Statement of main findings
Our two-step analysis shows a significant association for older people being included in the PMC program and making visits to government health posts on a planned rather than unplanned basis, when compared to a matched set of people not in PMC. Applying the same analytical method, we find being included in PMC was significantly associated with a higher proportion of rehabilitation visits to health posts, as opposed to other motives for visits.
Limitation of the study
Our study design has several limitations. First, the scarcity of individual-level data does not permit us to identify potential predictors of service use. To compensate for this, we use a rich set of neighborhood-level data. Second, although matching techniques enable balancing of observations based on observed covariates, unmeasured confounding variables may still be present in our analysis. We address any hidden bias through sensitivity tests and comparing post-match covariates between groups, and these indicate that the estimates of treatment effects we report are robust. Third, in neither the PMC nor the control groups do we observe any patients making many visits, which might have biased our estimates. As a robustness test, we performed the same analysis excluding people making more than four and five visits. The number of such people was very small and so the findings were almost identical to those of our analysis.
Using a complete dataset by discarding observations with missing data on covariates is done to prevent introducing bias on the matching process. As matching implies finding similar observations based on covariates, the lack of data relies on an assumption of non-relevance of that piece of information. As results are robust modelling with complete dataset, and discarded observations were not outliers, our approach is avoids introducing bias to the analysis.
Finally, we dismiss possible spillover effects over the health and care systems, which could occur under the scarcity of the limited resources available on outpatient health services. This is based on previous research that does not suggest substitution of services in LMICs [30], which was confirmed by our local stakeholders.
Comparisons to other studies
Our descriptive finding that 92% of people aged 60 or more in Belo Horizonte made no visit to a government health post between April and June 2018 indicates older people’s engagement with the city’s supposedly universal public primary health care system was limited. This finding does not match the results of other studies in Brazil, which report higher rates of health post utilization by people at older ages [31]. For example, a national survey of Brazilians aged 50 and over reported that 63.5% had made at least one visit to a government health post during the previous year [32]. One explanation for the lower rate of health post utilization reported in our study is that it refers to a three-month period rather than 12 months. Some specific local factors may also be relevant. For example, Belo Horizonte has a much hillier terrain than most Brazilian cities and its poorer neighborhoods are characterized by very steeply sloping streets, creating specific difficulties for accessing services for older people with limited mobility and who lack private transport.
Studies about the frequency and determinants of outpatient health service use by older people in Brazil and other countries do not look at the same outcomes of interest as in our study. With reference to the UK [33], it has been observed that: “there has been little attention to acuity of presentation to GPs during the working week, and in particular, multi-morbid community-dwelling older person’s utilization of planned and unplanned GP care.” Consequently, it is not possible to make direct comparisons between our two main findings and the wider literature.
Our findings accord with qualitative research on PMC which indicates participation in the programme promotes effective engagement with local health services [20]. The unusual nature of PMC limits comparisons with interventions involving teams of clinical and non-clinical professionals, such as home-based primary health care or hospital at home [34]. PMC’s use of lay carers shares elements with initiatives promoting community-based support for dependent older people in other developing countries [35]. However, these schemes rely on volunteers rather than paid carers and have not been subject to quantitative analysis.
Meaning of this study for policy
This study provides some evidence that interventions like PMC are associated with a more efficient use of scarce health resources. Participation in PMC is associated with a higher share of outpatient visits for the purpose of rehabilitation. This is in keeping with Brazil’s national health system protocol that primary health care providers should have lead responsibility for identifying and managing adult rehabilitation needs [29]. Leading causes of hospitalization of older people in Brazil include hip fracture and stroke, and studies in other countries demonstrate the benefits of outpatient rehabilitation for these conditions [36, 37]. They also demonstrate the potential cost savings from outpatient rehabilitation. For example, analysis of average monthly post-stroke care costs in the USA found that services provided in outpatient settings cost less than a sixth of those provided as inpatient services [1, 13] . When provided in the home setting as part of an inter-disciplinary intervention, this can reduce the need for inpatient hospital care.
This study finds an association between participation in PMC and a lower share of outpatient visits that were unplanned. This is likely to enhance both the technical and allocative efficiency of health services. A UK official review found that growing utilization of urgent and emergency outpatient care is leading to mounting costs and increased pressure on resources [38].
Unanswered questions and further research
Key related areas for future research fall into two broad areas. First, there is an urgent need to identify and categorize other examples of interventions and policy experimentation that share some elements with PMC. This will establish the degree to which PMC is a unique experience or is representative of wider policy trends in Brazil and beyond.
Second, there is a need to develop comparative evidence about the effects of these different interventions. Doing so will be vital for addressing pressures on health services resulting from population ageing, the COVID-19 pandemic and fiscal austerity. Currently, this comparative evidence remains very limited [4, 34, 39, 40]. Our findings contribute to that evidence base, with reference to a specific set of effects for a single intervention. Research from Brazil, the UK and other countries shows that inadequate social care for older people in the community can contribute significantly to otherwise avoidable hospital and care home admissions [41,42,43,44]. We were unable to explore whether PMC does this.