This cluster randomized clinical trial investigates if the community-based Sustainable Model of Early Intervention and Telerehabilitation for Children With Cerebral Palsy (SMART-CP) model improves early diagnosis, access to services, and survival in children with cerebral palsy.
Key Points
Question
Does the community-based Sustainable Model of Early Intervention and Telerehabilitation for Children With Cerebral Palsy (SMART-CP) model improve early diagnosis, access to services, and survival of children with CP?
Findings
In this cluster randomized clinical trial including 968 children with CP, compared with the control arm, the intervention arm exhibited significantly higher service access, earlier CP diagnosis and rehabilitation commencement, and had higher therapy sessions and assistive device use. The SMART-CP model did not affect survival outcomes.
Meaning
Results suggest that in low-resource settings, the SMART-CP model enhanced access to and utilization of early diagnosis, intervention, and rehabilitation services for children with CP.
Abstract
Importance
Access to early intervention and rehabilitation services among children with cerebral palsy (CP) remains limited in Bangladesh, which demands an innovative and sustainable service delivery model.
Objective
To evaluate the effectiveness of the Sustainable Model of Early Intervention and Telerehabilitation for Children With CP (SMART-CP) model compared with usual care in improving access to and utilization of early diagnosis, early intervention, and rehabilitation services in rural Bangladesh.
Design, Setting, and Participants
This was a 2-arm cluster randomized clinical trial, with 8 clusters (ie, subdistricts) randomly allocated to the intervention (SMART-CP model) or control arm. The setting was in Sirajganj, Bangladesh, and included children with CP 18 years or younger. Outcomes were measured at 0 and 12 months, and an intention-to-treat analysis was conducted. Data were analyzed from December 2023 to May 2024.
Interventions
The SMART-CP model comprised (1) a rural referral network involving key informants and caregiver peer groups (called mPower or mothers’ power), (2) subdistrict level SMART-CP centers, and (3) telerehabilitation services. Children in the intervention arm received weekly goal-directed therapy, mPower group meetings every 2 weeks, and monthly telerehabilitation sessions.
Main Outcomes and Measures
The primary outcome was whether a child with CP accessed any form of rehabilitation services, with secondary outcomes analyzed as hypothesis generating.
Results
Overall, 968 children with CP (mean [SD] age, 7.9 [4.9] years; 581 male [60.0%]) were enrolled, with 500 in the intervention arm and 468 in the control arm. Between baseline and endline, rehabilitation services uptake significantly increased in the intervention arm (70.2% [351 of 500] vs 99.4% [497 of 500]), compared with the control arm (63.9% [299 of 468] vs 68.2% [319 of 468]; P <.001). Children in the intervention arm were 1.5 times more likely to access rehabilitation than the control arm. Secondary analyses suggested that the intervention arm also facilitated early CP diagnosis (mean [SD] diagnosis time, 2.0 [2.0] years vs 3.8 [3.3] years; Cohen d = −0.7) and initiation of rehabilitation (mean [SD] rehabilitation time, 1.8 [1.8] years vs 3.6 [2.4] years; Cohen d = −0.9). Additionally, higher therapy session counts (mean [SD] session counts, 23.4 [31.7] vs 4.3 [20.8]; Cohen d = 0.7), increased assistive device utilization (20.8% [104 of 500] vs 3.0% [14 of 468]; risk ratio, 0.82; 95% CI, 0.78-0.86; P < .001), and lower out-of-pocket expenditure per month (mean [SD] expenditure, $1.5 [$1.6] vs $2.9 [$5.1]; Cohen d = −0.4) were found in the intervention arm. No significant difference in clinical outcomes and mortality rates was observed between the intervention and control groups.
Conclusions and Relevance
Results of this cluster randomized clinical trial reveal that the SMART-CP model improved access to and utilization of early diagnosis and intervention services for children with CP in rural Bangladesh. This model holds promise for global scalability.
Trial Registration
ANZCTR Trial Identifier: ACTRN12622000396729
Introduction
Children with cerebral palsy (CP) encounter complex challenges requiring multidisciplinary services, often lacking in low- and middle-income countries (LMICs).1 In Bangladesh, persistent shortages and uneven distribution of rehabilitation professionals1,2 severely limit access to services for children with CP.3,4 A situation analysis revealed only 6.8 centers per million people in Bangladesh, with 95.8% located in urban areas and 66.3% requiring out-of-pocket payments.2 Only 1.2% of these centers provide the multidisciplinary services essential for children with CP.2 Furthermore, the rehabilitation workforce is insufficient, with 1 physiotherapist for 106 000 people.2
Data from the Bangladesh CP Register (BCPR) highlight socioeconomic challenges, with 97% of families living in extreme poverty and 49.8% of children with CP lacking access to rehabilitation services.3,4 Furthermore, delayed diagnosis of CP is highly prevalent in rural areas.3 With 77% of Bangladesh’s population living in rural areas,5 financial barriers and limited government resources (0.3% of the annual budget allocated to disability)6 hinder service provision.2,3,4,7,8 These gaps underscore the need for an innovative and sustainable model to address the unmet needs.
Telecommunication technologies and rehabilitation service via referral networks could potentially improve access to rehabilitation services for children with disabilities in both developed and less developed countries, particularly in remote rural areas of LMICs.9,10,11,12 Leveraging these approaches, we designed the Sustainable Model of Early intervention and Telerehabilitation for children with CP (SMART-CP) model. This trial aimed to evaluate the effectiveness of the SMART-CP model compared with usual care in improving access to and utilization of early detection, intervention, and rehabilitation services for children with CP in rural Bangladesh. We hypothesized that the SMART-CP model would improve access to services, service utilization, early detection, assistive devices use, and survival of children with CP in rural Bangladesh.
