ambulant infants with CP until 12 months (p<0.001).
Cognitive trajectories
Infants with likely ambulant CP had significantly higher mean DAYC-2 CD raw scores compared with those with likely non-ambulant CP from 12 months onward (p<0.001). Infants with likely ambulant CP had significantly lower scores in-person compared with infants with DD at 12 and 18 months (p<0.001), but not at 24 months (p=0.063).
Communication, problem solving and personal and social trajectories
Infants with likely non-ambulant CP had significantly lower ASQ-3 Communication scores. Problem solving and personal social trajectories followed a similar pattern for assessments completed in-person.
Discussion
This prospective cohort study provides the first comprehensive examination of early survival, diagnostic, neurological and developmental outcomes among neonates at risk of developing CP in a NICU in Bangladesh. These findings contribute critical new insights into early trajectories of high-risk infants and support the global shift toward earlier detection of CP and intervention strategies.
Survival and early risk factors for mortality
In this cohort, nearly one-third (27%) of infants died before 24 months, with most deaths occurring in the first 3 months of life, consistent with global reports of the highest rate of mortality in the neonatal period.33 This high mortality rate reinforces the presence of survival bias in estimates of CP prevalence from LMICs.4 34 Although globally newborn mortality has declined from 37 to 17 deaths per 1000 live births since 1990,35 the rates remain the highest in sub-Saharan Africa and South Asia at 26 and 22 deaths per 1000 live births, respectively, in 2023.35 Further reductions in newborn mortality require improvements in care at birth and for sick newborns, with persistent challenges in delivering such care in underserved parts of the world.33
In this study, multiple birth and NE were significant contributors to newborn deaths, identifying potential targeted areas for prevention. While improving infant survival in LMICs may increase observed CP prevalence among surviving infants, evidence from HICs where preventive measures have reduced rates by up to 40%36 suggests that improved perinatal care may mitigate any such increases.
Multiple birth, NE grade and prematurity were identified as significant predictors of mortality. Increased odds of death for multiple births are well recognised in LMICs, and strengthening antenatal and perinatal care for multiple births has been shown to reduce mortality.37 Effective interventions for NE in LMICs remain limited, particularly following the HELIX trial which showed increased mortality associated with therapeutic hypothermia for neonates in such context.38 Improved definitions, prevention strategies and context-specific interventions for NE in LMICs are needed. Thin transparent skin was chosen as a proxy indicator for extreme prematurity; however, given its subjectivity and poor specificity, this finding should be interpreted cautiously.
Diagnostic groups
Among the 125 infants assessed in-person at 24 months, 29% were diagnosed with CP, 10% with DD and 61% had no known disability or developmental delay. Despite a relatively high proportion of infants with no known disability, these findings support the appropriateness of the inclusion criteria used for the identification of infants at risk of developing CP. As such, these criteria may be useful for early identification and monitoring of high-risk infants in LMICs. While the definition of DD as <2 SD in at least one developmental domain used in this study is standard in HICs,39,41 the absence of locally derived normative data and definitions in LMIC cohorts may have influenced the proportion of infants with DD.
Neurological trajectories across diagnostic groups
The HINE effectively differentiated between the three diagnostic groups at 12 and 24 months. Most infants with CP scored ≤66, consistent with findings from an NICU cohort in Italy.20 However, in this cohort most infants with no known disability scored ≥73, while HINE scores in the Italian cohort were predominantly >67. Similar to the Italian cohort,20 infants with DD in our cohort exhibited transitional profiles, scoring ≤66 at 3 months but ≥73 by 12 and 24 months. These observed differences suggest that thresholds developed in HICs may not fully capture normative neurological development in LMIC settings, underscoring the need for context-specific validation efforts in LMICs. Neuroimaging was not available in this cohort, which is reflective of many LMIC settings.14 These findings support the pragmatic use of the HINE as a scoreable neurological assessment tool in settings where neuroimaging is not routinely available.
Developmental trajectories across diagnostic groups
Motor development assessed using PDMS-2 TMQ and GMFM-66 demonstrated early and persistent differences between infants with CP, DD and no known disability with infants with CP showing significantly lower scores from 3 months of age. All three diagnostic groups demonstrated a decline in PDMS-2 standard scores between 12 and 24 months. This may reflect cultural and environmental influences on item performance, as some PDMS-2 items include tasks that may be less familiar to infants in rural Bangladesh. These findings highlight current concerns around cross-cultural validity of HIC-normed motor developmental assessments and the need for culturally responsive developmental assessment tools. While PDMS-2 remains the most valid and reliable for the assessment of infant motor development in CP, the reported scores should be interpreted with caution in this cohort.
