Rehabilitation status of children with cerebral palsy in Bangladesh: Findings from the Bangladesh Cerebral Palsy Register

Objective The objective of this study was to assess the rehabilitation status and factors associated with rehabilitation service utilisation among children with cerebral palsy (CP) in Bangladesh. Materials and methods This is a population-based surveillance study conducted among children with CP registered in the Bangladesh CP Register (BCPR), the first population-based register of children with CP aged <18 years (y) in Bangladesh. Children with CP were identified from the community using the key informant method and underwent a detailed neurodevelopmental assessment. Socio-demographic, clinical and rehabilitation status were documented. Unadjusted and adjusted analyses with a 95% confidence interval (CI) were used to identify potential predictors of rehabilitation service uptake. Results Between January 2015 and December 2019, 2852 children with CP were registered in the BCPR (mean (standard deviation, SD) age: 7 y 8 months (mo) (4 y 7 mo), 38.5% female). Of these, 50.2% had received rehabilitation services; physiotherapy was the most common type of service (90.0%). The mean (SD) age at commencement of rehabilitation services was 3 y 10 mo (3 y 1 mo). The odds of not receiving rehabilitation was significantly higher among female children (adjusted odds ratio (aOR) 1.3 [95% CI: 1.0–1.7], children whose mothers were illiterate and primary level completed (aOR 2.1 [95% CI: 1.4–3.1] and aOR 1.5 [95% CI: 1.1–2.1], respectively), fathers were illiterate (aOR 1.9 [95% CI: 1.3–2.8]), had a monthly family income ~US$ 59–118 (aOR: 1.8 [95% CI: 1.2–2.6]), had hearing impairment (aOR: 2.3 [95% CI: 1.5–3.5]) and motor severity (i.e. Gross Motor Function Classification System level III (aOR: 0.6 [95% CI: 0.3–0.9]) and level V (aOR: 0.4 [95% CI: 0.2–0.7])). Conclusions Rehabilitation status was poor among the majority of the children with CP in the BCPR cohort, limiting their opportunities for functional improvement. A community-based rehabilitation model focusing on socio-demographic and clinical characteristics should be a public health priority in Bangladesh.


Introduction
Childhood disability is a global public health concern due to its lifelong impact on physical and psychological wellbeing. An estimated 80% of childhood disability occurs in low-and middle-income countries (LMICs) [1]. Despite this high burden, there is limited information on access to rehabilitation services in LMICs [2]. The 2030 Agenda for Sustainable Development Goals warrants that children with disabilities should enjoy equal access to health care and rehabilitation (Goal #3 Good health and wellbeing) regardless of their abilities and socioeconomic status (Goal #10 Reduce inequality).
Cerebral palsy (CP) is one of the leading causes of childhood disability, with an estimated 50 million people living with CP worldwide [3,4]. CP is a clinical description for non-progressive motor disorders caused by injury to the developing brain [5]. The burden of CP is estimated to be substantially higher in LMICs compared to high-income countries (HICs) [6,7].
Children with CP require support from a multidisciplinary team of medical and rehabilitation professionals including physiotherapists, occupational therapists and speech and language therapists to improve function, prevent secondary complications and enhance autonomy [8]. However, such services are not always available for children with CP, particularly in LMICs [9,10]. In Bangladesh, the population-based prevalence of CP was estimated to be 3.4 (95% CI 3.2-3.7) per 1000 children, which approximates to~234,000 children with CP in a country of 166 million people [7]. However, this is likely to be an underestimation due to survival bias.
Population-based data in this regard is limited in LMICs. Such data are essential to identify potential scope for intervention, ensure optimal use of inadequate available resources and maximise the service coverage for children with disabilities (e.g. CP) in the resource-constrained settings of LMICs like Bangladesh. Therefore, this study aimed to assess the

Materials and methods
We established the first population-based surveillance of children with CP i.e. Bangladesh CP Register (BCPR) in rural Bangladesh in 2015. The BCPR is an ongoing surveillance that studies epidemiology, rehabilitation and intervention strategies to improve functional outcomes and limit associated impairments among children with CP in Bangladesh [29]. Initially, the surveillance activities (i.e. BCPR) were confined in one subdistrict (Shahjadpur,~325 square kilometres,~123,576 households) which represents rural and semi-urban Bangladesh (where the majority of the population (76.7%) lives [30]) in terms of demographic and other indicators (e.g. birth rate, immunisation rate, perinatal mortality rate, literacy rate). To date, this population-based surveillance has been maintained with high case ascertainment in Shahjadpur. Additionally, between 2015 and 2019 the BCPR has been scaled up to 17 other subdistricts (S4,338 square kilometres,~1,304,960 households) following opportunistic recruitment at the community level (Fig 1).