Methods
Consent and Ethical Considerations
The trial is registered with the Australian New Zealand Clinical Trials Registry. Ethical approval was obtained from the Bangladesh Medical Research Council. Written informed consent was obtained from the primary caregivers, who were provided with information in the local language to ensure understanding of the study’s purpose, procedures, risks, and their right to withdraw. To ensure confidentiality, only anonymized data were used in the analyses.
Study Design and Settings
This 2-arm cluster randomized clinical trial (RCT) compared the SMART-CP model with usual care across 8 randomly selected subdistricts in Sirajganj, Bangladesh, between April 2022 and March 2023. Sirajganj spans 2498 km2, with a child population of 1 249 2035 and an estimated 4997 children with CP. The trial protocol is available in Supplement 1. This study followed the Consolidated Standards of Reporting Trials (CONSORT) reporting guidelines.
Participants and Recruitment
The trial used the BCPR, an ongoing surveillance program for children with CP 18 years or younger in Bangladesh, as a sampling frame using the Key Informant Method (KIM).13 Recruitment for both intervention and control arms was conducted through trained key informants. Suspected cases identified through KIM underwent comprehensive neurodevelopmental assessments by a multidisciplinary team to confirm CP diagnosis and registration into the BCPR and subsequently enrolled in the trial across both arms. Eligibility criteria included confirmed CP diagnosis, registration in the BCPR, age 18 years or younger, had any severity of CP, and maintained a residence in the Sirajganj district. Children participating in any other trials or with life-limiting illnesses were excluded. Bangladesh is largely a homogenous country, with ethnic minorities comprising only 1% of the population.5 National surveys, such as the Bangladesh Demographic and Health Survey, do not include race and ethnicity data. Following this standard practice, we did not collect this information in our trial.
Cluster Formation and Randomization
Randomization unit was subdistricts, and 8 subdistricts were grouped into homogeneous pairs based on socioeconomic characteristics and CP prevalence. From each pair, 1 intervention subdistrict was selected randomly using random number generator, the remainder served as a control. The randomization schedule was computer generated.
Sample Size Calculation and Statistical Power
The sample size was calculated using the method for cluster RCT comparing proportions between 2 groups in a parallel-arm design.14 The calculation assumed a baseline rehabilitation service access rate of 50% in the control arm and a 20% improvement in the intervention arm, with an intraclass correlation coefficient of 0.083 and an average cluster size of 120 children.8 With these parameters, a total of 960 children with CP from 8 clusters (4 per arm) provided 80% power to detect the specified difference at a 5% significance level, accounting for clustering effects and 20% attrition. Participants and investigators were not blinded to the intervention received; however, outcome assessors remained blinded to group allocation.
Intervention Arm
The development of the SMART-CP model was a collaborative, codesign process, drawing on input from primary caregivers, service practitioners, and local community stakeholders. The SMART-CP model was implemented in 4 clusters within the intervention arm, incorporating the following activities (eFigure in Supplement 2):
-
Referral networks with key informants and caregiver peer groups (called mPower or mothers’ power)
Key informant: key informants know about children with disability due to their social role, and they act as a source of ongoing advocacy and referral.13 Considering 1 to 2 key informants per village, this study trained 1150 key informants. Although key informants from both arms were trained similarly to ensure consistent identification criteria, only the intervention arm received a structured referral network that included monthly follow-up calls.
mPower groups: the mPower initiative was designed to provide peer support and improve health literacy for mothers of children with CP. Each group, consisting of 10 to 12 mothers, met every 2 weeks to discuss CP management, address challenges, and offer psychosocial support. The Getting to Know CP15 manual guided discussions, empowering parents with practical knowledge. A total of 45 mPower groups were formed, facilitated by trained group leaders with support from a community mobilizer. All mothers in the intervention arm were invited; however, participation was voluntary. Further implementation details are published elsewhere.16
-
SMART-CP center
A subdistrict level early detection, intervention and telerehabilitation service center for children with CP, the SMART-CP center, was established. Each facility was staffed by a diploma physiotherapist, a community rehabilitation worker, and a community mobilizer and was equipped with basic physiotherapy equipment. The staff were recruited locally and trained on early intervention protocol.
-
Children in the intervention arm received a 60-minute goal directed training (GDT) session weekly at the SMART-CP centers for 12 months. GDT is a goal-oriented, activity-focused, parent-led intervention17 consisting 4 components:
Goal selection: realistic, time-framed goals set with parents, focusing on everyday tasks children found challenging, including gross motor, self-care, communication, play, and school activities.
Assessment: a detailed assessment identified factors affecting goal attainment, including physical requirements, resources, equipment, and task setting.
Intervention: weekly group-based GDT sessions were conducted by community rehabilitation workers supervised by diploma physiotherapists. Parents were guided to support their child in completing motor tasks, using play preferences to encourage self-generated activity. Community mobilizers visited homes every 2 weeks to support home-based GDT.
Outcome evaluation: adherence and progress were evaluated through attendance registers, checklists, and probing questions during GDT sessions, with goal achievement monitored using structured assessments.