DAYC-2 CD raw scores demonstrated clear differences between infants with CP and those with DD and no known disability from 6 months of age. Given that over one-third of the children with CP in Bangladesh have cognitive impairment,42 and that guidelines recommend early cognitive monitoring, these findings further highlight the importance of early identification to implement effective interventions.43 While effective cognitive interventions for infants with CP have been demonstrated in HIC43 44 and promising interventions exist in LMIC for children without CP,45 timely identification of infants at risk for cognitive impairment remains essential to enable access to appropriate interventions and support. Caution is warranted in interpreting DAYC-2 assessments completed remotely, as the tools were translated from English to Bangla, but not back-translated, potentially resulting in inconsistent caregiver understanding and reporting. Locally validated, phone-appropriate cognitive assessment tools are urgently needed, particularly where in-person follow-up may be challenging. Similar considerations apply to communication, problem solving and personal social development domains.
Cerebral palsy: likely ambulant versus non-ambulant
Among infants diagnosed with CP, 81% were likely non-ambulant, consistent with current reports from Bangladesh and other LMICs where 71–73% of children were classified as GMFCS levels III-V.31 42 Although the evidence for the effectiveness of assistive technologies such as mobility aids and standing frames for non-ambulant CP is apparent,46 access remains limited in LMICs with only 12.7% of children with CP reported to have access to assistive technologies.31 This enduring high rate of non-ambulant CP urgently calls for context-specific support that enables children with CP and their families to function and participate in everyday life. Developmental trajectories diverged sharply by ambulatory status, with likely non-ambulant infants with CP showing less motor development over time compared with those with likely ambulant CP. There was no consistent difference in cognitive trajectories between infants with likely ambulant CP and DD, suggesting that longitudinal follow-up of both groups is prudent to better understand cognitive development over time.
Strengths and limitations
This study represents the first prospective developmental cohort study in an NICU population in Bangladesh focusing on infants at risk of developing CP. Best practice assessment tools were used, and where possible, adapted for the local context. The integration of both in-person and remote assessments enabled continuity of data collection during COVID-19 pandemic-related disruptions, which represents a key strength of the study.
Several limitations should be considered. Remote assessment tools lacked back-translation, and their psychometric properties have not been established in LMIC settings. Although globally recognised, developmental tools such as PDMS-2 and DAYC-2 may not fully reflect culturally embedded developmental expectations. Key neonatal mortality risk factors including low birth weight or gestational age <32 weeks could not be comprehensively assessed due to missing data or reliance on recalled last menstrual period, which is known to have reduced accuracy in low-literacy settings.47 Future research would benefit from prospective newborn examinations, such as The New Ballard scoring or Dubowitz scoring to improve risk modelling around gestational age.47
Lastly, the proportion of infants who died and or were diagnosed with and showed signs of NE was higher than expected, even in a high-risk population. This may reflect delayed presentation to health facilities following complications at birth, often after initial care by traditional healers.48 Additional barriers, including out-of-pocket health expenditures, may further limit access to timely care.49 While access to care for pregnant women has seen substantial improvement,49 equitable access to NICU care for high-risk infants remains an ongoing challenge. Overestimation of NE may have occurred due to multiple comorbidities within this cohort.
Conclusions
The high mortality observed in this cohort confirms the presence of survival bias. Among the surviving infants, one-third were diagnosed with CP with the majority likely to be non-ambulant. Motor delays in CP were identifiable from 3 months, and cognitive delays from 6 months. These findings highlight the urgent need for culturally responsive, CP-specific evidence-based early assessment and intervention programmes in LMIC settings to improve survival as well as neurological and developmental outcomes for infants at risk of developing CP.
Supplementary material
Acknowledgements
Data collection was supported by Dr Md Jahangir Alam. We acknowledge all infants and their families for their contribution to this research.
The funder did not influence the study design, data collection and analysis, decision to publish or preparation of the manuscript despite authors affiliation with the funder.
Footnotes
Funding: This work was supported by the Cerebral Palsy Alliance Research Foundation (PG01318).
Provenance and peer review: Part of a Topic Collection; Not commissioned; externally peer reviewed.
Patient consent for publication: Not applicable.