Study participants and data collection method
As part of the ongoing surveillance (i.e. BCPR), children with suspected CP in a community were identified using the key informant method (KIM). The KIM is a validated method where local volunteers (e.g. religious leaders, teachers, community health workers etc.) are trained as key informants (KIs) to identify children with disabilities in their communities [31]. The KIs were identified (approximately 1 KI per village) by the Community Mobilisers (CMs-paid project staff) and received a day-long training on the identification of children with suspected CP, disability sensitisation, advocacy for a disability inclusive society using flip charts, group work and role play. Following the training, the KIs were provided 4-6 weeks to identify and enlist children with suspected CP and share their contact details with CMs to bring those children and their primary caregivers to the nearest medical assessment camps for a confirmed diagnosis, detailed neurodevelopmental assessments and registration in the BCPR. The clinical definition of CP adopted from the Surveillance of CP in Europe (SCPE) [32] and the Australian CP Register (ACPR) [6] were strictly followed during case ascertainment. The details of case identification and recruitment into the BCPR have been described in our previous publication [7]. Data were collected using a standard case-record form (adapted from the ACPR) through interviews with the primary caregivers, as well as a clinical assessment and a review of medical records (if available). The detailed data collection method and the variables included in the BCPR are available in our previous publication [7]. In the current analyses, we used the following variables: (i) socio-demographic characteristics (e.g. age, gender, educational level of parents, monthly family income); (ii) clinical characteristics (e.g. Gross Motor Function Classification System (GMFCS) level, Manual Ability Classification System (MACS) level, predominant CP motor type, CP topography, associated impairments) and (iii) information on rehabilitation (i.e. whether the child ever received rehabilitation, the type and location/source of services received, age of commencement of rehabilitation). Here, the main outcome variable was the rehabilitation status (i.e. whether he/she ever received rehabilitation) of a child with CP registered in the BCPR, which is a binary variable with 'yes' and 'no' as responses. Information on rehabilitation was collected from primary caregivers of children with CP by a physiotherapist at the medical assessment camps in the context of BCPR recruitment. To document the type of rehabilitation services received, multiple responses were allowed and, therefore, the numbers presented for this variable are not mutually exclusive. Additionally, the medical records available from primary caregivers were reviewed for any documentation of rehabilitation services.

Data management and analysis
Following collection, the data were entered electronically into the online password-protected BCPR data repository (http://bangladesh.cpregister.com/), access to which was limited to named investigators only. A dedicated data management team located at CSF Global in Dhaka, Bangladesh, with support from investigators conducted the data management and analyses.
To maintain the quality of the data, we used a double data entry method. Subsequently, data-entry error checks were performed by running frequencies of all variables to identify any outliers. In case of any missing data or incorrect/suspicious information, the BCPR case record forms (i.e. source documents) were reviewed. If the information was not found in the BCPR form, it was relayed to the field team and where possible the participants were contacted to gather the missing information. Continuous variables were collected as exact values and later recoded and categorised into groups (e.g. age was categorised as follows: 0-4 years, 5-9 years, 10-14 years and 15-18 years). Similarly, the monthly family income data were converted to United States Dollar (US$) (considering 1 US$ = 84.43 Bangladeshi taka (BDT)) and categorised into four family-income groups (i.e. BDT 500-4999 (US$ S6-59), BDT 5000-9999 (US$ S59-118), BDT 10,000-14,999 (US$ S118-178) and BDT 15,000 and above (US$ S178 and above). Comparison between the BCPR cohort and the general population was performed using the 2014 Bangladesh Demographic and Health Survey (BDHS) [33] and Household Income and Expenditure Survey [34] data from 2016. The poverty level among the families in the BCPR cohort was estimated using the national poverty lines (at the divisional level) as a cut-off. The proportion of families living below the poverty lines was then compared with the general population of the respective divisions as reported in the 2016 Household Income Expenditure Survey (HIES) in Bangladesh [34]. Bivariate analyses were completed to assess the impact of socio-demographic and clinical factors on rehabilitation status. For regression models, 'Not receiving rehabilitation' was considered as the main outcome of interest. Factors that were found to be statistically significant in unadjusted logistic regression were fitted into an adjusted logistic regression model. Odds ratios with 95% confidence intervals (CI) were reported. A p value <0.05 was considered significant. All data were analysed using the Statistical Package for the Social Sciences (SPSS) software, version 26 (IBM, Armonk, NY, USA).