-
Telerehabilitation
A centralized telerehabilitation team, including physicians, physiotherapists, speech therapists, and nutritionists, supported the SMART-CP centers. Each child received a 30-minute telerehabilitation session monthly for 12 months. The team supervised interventions, tracked progress, and adjusted plans as needed. The first session involved clinical assessment, goal setting, and planning, with subsequent sessions focused on treatment progress, goal evaluation, and guidance for staff.
Control Arm
Children with CP identified from control clusters were provided with basic education on early intervention and encouraged to access health care from local sources.
Primary and Secondary Outcomes
Outcomes were assessed at 0 and 12 months. The primary study outcome was rehabilitation service uptake, which is whether a child with CP accessed any rehabilitation services at baseline and endline. This includes physiotherapy, occupational therapy, speech and language therapy, the provision of prostheses and orthoses, and the use of assistive devices such as wheelchairs, walkers, or communication aids. Access was determined based on caregiver-reported service use.
Secondary study outcomes included the following: (1) age at CP diagnosis (age at CP diagnosis among children recruited during the study period), (2) age at rehabilitation service commencement (age at which rehabilitation services initiated among children recruited during the study period), (3) access to assistive devices (number of children receiving assistive devices at baseline and endline), (4) number of therapy sessions (total therapy sessions received per child during the study period), (5) mortality rate (number of children who died during the study period), (6) motor functions (gross motor functions assessed using the Gross Motor Function Classification System [GMFCS]18 at baseline and endline and hand functions assessed using Manual Ability Classification System19 at baseline and endline), (7) out-of-pocket expenditure (total amount of out-of-pocket expenses incurred per child for rehabilitation services during the study period, collected through caregiver self-reports at endline), (8) travel distance (distance in kilometers traveled by each child to access rehabilitation services during the study period, (9) episode of illnesses (number of illnesses requiring medical attention per child with CP during the study period), and (10) episode of hospitalizations (number of hospitalizations per child with CP during the study period). A Theory of Change framework of the SMART-CP model is illustrated in eTable 1 in Supplement 2.
Statistical Analysis
Missing data (eTable 2 in Supplement 2) were handled using multiple imputation, assuming data were missing at random. Multiple imputation was performed for all variables with missing values.
An intention-to-treat analysis was used to compare improvements in primary and secondary outcomes between the intervention and control arms. Changes in rehabilitation service uptake within arms were assessed using χ2 tests, as well as the risk ratio (RR) with 95% CI between arms at endline. A baseline-adjusted regression model was developed to adjust for potential confounders. A mixed-effects model was used to account for intracluster correlation. Both χ2 tests and risk ratios were calculated for categorical outcomes. For continuous outcomes, mean differences with 95% CI and Cohen d effect sizes were computed. All analyses were performed using R Studio, version 4.3.3 (R Project for Statistical Computing), with statistical significance set at a 2-sided P value < .05. Data were analyzed from December 2023 to May 2024.
Results
A total of 1044 children with CP were screened, of which 76 were excluded based on the inclusion and exclusion criteria and 968 were enrolled (mean [SD] age, 7.9 [4.9] years; 387 female [40.0%]; 581 male [60.0%]) in the trial (500 in the intervention arm and 468 children in the control arm). All 968 children completed the baseline assessment, and 889 (91.8%) completed the endline assessments. Of the 79 children lost to follow-up, 48 moved out of the study site (intervention arm: 27 and control arm: 21), and 31 died during the study (intervention arm: 14 and control arm: 17) (Figure).
Figure. Consolidated Standards of Reporting Trials Diagram of the Sustainable Model of Early Intervention and Telerehabilitation for Children With Cerebral Palsy (SMART-CP) Trial.
GDT indicates goal-directed training.
Primary Outcome: Rehabilitation Service Uptake Among Children With CP
Between baseline and endline, rehabilitation services uptake significantly increased in the intervention arm (70.2% [351 of 500] vs 99.4% [497 of 500]), compared with the control arm (63.9% [299 of 468] vs 68.2% [319 of 468]). The intervention arm was 1.5 times more likely to access rehabilitation (Table 1).
Table 1. Outcome Variables at Baseline and End Line in the Intervention and Control Arms.
| Outcomes | Intervention | Control | Between group change at endline | |||||
|---|---|---|---|---|---|---|---|---|
| No. (%) | Within group change, P value | No. (%) | Within group change, P value | Risk ratio (95% CI) | P value | |||
| Baseline (n = 500) | Endline (n = 500) | Baseline (n = 468) | Endline (n = 468) | |||||
| Received rehabilitation services | ||||||||
| Yes | 351 (70.2) | 497 (99.4) | <.001a | 299 (63.9) | 319 (68.2) | .19b | 1.52 (1.44-1.61) | <.001 |
| No | 149 (29.8) | 3 (0.6) | 169 (36.1) | 149 (31.8) | ||||
Fisher exact test.
χ2 Test.
The demographics of the children with CP and their parents are outlined in eTable 3 in Supplement 2. Both study arms had similar sociodemographic and clinical characteristics, except for age (median [IQR] age, intervention: 6.3 [3.4-11.0] years vs control: 7.9 [4.4-12.8] years; P < .001), mothers’ education (eg, primary, intervention: 303 of 500 [60.6%] vs control: 232 of 468 [49.7%]; P = .002), fathers’ education (eg, primary, intervention: 270 of 500 [54.0%] vs control: 222 of 468 [47.4%]; P = .05), fathers’ occupation (eg, agriculture/farming, intervention: 127 of 500 [25.4%] vs control: 192 of 468 [41.0%]; P < .001), CP topography (eg, diplegia, intervention: 101 of 393 [26.0%] vs control: 60 of 369 [16.3%]; P = .006), and Manual Ability Classification System (MACS) level (eg, level I-III, intervention: 313 of 500 [62.6%] vs control: 259 of 468 [55.3%]; P = .02).