Ethics approval: Ethical approval was obtained from Asian Institute of Disability and Development (southasia-hrec-2019-5-02) and Bangladesh Medical Research Council (BMRC/NREC/2016-2019/72). Participants gave informed consent to participate in the study before taking part.
Patient and public involvement: Patients and/or the public were not involved in the design, or conduct, or reporting, or dissemination plans of this research.
Data availability statement
Data are available upon reasonable request.
References
- 1.Council UNGAEaS . Progress towards the Sustainable Development Goals: Report of the Secretary-General. New York: United Nations; 2024. [Google Scholar]
- 2.Rosenbaum P, Paneth N, Leviton A. The Definition and Classification of Cerebral Palsy. Dev Med Child Neurol. 2007;49:1–44. doi: 10.1111/j.1469-8749.2007.00001.x. [DOI] [PubMed] [Google Scholar]
- 3.Global Burden of Disease; 2019. Global health data exchange. [Google Scholar]
- 4.Khandaker G, Muhit M, Karim T, et al. Epidemiology of cerebral palsy in Bangladesh: a population-based surveillance study. Dev Med Child Neurol. 2019;61:601–9. doi: 10.1111/dmcn.14013. [DOI] [PubMed] [Google Scholar]
- 5.McIntyre S, Goldsmith S, Webb A, et al. Global prevalence of cerebral palsy: A systematic analysis. Develop Med Child Neuro. 2022;64:1494–506. doi: 10.1111/dmcn.15346. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Smythe T, Kuper H. The association between disability and all-cause mortality in low-income and middle-income countries: a systematic review and meta-analysis. Lancet Glob Health. 2024;12:e756–70. doi: 10.1016/S2214-109X(24)00042-1. [DOI] [PubMed] [Google Scholar]
- 7.Andrews C, Namaganda L, Eliasson A-C, et al. Functional development in children with cerebral palsy in Uganda: population-based longitudinal cohort study. Dev Med Child Neurol. 2022;64:70–9. doi: 10.1111/dmcn.14996. [DOI] [PubMed] [Google Scholar]
- 8.Lidbeck C, Häbel H, Martinsson C, et al. Motor Development in Children with Cerebral Palsy in Sweden-A Population-Based Longitudinal Register Study. Children (Basel) 2023;10:1864. doi: 10.3390/children10121864. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Klevberg GL, Elvrum A-KG, Zucknick M, et al. Development of bimanual performance in young children with cerebral palsy. Dev Med Child Neurol. 2018;60:490–7. doi: 10.1111/dmcn.13680. [DOI] [PubMed] [Google Scholar]
- 10.Eliasson AC, Nordstrand L, Backheden M, et al. Longitudinal development of hand use in children with unilateral spastic cerebral palsy from 18 months to 18 years. Dev Med Child Neurol. 2023;65:e101–e01. doi: 10.1111/dmcn.15735. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.te Velde A, Tantsis E, Novak I, et al. Age of Diagnosis, Fidelity and Acceptability of an Early Diagnosis Clinic for Cerebral Palsy: A Single Site Implementation Study. Brain Sci. 2021;11:1074. doi: 10.3390/brainsci11081074. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Al Imam MH, Jahan I, Das MC, et al. Population-based surveillance of children with cerebral palsy enables early diagnosis and intervention. Dev Med Child Neurol. 2021;63:883–4. doi: 10.1111/dmcn.14861. [DOI] [PubMed] [Google Scholar]
- 13.te Velde A, Morgan C, Novak I, et al. Early Diagnosis and Classification of Cerebral Palsy: An Historical Perspective and Barriers to an Early Diagnosis. JCM. 2019;8:1599. doi: 10.3390/jcm8101599. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Novak I, Morgan C, Adde L, et al. Early, Accurate Diagnosis and Early Intervention in Cerebral Palsy: Advances in Diagnosis and Treatment. JAMA Pediatr. 2017;171:897–907. doi: 10.1001/jamapediatrics.2017.1689. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Karim T, Te Velde A, Webb A, et al. Early detection of cerebral palsy among a high-risk cohort in Bangladesh. BMJ Paediatr Open . 2025;9:e003921. doi: 10.1136/bmjpo-2025-003921. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Sarnat HB, Sarnat MS. Neonatal encephalopathy following fetal distress. Obstet Gynecol Surv. 1977;32:295. doi: 10.1097/00006254-197705000-00011. [DOI] [Google Scholar]
- 17.Folio M, Fewell R. Peabody Developmental Motor Scales. 2nd. Austin, Texas: Pro-ed; 2000. edn. [Google Scholar]
