Stroke rehabilitation is integral to enhancing recovery and restoring patients’ ability to perform ADLs, which are often compromised following a stroke. Effective rehabilitation can reduce the burden on caregivers, improve patients’ functional status and lower healthcare costs by decreasing the likelihood of readmission and long-term institutional care (Langhorne, Bernhardt & Kwakkel 2011).

However, the challenges of stroke rehabilitation in private facilities are threefold. Firstly, rehabilitation case management and care coordination are driven from a funder or medical insurance perspective with a focus on cost containment and cost savings (Medford-Davis et al. 2016). Consequently, value-based stroke care that occurs at institutional level may be thwarted if the concept of value is defined by the health outcomes achieved in monetary currency (Porter & Kaplan 2016; Zafar et al. 2020). This, in turn, may lead to poor rehabilitation outcomes and suboptimal integration into communities of stroke survivors, with little funds allocated to prevent the readmission of discharged patients. Secondly, the role of nurses in stroke rehabilitation is poorly defined, including their responsibilities in evaluating functional outcomes using measures such as the BETA scale. Discharge planning, especially for early supported discharge, is a key nursing responsibility, and with appropriate post-registration training in rehabilitation (Byrne et al. 2025; Zhao et al. 2024), nurses can play a significant role within multidisciplinary rehabilitation teams. Thirdly, social workers and occupational therapists are integral to the discharge, return to work and reintegration into communities of stroke survivors. Although they are responsible for developing seamless community integration pathways and support systems for patients with functional loss, social workers are neither acknowledged nor supported by the funder case management system.

In the public sector too, rehabilitation capacity in South Africa is severely constrained relative to the population’s needs. As recently as 2 years ago, the public sector had 286 rehabilitation beds designated for long-term stroke care for adult patients – a fraction of what is needed given the high incidence of stroke (Louw et al. 2023). Rehabilitation facilities are generally indiscriminate and may include any setting outside of acute care, e.g. step-down units, general wards, outpatient departments (OPD’s), elderly care, designated rehabilitation units, etc. However, most of these facilities are in the private sector, making them inaccessible to large parts of the indigent population. Furthermore, rehabilitation authorisation decisions such as triage, follow-up and discharge are guided by clinical data interpreted by nursing case managers – a process that is discipline-specific rather than comprehensive functional assessments such as ADLs. This can lead to suboptimal care pathways and hinder reasonable value-based rehabilitation outcomes for stroke patients.

To address the need for more structured outcomes data in rehabilitation, the South African Database for Functional Medicine (SADFM) was developed. The SADFM collects real-time data on patient functional outcomes over time and settings, using a suite of routine outcome measures. One of these outcome measures is the 18-item BETA scale (Loubser, Casteleijn & Bruce 2015).

This study aimed to evaluate functional outcomes in patients with stroke at private rehabilitation facilities in South Africa over an 18-year period, using the BETA scale data associated with functional gain and deterioration. The objectives were to: Establish baseline demographic and functional variables of patients admitted with stroke to private facilities providing rehabilitation services; determine the distribution of functional gain and/or deterioration between the date of discharge and admission; and investigate the association between demographic and functional variables.

Patients diagnosed with stroke in private rehabilitation facilities nationally were the target population (N = 9010); the study population were the electronic records in SADFM, of stroke patients admitted to these facilities between January 2006 and May 2023. No other selection criteria were applied, and the total number of records (n = 9010) was included in the analysis.

Rehabilitation facilities were purposively selected and included if they were registered with the South African Board of Healthcare Funders and had a practice number (n = 47). Facilities offering rehabilitation services included dedicated sub-acute care facilities (86.25%), general wards in hospitals (12.82%) and the remainder (< 1%) received rehabilitation services at home (home-based care), elderly care facilities or as outpatients.

The data source for secondary analysis was the SADFM electronic records, which comprised patient ADL data. Using direct observation, these retrospective data were collected and entered by nurses, therapists and caregivers who were trained, accredited and tested for competence (set at 80%), and reliability in using the 7-point BETA scale. Although BETA scale data were routinely collected during the length of stay, BETA scores at only two key time points, namely, on admission to the rehabilitation facility and at discharge, were applicable in this study.