In the baseline-adjusted model, the intervention arm (odds ratio [OR], 376.5; 95% CI, 107.4-1904.7) and baseline rehabilitation status (OR, 244.7; 95% CI, 96.1-741.4) were strongly associated with a higher likelihood of service uptake. Maternal education also predicted service uptake (illiterate: OR, 0.1; 95% CI, 0-4.8) (Table 2).
Table 2. Baseline-Adjusted and Clustering-Adjusted Models.
| Variable | Category | Baseline-adjusted OR (95% CI) | P value | Clustering-adjusted OR (95% CI) | P value |
|---|---|---|---|---|---|
| Intervention arm | Control | 1 [Reference] | NA | 1 [Reference] | NA |
| Intervention | 376.48 (107.41-1904.73) | <.001 | 335.10 (81.82-3543.72) | <.001 | |
| Baseline rehabilitation | No | 1 [Reference] | NA | 1 [Reference] | NA |
| Yes | 244.72 (96.07-741.42) | <.001 | 1.16 (107.53-1044.31) | <.001 | |
| Age group, y | <6 | 1 [Reference] | NA | 1 [Reference] | NA |
| 6-10 | 1.02 (0.35-2.95) | .97 | 0.70 (0.34-3.01) | .91 | |
| 11-15 | 1.25 (0.39-4.07) | .70 | 1.10 (0.34-3.91) | .81 | |
| 16-18 | 0.95 (0.26-3.55) | .94 | 1.01 (0.27-4.44) | .90 | |
| CP topography | Diplegia | 1 [Reference] | 1 [Reference] | ||
| Hemiplegia/monoplegia | 0.71 (0.22-2.26) | .56 | 1.02 (0.21-2.31) | .80 | |
| Triplegia/quadriplegia | 0.90 (0.21-4.05) | .90 | 0.07 (0.23-4.59) | .98 | |
| Mother’s education | Higher education | 1 [Reference] | NA | 1 [Reference] | NA |
| Illiterate | 0.06 (0.004-0.75) | .03 | 0.19 (0.01-1.00) | .05 | |
| Primary | 0.17 (0.02-1.32) | .09 | 0.84 (0.02-1.57) | .12 | |
| Secondary | 0.62 (0.08-4.92) | .64 | 4.44 (0.10-6.70) | .87 | |
| Father’s education | Higher education | 1 [Reference] | NA | 1 [Reference] | NA |
| Illiterate | 3.81 (0.46-33.86) | .23 | 1.39 (0.46-43.20) | .30 | |
| Primary | 1.74 (0.29-10.59) | .54 | 2.34 (0.21-9.47) | .40 | |
| Secondary | 2.92 (0.48-18.57) | .25 | 1.37 (0.34-16.22) | .44 | |
| Father’s occupation | White-collar job | 1 [Reference] | NA | 1 [Reference] | NA |
| Farming/blue-collar job | 1.32 (0.41-4.37) | .64 | 0.97 (0.39-4.81) | .80 | |
| MACS level | I-III | 1 [Reference] | NA | 1 [Reference] | NA |
| IV-V | 0.93 (0.26-3.24) | .91 | 0.21 (0.28-3.39) | .95 |
Abbreviations: CP, cerebral palsy; MACS, Manual Ability Classification System; NA, not applicable; OR, odds ratio.
In the cluster-adjusted model, the intervention arm remained significantly associated with higher rehabilitation service uptake (OR, 335.1; 95% CI, 81.8-3543.7). Clustering was present (sig01 = 0.21) but contributed little to variance (Table 2).
Secondary Outcomes
At endline, the mean (SD) age at CP diagnosis was significantly earlier in the intervention arm (2.0 [2.0] years) than in the control arm (3.8 [3.3] years), with a mean difference of −1.9 years (95% CI, −2.91 to −0.84; Cohen d = −0.7). Similarly, the mean (SD) age at first rehabilitation commencement was earlier in the intervention arm (1.8 [1.8] years vs 3.6 [2.4] years; mean difference, −1.8 years, 95% CI, −2.64 to −1.05; Cohen d = −0.9). The intervention arm also received more therapy sessions (mean [SD], 23.4 [31.7] vs 4.3 [20.8]; mean difference, 19.1; 95% CI, 15.72-22.44; Cohen d = 0.7) than the control arm.
Episodes of illness (mean [SD], 6.7 [5.5] vs 5.9 [4.4]; mean difference, 0.9; 95% CI, 0.22; 1.48; Cohen d = 0.2) and doctor consultations (mean [SD], 1.8 [3.4] vs 1.8 [3.4]; mean difference, 0.7; 95% CI, 0.26-1.16; Cohen d = 0.1) were higher in the intervention arm. Hospitalizations and length of stay were comparable between arms, with small effect sizes. Travel distances for rehabilitation services were significantly shorter in the intervention arm (mean [SD], 18.1 [36.5] km vs 113.4 [66.6] km; mean difference, −95.3 km; 95% CI, −102.12 to −88.46; Cohen d = −1.8). Additionally, out-of-pocket expenditure was significantly lower in the intervention arm (mean [SD] expenditure, $1.5 [$1.6] vs $2.9 [$5.1]; mean difference, −1.4; 95% CI, −1.89 to −0.93; Cohen d = −0.4). Mortality rate over 12 months was comparable between arms (2.8% [14 of 500] vs 3.6% [17 of 468]; RR, 0.77; 95% CI, 0.38-1.55; P = .47) (Table 3).