- 18.Voress JK, Maddox T.Developmental Assessment of Young Children - Second Edition Examiner’s Manual Austin .Texas: Pro-Ed; 2013 [Google Scholar]
- 19.Squires J, Bricker D. Ages & Stages Questionniares, Third Edition (ASQ-3TM) 3rd. Paul H. Brookes Publishing Co; 2009. edn. [Google Scholar]
- 20.Romeo DMM, Cioni M, Palermo F, et al. Neurological assessment in infants discharged from a neonatal intensive care unit. Eur J Paediatr Neurol. 2013;17:192–8. doi: 10.1016/j.ejpn.2012.09.006. [DOI] [PubMed] [Google Scholar]
- 21.Tripathi R, Joshua AM, Kotian MS, et al. Normal Motor Development of Indian Children on Peabody Developmental Motor Scales-2 (PDMS-2) Pediatr Phys Ther. 2008;20:167–72. doi: 10.1097/PEP.0b013e3181710340. [DOI] [PubMed] [Google Scholar]
- 22.Mendonça B, Sargent B, Fetters L. Cross-cultural validity of standardized motor development screening and assessment tools: a systematic review. Dev Med Child Neurol. 2016;58:1213–22. doi: 10.1111/dmcn.13263. [DOI] [PubMed] [Google Scholar]
- 23.Connolly BH, McClune NO, Gatlin R. Concurrent validity of the Bayley-III and the Peabody Developmental Motor Scale-2. Pediatr Phys Ther. 2012;24:345–52. doi: 10.1097/PEP.0b013e318267c5cf. [DOI] [PubMed] [Google Scholar]
- 24.Wang H-H, Liao H-F, Hsieh C-L. Reliability, sensitivity to change, and responsiveness of the peabody developmental motor scales-second edition for children with cerebral palsy. Phys Ther. 2006;86:1351–9. doi: 10.2522/ptj.20050259. [DOI] [PubMed] [Google Scholar]
- 25.Russell DJ, Rosenbaum PL, Wright M, et al. Gross Motor Function Measure (GMFM-66 & GMFM-88) User’s Manual. 2nd. Mac Keith Press; 2013. edn. [Google Scholar]
- 26.Rivera-Rujana DM, Muñoz-Rodríguez DI, Agudelo-Cifuentes MC. Reliability of the Gross Motor Function Measure-66 scale in the evaluation of children with cerebral palsy: validation for Colombia. Bol Med Hosp Infant Mex . 2022;79:33–43. doi: 10.24875/BMHIM.21000094. [DOI] [PubMed] [Google Scholar]
- 27.Munoz-Chereau B, Ang L, Dockrell J, et al. Measuring early child development across low and middle-income countries: A systematic review. J Early Child Res. 2021;19:443–70. doi: 10.1177/1476718X211020031. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Sabanathan S, Wills B, Gladstone M. Child development assessment tools in low-income and middle-income countries: how can we use them more appropriately? Arch Dis Child. 2015;100:482–8. doi: 10.1136/archdischild-2014-308114. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Palisano R, Rosenbaum P, Walter S, et al. Development and reliability of a system to classify gross motor function in children with cerebral palsy. Dev Med Child Neurol. 1997;39:214–23. doi: 10.1111/j.1469-8749.1997.tb07414.x. [DOI] [PubMed] [Google Scholar]
- 30.Eliasson A-C, Ullenhag A, Wahlström U, et al. Mini-MACS: development of the Manual Ability Classification System for children younger than 4 years of age with signs of cerebral palsy. Dev Med Child Neurol. 2017;59:72–8. doi: 10.1111/dmcn.13162. [DOI] [PubMed] [Google Scholar]
- 31.Jahan I, Muhit M, Hardianto D, et al. Epidemiology of cerebral palsy in low- and middle-income countries: preliminary findings from an international multi-centre cerebral palsy register. Dev Med Child Neurol. 2021;63:1327–36. doi: 10.1111/dmcn.14926. [DOI] [PubMed] [Google Scholar]
- 32.von Elm E, Altman DG, Egger M, et al. The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) Statement: Guidelines for Reporting Observational Studies. PLoS Med. 2007;4:e296. doi: 10.1371/journal.pmed.0040296. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.UN UNI-aGfCME, IGME . Levels and Trends in Child Mortality Report 2024: Estimates Developed by the United Nations Inter-Agency Group for Child Mortality Estimation. New York: United Nations Children’s Fund; 2025. [Google Scholar]
- 34.Kakooza-Mwesige A, Andrews C, Peterson S, et al. Prevalence of cerebral palsy in Uganda: a population-based study. Lancet Glob Health. 2017;5:e1275–82. doi: 10.1016/S2214-109X(17)30374-1. [DOI] [PubMed] [Google Scholar]
- 35.UNICEF Neonatal mortality: UNICEF. 2025. https://data.unicef.org/topic/child-survival/neonatal-mortality/ Available.