The BETA scale is embedded in the nursing process and used as a routine observational measure of the patient’s functional ability to execute their ADLs in motor and cognitive domains, rated between one and seven. It comprises 18 items: 13 items measure motor ability associated with self-care (eating, grooming, bathing, dressing upper and lower body and toileting), sphincter control (bowel and bladder control) and mobility (transfers from bed or chair, toilet and bath, walking or wheelchair ambulation and stairs); five items measure cognitive abilities, including comprehension, expression, problem solving, social interaction and memory.

The analysis included demographic, clinical and facility-related variables available in the database to determine their potential influence on functional outcomes. Demographic factors were gender, race, age on admission, length of stay and the only clinical variable, stroke type or stroke description. Stroke description documented by healthcare staff included stroke-bilateral involvement, stroke-left body involvement, stroke-right body involvement, stroke-no paresis and stroke-unspecified.

Secondary data analysis was performed using Stata version 16. Descriptive statistics summarised demographic and clinical characteristics of the patient cohort, with parametric or nonparametric methods applied based on the data distribution. Functional outcomes were assessed by calculating the percentage change in BETA scores between admission and discharge. The percentage was derived by dividing the BETA score at each time point by the total possible score, then multiplying by 100 to standardise the measure. Patients were classified into eight distinct outcome groups: < 0% = functional deterioration, 0% = no gain, 1% – 20% = gain, 20% – 40% gain, 40% – 60% gain, 60% – 80% gain, 80% – 85% gain and 85% – 100% gain. This classification allowed for a view of the range of functional gains and losses experienced by patients during rehabilitation.

A trend analysis was conducted to assess the distribution of functional gain scores (any positive increase in BETA score percentage from admission to discharge), identifying any changes in rehabilitation outcomes over time. Linear regression models were computed to evaluate the predictors of functional gain as a continuous outcome, with functional gain defined as a positive increase in BETA scores from admission to discharge. Logistic regression analysis was performed with binary outcomes for functional gain (functional gain = yes) and functional deterioration (functional gain = no), incorporating all available demographic and facility-related variables as predictors. Functional deterioration was defined as any decrease in BETA scores from admission to discharge. This approach allowed for the identification of key factors associated with both positive and negative functional outcomes, providing insight into the influence of patient and facility characteristics on rehabilitation effectiveness. Missing data were reported in the respective tables. A level of significance of p < 0.05 was used.

Ethical clearance to conduct this study was obtained from the University of the Witwatersrand Human Research Ethics Committee (Medical). The ethical clearance number is M231140 M240802-A0002. The BETA score data used for this study had been previously collected and secured in the SADFM database. A facility licence was issued by the SADFM to participating rehabilitation facilities to collect the BETA scale data as patient-based evidence of their outcome performance achieved. To ensure confidentiality and anonymity, data collectors signed a compliance agreement in terms of the Protection of Personal Information Act (South Africa 2013) and received a personalised, unique login and password to enter patient scores into the SADFM database. All data columns with patient and facility identifiable data were encrypted by the SADFM technical team before they were made available to the researchers. The SADFM and the University of the Witwatersrand have signed a collaboration agreement, whereby de-identified data sets are accessible for further analyses by researchers.

Table 2 shows the functional outcomes assessed using the BETA scale. On admission, the median BETA score was 48% (IQR: 28%–71%), indicating substantial functional limitations amongst patients upon entering rehabilitation. At discharge, the median BETA score was 47% (IQR: 27%–70%), reflecting deterioration in functional ability over the rehabilitation period. Cognitive function and motor function scores on admission and discharge showed negative change or no change, respectively, underscoring poor functional gains overall.

Figure 1 illustrates the distribution of functional gain amongst patients, post-stroke, during rehabilitation. A substantial proportion of patients (40.83%) experienced a decline in functional ability, with scores decreasing from admission to discharge. An additional 19.47% of patients showed no functional gain, resulting in a combined 60.3% of patients who either deteriorated or showed no gains at all. Of those who achieved positive gains, the majority (31.9%) recorded only modest improvements in the 1% – 20% range. A smaller fraction of patients reached higher levels of functional gain: 6.08% achieved functional gains between 20% and 40%, 1.28% between 40% and 60% and 0.39% between 60% and 80%, whilst gains above 80% were extremely rare.

Figure 2 shows the number of patients with stroke admitted to rehabilitation facilities between 2006 and 2023, alongside the trend in functional gain percentages over time. Despite fluctuations in the number of patients admitted, the functional gain percentages remained relatively stable, indicating a consistent, limited improvement in functional recovery across the years.