Table 3. Secondary Crude Outcomes Among Children With Cerebral Palsy (CP) Between Intervention and Control Arms.
| Outcomes | Intervention | Control | Mean difference (95% CI); Cohen d value |
|---|---|---|---|
| No. of new children with CP recruited at endline | 124 | 43 | NA |
| Age at CP diagnosis in newly recruited children with CP, mean (SD), y | 2.0 (2.0) | 3.8 (3.3) | −1.9 (−2.91 to −0.84); −0.7 |
| Age at first commencement of rehabilitation in newly recruited children with CP, mean (SD), y | 1.8 (1.8) | 3.6 (2.4) | −1.8 (−2.64 to −1.05); −0.9 |
| Therapy sessions received in the last 12 mo, mean (SD) | 23.4 (31.7) | 4.3 (20.8) | 19.1 (15.72 to 22.44); 0.7 |
| Episode of illness in the last 12 mo, mean (SD) | 6.7 (5.5) | 5.9 (4.4) | 0.9 (0.22 to 1.48); 0.2 |
| Frequency of doctor’s consultations in the last 12 mo, mean (SD) | 1.8 (3.4) | 1.8 (3.4) | 0.7 (0.26 to 1.12); 0.2 |
| No. of hospitalizations in the last 12 mo, mean (SD) | 0.2 (0.7) | 0.1 (0.6) | 0.1 (0.00 to 0.16); 0.1 |
| Hospital length of stay in the last 12 mo, mean (SD), d | 1.1 (4.0) | 0.8 (4.0) | 0.3 (−0.24 to 0.76); 0.1 |
| Distance traveled to receive rehabilitation services (in kilometers) in the last 12 mo, mean (SD) | 18.1 (36.5) | 113.4 (66.6) | −95.3 (−102.12 to −88.46); −1.8 |
| Out-of-pocket expenditure for receiving rehabilitation services in the last 12 mo, mean (SD), $ | 1.5 (1.6) | 2.9 (5.1) | −1.4 (−1.89 to −0.93); −0.4 |
| Deceased in the last 12 mo, No. (%) | |||
| Yes | 14 (2.8) | 17 (3.6) | 0.77 (0.38 to 1.55); P = .47a |
| No | 486 (97.2) | 451 (96.4) |
Abbreviation: NA, not applicable.
Fisher exact test and risk ratio (95% CI).
The percentage of children receiving assistive devices increased significantly in the intervention arm (1.6% [8 of 500] to 20.8% [104 of 500]; P < .001), whereas the control arm showed a marginal increase (1.3% [6 of 468] to 3.0% [14 of 468]; P = .11). The between-group difference was statistically significant, with a risk ratio of 0.82 (95% CI, 0.78-0.86; P < .001), favoring the intervention arm (Table 4).
Table 4. Secondary Outcomes of Children With Cerebral Palsy (CP) at Baseline and Endline Between the Intervention and Control Arms.
| Clinical characteristics | Intervention | Control | Between-group changes at endline | |||||
|---|---|---|---|---|---|---|---|---|
| No. (%) | Within-group change, P valuea | No. (%) | Within-group change, P valuea | Risk ratio (95% CI) | P value | |||
| Baseline | Endline | Baseline | Endline | |||||
| GMFCS | ||||||||
| Level I-III | 284 (56.8) | 310 (62.0) | .09 | 244 (52.1) | 276 (59.0) | .09 | 0.91 (0.78-1.06) | .24 |
| Level IV-V | 216 (43.2) | 190 (38.0) | 224 (47.9) | 192 (41.0) | ||||
| MACS | ||||||||
| Level I-III | 313 (62.6) | 318 (63.6) | .39 | 262 (56.0) | 282 (60.3) | .16 | 0.87 (0.74-1.03) | .11 |
| Level IV-V | 187 (37.4) | 182 (36.4) | 206 (44.0) | 186 (37.9) | ||||
| Associated impairments | ||||||||
| No | 95 (19.0) | 87 (17.4) | .57 | 87 (18.6) | 83 (17. 7) | .80 | 1.01 (0.95-1.07) | .87 |
| Yes | 405 (81.0) | 413 (17.4) | 381 (81.4) | 385 (82.3) | ||||
| Received assistive devices | ||||||||
| Yes | 8 (1.6) | 104 (20.8) | <.001 | 6 (1.3) | 14 (3.0) | .11 | 0.82 (0.78-0.86) | <.001 |
| No | 492 (98.4) | 396 (79.2) | 462 (98.7) | 454 (97.0) | ||||
Abbreviations: GMFCS, Gross Motor Function Classification System; MACS, Manual Ability Classification System.
χ2 Test.
At endline, no statistically significant changes were observed in terms of GMFCS level (RR, 0.91; 95% CI, 0.87-1.06; P = .24), MACS level (RR, 0.87; 95% CI, 0.74-1.03; P = .11), and associated impairments (RR, 1.01; 95% CI, 0.95-1.07; P = .87) in either the intervention or control arm (Table 4).
mPower Group Session
A total of 45 mPower groups were formed, comprising mothers of children with CP in the intervention arm. Of these, 448 mothers attended at least 1 session (maximum: 22 sessions; mean [SD], 7.1 [6.4] sessions; median [IQR], 6 [2-13] sessions).