Ethical considerations
Ethical approval for the BCPR study was obtained from the Cerebral Palsy Alliance

Factors influencing receipt of rehabilitation services among children with CP
Age and sex of children with CP. Children in the BCPR cohort aged 10-14 y and 15-18 y had 1.3 times (95% confidence interval (CI) 1.1-1.7) and 1.6 times (95% CI 1.2-2.2) lower likelihood of receiving rehabilitation services compared with children aged 0-4 y, respectively. Female children were 1.2 times (95% CI 1.0-1.4) less likely to receive rehabilitation services compared to male children ( Table 3).
Education of parents. Parental education was significantly related to rehabilitation service uptake among children with CP in our cohort (p<0.001). Of the mothers and fathers of children with CP who had received rehabilitation services, 77.8% (n = 1105/1421) and 69.5% (n = 979/1409) had completed primary orsecondary education. Children of illiterate mothers had a 3.4 times (95% CI 2.8-4.2) lower chance of receiving rehabilitation services. A similar observation was reported for the father's education (Table 3).

Discussion
To the best of our knowledge, this is the first population-based study reporting the rehabilitation status and predictors of rehabilitation service uptake among children with CP in an LMIC. We observed that a large number of children with CP in the BCPR had never received any rehabilitation services. Children who had received services were more likely to be female and have educated parents, a higher socio-economic status and severe gross motor impairment.
Similar to this study, poor rehabilitation coverage has been reported among children with CP in India [22]. In contrast, Schmidt et al. [35] reported that 98.9% of children with CP in seven HICs had received rehabilitation services within one year. The proportion of children receiving rehabilitation services varies widely, even between LMICs. A higher proportion of rehabilitation service uptake has been observed in hospital-/institution-based studies, ranging from 55.6% in India [22] to 90.4% in Jordan [13], whereas studies conducted in communitybased settings identified considerably poorer rehabilitation service uptake in Uganda (9.7%) [10] and South Africa (26.0%) [25]. The observed differences are most likely due to selection bias in hospital-/institution-and community-based settings and the socioeconomic conditions of study participants (i.e. more affluent people with a higher level of education are more likely to access services).
The age for commencing rehabilitation services among participants was substantially delayed when compared with HICs (3 y 8 mo in Bangladesh vs 1 y 5 mo in Australia) [36]. The reported delays in the diagnosis of CP in rural Bangladesh might play a key role here [7]. Recent evidence suggests that early initiation of rehabilitation services is crucial for the best possible motor outcomes [36].
Among children who had received rehabilitation services in the BCPR cohort, the majority received physiotherapy. Similar to our study, physiotherapy was frequently reported in studies conducted in Jordan (90.4%) [13] and Korea (81.3%) [12]. We found that more than two third of the children registered in the BCPR had MACS level III-V (68.6%) and could have benefitted from occupational therapy. Furthermore, 75.2% of children with speech impairment could have benefitted from speech therapy. However, our findings indicate that none of these children had received the required services. The low number of trained occupational therapists and speech and language therapists compared to physiotherapists (250 vs 260 vs 2400, respectively) with a higher availability in major cities in Bangladesh [37] might be responsible for this disparity. We also found that only 4.7% of children with a GMFCS level III-V had received assistive devices. The poor access to assistive devices found in this study is consistent with earlier studies conducted in LMICs [7,10,24]. Without mobility aids, children with severe functional motor limitations are likely to be bedridden and unable to participate in family, school and community life [10]. We also identified several socio-demographic and clinical factors as barriers in rehabilitation service uptake among our participating children. In terms of gender, female children with CP had a lower likelihood of receiving rehabilitation services when compared to male children. However, the current literature on this issue is conflicting. Young females with CP were found four times more likely to utilise rehabilitation services compared to males in the USA [38]. Whereas Sinha and Sharma [22] reported that there is no relationship between sex and rehabilitation service utilisation of children with CP in India. Women with disabilities face a double burden, because of their gender roles and disabilities, in LMICs [39]. Disability for a female becomes a greater barrier in terms of accessing opportunities such as rehabilitation [39,40].
McConachie et al. [20] described that having a male child can influence parents to seek rehabilitation services, particularly in rural settings. This might be related to the notion that male children need to be able to support a family in the future.
Parental education, and in particular maternal education, was significantly associated with rehabilitation service uptake among children with CP in the BCPR. Children whose parents were literate had significantly higher odds of receiving rehabilitation services. The findings are consistent with studies conducted in India [22] and the USA [38]. It is likely that parents who are educated are more aware of their child's health condition and needs, and understand the significance of providing rehabilitation to their children with CP. Additionally, parents with less education are more likely to be engaged in daily-basis low-paid jobs, which can make it difficult for them to take time away from work or to cover travel and/or service costs to get their children to rehabilitation centres. Our findings suggest that the majority of children in the BCPR were from impoverished families, and this cohort had higher odds of not receiving rehabilitation services. Financial constraints have been identified as a major barrier to utilising rehabilitation services in studies conducted both in HICs [38] and LMICs [10,22,24]. McConachie et al. [20] describe that a majority of children with CP cannot access rehabilitation services because of the costs, both direct (e.g. rehabilitation service charge) and indirect (e.g. transport cost, accommodation cost, food cost), associated with these services.
In terms of clinical factors, we found that children with GMFCS level III and V had a significantly higher likelihood of receiving rehabilitation services. This result is consistent with earlier studies conducted in Canada [17,26], the USA [26] and Australia [19]. The higher utilisation of rehabilitation services among severely motor-impaired children may be because of their increased rehabilitation needs in order to improve pain, comfort and quality of life. While the importance of rehabilitation for children with GMFCS level III-V cannot be undervalued, recent evidence suggests that the activity, function and participation of children with GMFCS level I-II could be enhanced through early intervention and rehabilitation services [41].
We also found that children with hearing impairment had a significantly lower probability of receiving rehabilitation services. Similar to our findings, Liljenquist et al. [38] found that children with CP and associated impairments had lower odds of utilising physiotherapy services in the USA. In contrast, Majnemer et al. [17] reported that children with lower intellectual impairment had higher odds of rehabilitation service uptake in Canada. It is not clear why children with hearing impairment have a lower likelihood of receiving rehabilitation in our cohort. Further study is required to investigate the effect of associated impairments on rehabilitation service uptake among children with CP in LMICs like Bangladesh.