Linear regression analysis (Table 3) identified several key predictors of functional gain. Female patients showed higher mean gain scores compared to males, suggesting gender-based differences in functional outcomes; these differences were statistically significant (p = 0.006). Racial disparities were also observed, with Black, Mixed-race and White patients exhibiting significantly lower functional gains than those in the ‘Other’ category, with Black patients having the largest negative association (p < 0.001). The type of facility influenced functional outcomes; patients receiving rehabilitation whilst in general wards showed lower functional gains compared to those in sub-acute facilities (p = 0.001). The stroke involvement or description category also impacted outcomes, with patients who suffered a left-sided stroke demonstrating higher functional gains (p = 0.005). Higher baseline motor function scores at admission were positively correlated with functional gains (p < 0.001), highlighting the importance of initial motor abilities for recovery potential.

Logistic regression analysis (Table 4) identified predictors of functional deterioration; regression analysis was limited to 7298 patients who experienced either gain or deterioration; patients with 0% gain were removed from the analysis (n = 1712). Black and White patients had higher odds of deterioration compared to other racial groups (p < 0.001). Facility type was also a significant factor as admission to general wards increased the likelihood of deterioration (p = 0.001), further supporting the potential benefit of sub-acute care facilities or similar forms of transitional care in promoting recovery. Stroke clinical characteristics that influenced deterioration risks included patients who had right-sided body stroke involvement and those without paresis, showing higher odds of deterioration. Conversely, higher motor scores on admission were associated with a reduced risk of deterioration (p < 0.001), reinforcing the impact of initial functional status on recovery outcomes. Additionally, cognitive scores on admission were significantly associated with an increased risk of functional deterioration at discharge (p < 0.001). This suggests that the higher the BETA cognitive functioning of stroke patients on admission, the more likely their total BETA scores will show a decline by discharge.

This study has several limitations that should be considered when interpreting the findings. The SADFM, which primarily includes patients from private rehabilitation facilities, may limit the generalisability of results to the broader South African population, particularly those receiving care in public or rural settings. Whilst efforts were made to include diverse facility types within the private sector, findings may not fully represent the rehabilitation experiences of patients in resource-limited environments. Additionally, the analysis was based solely on functional scores recorded at admission and discharge, without follow-up data to assess long-term recovery outcomes. This limitation restricts insights into the sustainability of functional improvements and potential regression post-rehabilitation.

Although the BETA scale has been validated in clinical settings, it may not capture all aspects of functional recovery, especially subtle changes in ADL and quality of life. The decision to report functional scores as percentages rather than raw BETA scale scores was made to improve interpretability. However, this approach may not fully convey the clinical significance of changes in functional ability. Future studies should explore calibration methods to improve the interpretability of percentage-based functional scores and ensure consistency with rehabilitation benchmarks.

Confounding variables, such as socioeconomic status, comorbidities and differences in rehabilitation protocols across facilities, could not be fully accounted for in this study. Whilst regression models were used to adjust for key demographic and clinical factors (e.g. age, gender, stroke type and facility type), residual confounding variables remain a possibility.

Finally, patients’ length of stay posed a potential limitation, as longer periods of rehabilitation may have influenced functional outcomes. To minimise this impact, length of stay was included as an adjustment variable in the analysis; however, future studies should investigate how rehabilitation intensity, duration and facility-specific practices contribute to stroke recovery. Despite these limitations, the study provides valuable insights into stroke patients’ functional outcomes in private rehabilitation facilities, highlighting critical areas for improvement in rehabilitation practices and policy.

Standardised rehabilitation guidelines grounded in evidence, and practical and applicable to local contexts, are key. Policymakers are required to focus on expanding access to multidisciplinary rehabilitation teams throughout the services, whilst developing and implementing policies that incentivise specialised staff availability in underserved areas.

Variation in recovery results across different facility types indicates that sub-acute care settings likely offer a better rehabilitation environment for stroke survivors. The healthcare system could see improved stroke outcomes through the development of sub-acute facilities and the adoption of these models in general wards. The implementation of such policy changes could serve public health interests because rehabilitation plays a vital role in both reducing disability and improving the quality of life amongst stroke survivors.

Further studies are necessary to determine stroke rehabilitation results in the public sector and assess interventions that can improve recovery across both public and private facilities. The sustainability of functional recovery after discharge could be better understood through longitudinal research that tracks patient outcomes and highlights critical factors for long-term rehabilitation success. Qualitative research that examines patient and provider perspectives can reveal facilitators of and obstacles to successful rehabilitation and direct service improvements.