Discussion
This study was the first RCT, to our knowledge, to investigate the effectiveness of a sustainable model for early detection, intervention, and telerehabilitation in children with CP in rural Bangladesh. Our findings demonstrate that the SMART-CP model significantly enhanced access to rehabilitation services in the intervention arm. Secondary analyses suggest additional benefits, such as earlier diagnosis of CP, increased therapy session frequency, improved access to assistive devices, decreased travel distance, and reduced out-of-pocket expenses; although, these should be considered as hypothesis-generating outcomes. Adjusted analyses identified maternal education and baseline rehabilitation status as significant predictors of rehabilitation service uptake. Although the SMART-CP model facilitated service uptake, its influence on survival outcomes remains inconclusive.
The increase in service uptake, including assistive device use, in the intervention arm, demonstrates the model’s effectiveness in improving service accessibility through referral networks, rehabilitation centers, and telerehabilitation services. Additionally, the intervention arm had significantly lower travel distance and out-of-pocket expenses, making services more accessible and affordable for families. A study in Mozambique showed that strengthening referral systems including training of community health workers improved health care access in rural areas.20 Our previous studies also found that community-based early intervention and rehabilitation programs increased access to services,7 improved health outcomes21 and enhanced quality of life of children with CP8 in rural Bangladesh.
In LMICs, where mothers often face stigma22 and balance caregiving with other responsibilities,23 peer support groups like mPower could help reduce social isolation and build community.24 Although this RCT did not focus on caregiver outcomes, a secondary analysis of mPower data showed significant improvement in health literacy and rehabilitation service utilization among mothers in the intervention arm.16 Several models have been developed to empower caregivers of children with CP in LMICs, each with strengths and weaknesses. For instance, the Getting to Know CP model raises CP awareness but lacks peer support.15 The Ubuntu model emphasizes community interconnectedness with a holistic approach, complementing mPower’s structural support.25 The Juntos model involves both parents and children in educational sessions to improve CP knowledge and foster community.26 Together, these models create a comprehensive framework blending education, peer support, and community engagement.
Integrating telerehabilitation with conventional early intervention programs effectively increased service utilization in our study. A recent systematic review found that telerehabilitation improves clinical outcomes and enhances service uptake, especially in remote areas.27 Even in high-income countries like Australia, traditional center-based rehabilitation services are insufficient for children with CP, leading to recommendations for telerehabilitation to improve service quality and uptake.28 These findings underscore the need to expand telerehabilitation to overcome access barriers due to a shortage of trained professionals. However, sustainability remains a challenge in LMICs.29 Private sector partnerships, payment-based services, and scalable technologies could help reduce costs while maintaining high-quality care.
Our data also suggest that the SMART-CP model facilitated earlier diagnosis of CP, which is crucial for timely intervention during the neuroplastic period and improved long-term outcomes.30 A parent-led community-based trial in India also showed that such approaches enhances early identification and developmental outcomes for children with CP.31 This highlights the effectiveness of community-based models in promoting early detection and improving developmental trajectories for children with CP.
However, no statistically significant difference was observed in the clinical measures (eg, GMFCS, MACS, and associated impairments) in either arm. This could be attributed to the relatively short duration of the intervention,32,33 or the insensitivity of the assessment tools used to capture subtle changes over time.34 Nevertheless, these classification systems were said to describe the groups and rehabilitation needs. Future studies should use more sensitive motor assessment tools (eg, Gross Motor Function Measures 66 [GMFM-66]35) and extend the intervention period to better detect clinical improvements among children with CP and well-being (eg, stress and anxiety) of their primary caregivers.
Interestingly, the baseline and cluster adjusted analyses identified maternal education as a significant positive predictor of rehabilitation service uptake, underscoring the importance of maternal education in health-seeking behaviors. Similar findings were reported in Bangladesh, Nepal, Indonesia, and Ghana,4,12 indicating the need for integrating maternal education4 in service delivery models for better outcome.
Our findings highlight the need for equitable, accessible, and sustainable service delivery models for children with CP in LMICs. This finding offers a foundation for future research and policies prioritizing culturally sensitive, evidence-based interventions, leveraging community-based rehabilitation, task shifting, and low-cost assistive devices to address the unmet needs in LMICs. Scalability requires training health care workers for early detection and intervention, strengthening referral systems, and integrating CP care into national policies to streamline resources and infrastructure. Technology-enhanced solutions, such as telehealth, could improve follow-up and accessibility for remote families. Collaboration among governments, nongovernment, and international organizations, including the World Health Organization and United Nations International Children’s Emergency Fund, could drive resource mobilization, ensuring sustainable, appropriate CP care in LMICs.
Our study prioritized representativeness and inclusivity by incorporating diverse perspectives and addressing systemic barriers faced by children with CP and their families in LMICs. Recruitment for this trial spanned varied sociodemographic/economic groups across rural and semi-urban areas. Gender equality was maintained in offsetting up key informant networks. Primary caregivers were involved as a coinvestigator and advisor for trial implementation to amplify lived experience. The mPower groups fostered empowerment and provided a safe space for mothers to combat stigma. By integrating community-driven solutions and stakeholder codesign, our findings offer a scalable and inclusive model for LMICs.