- 36.Australian Cerebral Palsy Register Group Australian cerebral palsy register bulletin birth years 1995-2018: australian cerebral palsy register, 2025. 2025
- 37.Vogel JP, Torloni MR, Seuc A, et al. Maternal and perinatal outcomes of twin pregnancy in 23 low- and middle-income countries. PLoS One. 2013;8:e70549. doi: 10.1371/journal.pone.0070549. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38.Molloy EJ, Branagan A, Hurley T, et al. Neonatal encephalopathy and hypoxic–ischemic encephalopathy: moving from controversy to consensus definitions and subclassification. Pediatr Res. 2023;94:1860–3. doi: 10.1038/s41390-023-02775-z. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39.Putnick DL, Bell EM, Tyris J, et al. Place-Based Child Opportunity at Birth and Child Development from Infancy to Age 4. J Pediatr. 2024;267:113909. doi: 10.1016/j.jpeds.2024.113909. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40.Alvarado-Domenech LI, Rivera-Amill V, Appleton AA, et al. Early Childhood Neurodevelopmental Outcomes in Children with Prenatal Zika Virus Exposure: A Cohort Study in Puerto Rico. J Pediatr. 2022;247:38–45. doi: 10.1016/j.jpeds.2022.05.016. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 41.Lee N-C, Chang K-L, In ’t Groen SLM, et al. Outcome of Later-Onset Pompe Disease Identified Through Newborn Screening. J Pediatr. 2022;244:139–47. doi: 10.1016/j.jpeds.2021.12.072. [DOI] [PubMed] [Google Scholar]
- 42.Narayan A, Muhit M, Whitehall J, et al. Associated Impairments among Children with Cerebral Palsy in Rural Bangladesh-Findings from the Bangladesh Cerebral Palsy Register. J Clin Med. 2023;12:1597. doi: 10.3390/jcm12041597. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 43.Morgan C, Fetters L, Adde L, et al. Early Intervention for Children Aged 0 to 2 Years With or at High Risk of Cerebral Palsy: International Clinical Practice Guideline Based on Systematic Reviews. JAMA Pediatr. 2021;175:846–58. doi: 10.1001/jamapediatrics.2021.0878. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 44.Novak I, Morgan C, Fahey M, et al. State of the Evidence Traffic Lights 2019: Systematic Review of Interventions for Preventing and Treating Children with Cerebral Palsy. Curr Neurol Neurosci Rep. 2020;20:3. doi: 10.1007/s11910-020-1022-z. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 45.Zhang L, Ssewanyana D, Martin M-C, et al. Supporting Child Development Through Parenting Interventions in Low- to Middle-Income Countries: An Updated Systematic Review. Front Public Health. 2021;9 doi: 10.3389/fpubh.2021.671988. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 46.Kija E, Namata TT, Sebunya R. Management of cerebral palsy in low- and middle-income countries. Semin Pediatr Neurol. 2025;55:101202. doi: 10.1016/j.spen.2025.101202. [DOI] [PubMed] [Google Scholar]
- 47.Ravi A, Singh A. Born Too Soon: The Global Action Report on Preterm Birth. Floreat: Australasian Medical Journal. 2012;598 [Google Scholar]
- 48.Abdullah ASM, Dalal K, Yasmin M, et al. Perceptions and practices on newborn care and managing complications at rural communities in Bangladesh: a qualitative study. BMC Pediatr. 2021;21:168. doi: 10.1186/s12887-021-02633-z. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 49.Nandi A, Charters TJ, Quamruzzaman A, et al. Health care services use, stillbirth, and neonatal and infant survival following implementation of the Maternal Health Voucher Scheme in Bangladesh: A difference-in-differences analysis of Bangladesh Demographic and Health Survey data, 2000 to 2016. PLoS Med. 2022;19:e1004022. doi: 10.1371/journal.pmed.1004022. [DOI] [PMC free article] [PubMed] [Google Scholar]