Strengths and limitations
This study used population-based data from an established register of children with CP in Bangladesh. Another methodological strength of this research is the adoption of the case definition of CP from the ACPR and the SCPE to ensure international consensus for the clinical diagnosis [6] and measurement of motor functions [42]. Despite our extensive efforts, this study had several limitations, however. Whilst the KIM is cost-effective in the identification of children with disabilities in LMICs [24,43], the BCPR recruitment efforts might have missed some children with CP who have mild motor limitations as the KIM has a 77.6% case-ascertainment rate compared with door-to-door surveying [31]. Therefore, children with severe motor limitations may have been overrepresented in this study. Secondly, the assessment of rehabilitation status was mostly based on the primary caregiver's responses, due to a lack of service utilisation records. Although there is a chance of recall bias, such a method has previously been used in HICs [17,19] and LMICs [10,24]. Thirdly, due to a lack of medical records, BCPR data collection has to rely on caregiver responses, in addition to the clinical examination, when assessing the severity of associated impairments in some of cases. Although this might have introduced information bias in determining the severity of associated impairments, such methods have also been previously used in other large-scale studies conducted in LMICs and HICs [7,44]. Finally, the national poverty lines were estimated based on household per capita consumption (food and non-food consumption/expenditure) using a detailed questionnaire from the HIES in Bangladesh [34]. Although information with that level of detail is not collected as part of the BCPR, the methodology is similar to some extent. However, there is still a risk of overreporting the poverty level with BCPR data due to the differences in survey tools and depth of information collected compared to the HIES, as well as several other factors.

Conclusions
Nearly half of the children with CP in our study had not have access to rehabilitation services. A majority of children who were in need of assistive devices could not access them. Additionally, the age at commencement of rehabilitation was substantially delayed, limiting the opportunity to improve function and independence. Socio-demographic (i.e. sex, parental education and monthly family income) and clinical factors (i.e. GMFCS level and associated impairments) were significantly associated with rehabilitation service uptake. This evidence has important implications for policy formation and the improvement of rehabilitation services for children with CP in Bangladesh. Locally available and affordable early intervention and rehabilitation service delivery models, including training of rehabilitation professionals regarding community-based management of CP, should be seen as a priority for the strategic development of rehabilitation service coverage among this vulnerable population.