Limitations
Despite our efforts, this study had several limitations. Although the large sample size is a strength, the small number of clusters (n = 8) may have reduced precision and generalizability. To address this, mixed-effects logistic regression was used to account for intracluster correlation. Although rehabilitation access is a key outcome, it inherently reflects the effectiveness of the intervention itself, which may limit its utility as the sole primary outcome measure. Some baseline demographic differences were observed between study arms, likely due to random variation; confounding adjustments were made. Secondary outcomes lacked multiplicity adjustments; therefore, these results should be interpreted as hypothesis generating. The absence of sensitive motor assessment tools may have underestimated clinical outcomes. The study also lacked a process evaluation and cost-effectiveness analysis, thereby limiting insights into scalability and sustainability. The study design did not allow disaggregated analysis of the individual components of the SMART-CP model, making it impossible to determine whether the observed impact was attributable to the SMART-CP centers, the mPower groups, or their combined effect. The short duration of the intervention may have constrained our ability to assess the long-term impacts of the SMART-CP model on rehabilitation outcomes.
Future studies should evaluate the individual components of SMART-CP model (parent groups, SMART-CP centers, and telerehabilitation) on access and outcomes. Longer intervention periods and sensitive tools like the GMFM-66 are required to better track functional changes. Research should also assess parent-level outcomes, such as mental health and quality of life, to capture the broader family impact.
Conclusions
In this cluster RCT, the SMART-CP model demonstrated promising outcomes by significantly improving rehabilitation service uptake and access to assistive devices among children with CP in rural Bangladesh. It also facilitated early CP diagnosis and reduced travel distances and out-of-pocket expenditures associated with accessing rehabilitation services. These findings suggest potential for replication in similar economic settings.
Trial Protocol.
eFigure. Components of the SMART CP Model
eTable 1. Theory of Change Framework for SMART CP Model
eTable 2. Distribution of Missing Data
eTable 3. Characteristics of SMART CP Trial Participants
Data Sharing Statement.
References
- 1.Bright T, Wallace S, Kuper H. A systematic review of access to rehabilitation for people with disabilities in low-and middle-income countries. Int J Environ Res Public Health. 2018;15(10):2165. doi: 10.3390/ijerph15102165 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Al Imam MH, Jahan I, Das MC, et al. Situation analysis of rehabilitation services for persons with disabilities in Bangladesh: identifying service gaps and scopes for improvement. Disabil Rehabil. 2022;44(19):5571-5584. doi: 10.1080/09638288.2021.1939799 [DOI] [PubMed] [Google Scholar]
- 3.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(5):601-609. doi: 10.1111/dmcn.14013 [DOI] [PubMed] [Google Scholar]
- 4.Al Imam MH, Jahan I, Das MC, et al. Rehabilitation status of children with cerebral palsy in Bangladesh: findings from the Bangladesh Cerebral Palsy Register. PLoS One. 2021;16(5):e0250640. doi: 10.1371/journal.pone.0250640 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Bangladesh Bureau of Statistics . Population and Housing Census 2011: National Volume 2-Union Statistics. Accessed March 22, 2024. https://bbs.gov.bd/site/page/47856ad0-7e1c-4aab-bd78-892733bc06eb/Population-and-Housing-Census
- 6.Social Security Programs . Fiscal Year 2023-24. Ministry of Finance; 2024. [Google Scholar]
- 7.Al Imam MH, Das MC, Jahan I, et al. A social business model of early intervention and rehabilitation for people with disability in rural Bangladesh. Brain Sci. 2022;12(2):264. doi: 10.3390/brainsci12020264 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Al Imam MH, Jahan I, Das MC, et al. Supporting People in Extreme Poverty With Rehabilitation and Therapy (SUPPORT CP): a trial among families of children with cerebral palsy in Bangladesh. Dev Med Child Neurol. 2023;65(6):773-782. doi: 10.1111/dmcn.15445 [DOI] [PubMed] [Google Scholar]
- 9.Zampolini M, Todeschini E, Guitart MB, et al. Tele-rehabilitation: present and future. Ann Ist Super Sanita. 2008;44(2):125-134. [PubMed] [Google Scholar]
- 10.Alonazi A. Effectiveness and acceptability of telerehabilitation in physical therapy during COVID-19 in children: findings of a systematic review. Children (Basel). 2021;8(12):1101. doi: 10.3390/children8121101 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Cristinziano M, Assenza C, Antenore C, et al. Telerehabilitation during COVID-19 lockdown and gross motor function in cerebral palsy: an observational study. Eur J Phys Rehabil Med. 2022;58(4):592-597. doi: 10.23736/S1973-9087.21.07132-X [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Al Imam MH, Jahan I, Muhit M, et al. Predictors of rehabilitation service utilization among children with cerebral palsy (CP) in low-and middle-income countries (LMIC): findings from the Global LMIC CP Register. Brain Sci. 2021;11(7):848. doi: 10.3390/brainsci11070848 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Muhit MA, Shah SP, Gilbert CE, Hartley SD, Foster A. The key informant method: a novel means of ascertaining blind children in Bangladesh. Br J Ophthalmol. 2007;91(8):995-999. doi: 10.1136/bjo.2006.108027 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Donner A, Klar N. Design and Analysis of Cluster Randomization Trials in Health Research. John Wiley & Sons, Inc; 2000. [Google Scholar]
- 15.Cerebral Palsy Association . Getting to Know Cerebral Palsy. Hambisela; 2008. [Google Scholar]
- 16.Perrins G, Jahan I, Khan MN, et al. The mPower (mother’s power) initiative: improving health behavior through peer support and health literacy for mothers of children with cerebral palsy in rural Bangladesh. Children (Basel). 2024;11(12):1438. doi: 10.3390/children11121438 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Mastos M, Miller K, Eliasson AC, Imms C. Goal-directed training: linking theories of treatment to clinical practice for improved functional activities in daily life. Clin Rehabil. 2007;21(1):47-55. doi: 10.1177/0269215506073494 [DOI] [PubMed] [Google Scholar]
- 18.Palisano R, Rosenbaum P, Walter S, Russell D, Wood E, Galuppi B. Development and reliability of a system to classify gross motor function in children with cerebral palsy. Dev Med Child Neurol. 1997;39(4):214-223. doi: 10.1111/j.1469-8749.1997.tb07414.x [DOI] [PubMed] [Google Scholar]
- 19.Eliasson AC, Krumlinde-Sundholm L, Rösblad B, et al. The Manual Ability Classification System (MACS) for children with cerebral palsy: scale development and evidence of validity and reliability. Dev Med Child Neurol. 2006;48(7):549-554. doi: 10.1017/S0012162206001162 [DOI] [PubMed] [Google Scholar]
- 20.Give C, Ndima S, Steege R, et al. Strengthening referral systems in community health programs: a qualitative study in 2 rural districts of Maputo Province, Mozambique. BMC Health Serv Res. 2019;19(1):263. doi: 10.1186/s12913-019-4076-3 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Karim T, Muhit M, Jahan I, et al. Outcome of community-based early intervention and rehabilitation for children with cerebral palsy in rural Bangladesh: a quasi-experimental study. Brain Sci. 2021;11(9):1189. doi: 10.3390/brainsci11091189 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.McLean S, Halstead EJ. Resilience and stigma in mothers of children with emotional and behavioral difficulties. Res Dev Disabil. 2021;108:103818. doi: 10.1016/j.ridd.2020.103818 [DOI] [PubMed] [Google Scholar]
- 23.Gagnon RJ, Garst BA, Kouros CD, Schiffrin HH, Cui M. When overparenting is normal parenting: examining child disability and overparenting in early adolescence. J Child Fam Stud. 2020;29:413-425. doi: 10.1007/s10826-019-01623-1 [DOI] [Google Scholar]
- 24.Chakraborti M, Gitimoghaddam M, McKellin WH, Miller AR, Collet JP. Understanding the implications of peer support for families of children with neurodevelopmental and intellectual disabilities: a scoping review. Front Public Health. 2021;9:719640. doi: 10.3389/fpubh.2021.719640 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Mbigi L, Maree J. Ubuntu, the Spirit of African Transformation Management. Knowledge Resources; 1995. [Google Scholar]
- 26.Ubuntu-Hub . Juntos. https://www.ubuntu-hub.org/resources/juntos/
- 27.Wang Z, He K, Sui X, et al. The effect of web-based telerehabilitation programs on children and adolescents with brain injury: systematic review and meta-analysis. J Med Internet Res. 2023;25:e46957. doi: 10.2196/46957 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Edirippulige S, Reyno J, Armfield NR, Bambling M, Lloyd O, McNevin E. Availability, spatial accessibility, utilization and the role of telehealth for multidisciplinary paediatric cerebral palsy services in Queensland. J Telemed Telecare. 2016;22(7):391-396. doi: 10.1177/1357633X15610720 [DOI] [PubMed] [Google Scholar]
- 29.Handicap International–Luxembourg Aid & Development . Barriers and levers for the use of telerehabilitation through experimentation in 3 countries. Accessed May 5, 2024. https://www.hi.org/sn_uploads/document/barriersandlevers_telerehabilitation_rs16.pdf
- 30.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(9):897-907. doi: 10.1001/jamapediatrics.2017.1689 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Benfer KA, Whittingham K, Ware RS, et al. Efficacy of early intervention for infants with cerebral palsy in an LMIC: an RCT. Pediatrics. 2024;153(4):e2023063854. doi: 10.1542/peds.2023-063854 [DOI] [PubMed] [Google Scholar]
- 32.Xie J, Jiang L, Li Y, et al. Rehabilitation of motor function in children with cerebral palsy based on motor imagery. Cogn Neurodyn. 2021;15(6):939-948. doi: 10.1007/s11571-021-09672-3 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Kushnir A, Kachmar O. Intensive neurophysiological rehabilitation system for children with cerebral palsy: a quasi-randomized controlled trial. BMC Neurol. 2023;23(1):157. doi: 10.1186/s12883-023-03216-4 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Piscitelli D, Ferrarello F, Ugolini A, Verola S, Pellicciari L. Measurement properties of the Gross Motor Function Classification System, Gross Motor Function Classification System-Expanded & Revised, Manual Ability Classification System, and Communication Function Classification System in Cerebral Palsy: a systematic review with meta-analysis. Dev Med Child Neurol. 2021;63(11):1251-1261. doi: 10.1111/dmcn.14910 [DOI] [PubMed] [Google Scholar]
- 35.Brunton LK, Bartlett DJ. Validity and reliability of 2 abbreviated versions of the Gross Motor Function Measure. Phys Ther. 2011;91(4):577-588. doi: 10.2522/ptj.20100279 [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Trial Protocol.
eFigure. Components of the SMART CP Model
eTable 1. Theory of Change Framework for SMART CP Model
eTable 2. Distribution of Missing Data
eTable 3. Characteristics of SMART CP Trial Participants
Data Sharing Statement.
