Patient choice and provider incentives : socioeconomic differentials in effects from market reform in Swedish primary care

During the last decades, market reforms have been increasingly implemented in publicly financed health care systems. Still, there is a lack of knowledge of how these reforms affect inequality in access to care and in turn health inequality. This thesis aims to provide such evidence.

The overarching aim of the thesis is to increase the understanding of how market mechanisms affect the socioeconomic (SES) distribution of access to, utilization of, and outcomes from, primary health care (PHC) in a publicly financed health care system. The thesis also aims to further disentangle the effect of three interacting market mechanisms: patient choice, provider location, and changes in reimbursement models. We operationalize our aim using a set of reforms implemented in Swedish PHC in 2007-2009.

The thesis consists of four sub-studies which are all based on comprehensive individual level total population registry data. Study I explores SES differentials in PHC utilization, using concentration indices and data for the three largest Swedish regions, which implemented the reforms at slightly different points in time. Study II estimates the causal separate but interdependent effects of PHC unit location and patient choice on SES differences in patient mix across PHC units, using data from two regions, a counterfactual approach and a decomposable segregation index. Study III assesses to what extent reform exposure had differential impact on PHC outcome quality for low-income elders as compared to matched peers, using a difference-in-difference-in-differences approach, comparing the development in one region exposed to reform, to that of another (at the time) unexposed region. Outcome quality is approximated using avoidable hospitalizations. Finally, Study IV exploits the fact that a smaller set of PHC units in one region for a transitional period did not enter into a new reimbursement design that included part per visit payments. We leverage this variation using difference-in-differences models, and compare the development of PHC visits, secondary care outpatient visits and avoidable hospitalizations by age group, chronic disease and SES factors.

We find that following market reform, individuals with lower SES increased their number of visits more than those with higher SES. This occurred to some extent across all studied regions, but was more pronounced and clearly reform-timed in the region that implemented part per visit payments. Further, within the region with part per visit payments, the effect was slightly larger for PHC units exposed to the new reimbursement design, compared to a subset of units that temporarily remained unexposed. In the latter case, differences were however modest and specific to older individuals. The higher increase in the number of visits for individuals with lower SES was partly explained by visits close in time to a previous visit, and particularly so in the presence of part per visit payments. This may reflect a subdivision of visits in response to reimbursement incentives particularly for individuals with lower SES, but since data do not reveal the relevance, content, length or quality of each visit, the exact interpretation of the increases in recurrent visits remains to be further understood. We also find indications that PHC visits may have substituted secondary outpatient care when PHC units were in part paid per visit, but these findings, and their qualitative implications for individuals with different SES, also need to be further explored.

We further find that market reform increased differences in SES patient mix between PHC units. This was mainly due to a smaller group of new PHC units in the capitation-based region locating in low SES areas, rather than PHC units selecting high SES patients or patients clustering to certain providers through choice patterns. It remains to be explored if the geographical access increases in lower SES areas translated into increased utilization of PHC or, for that matter, better health outcomes.

Finally, we find that market reform had on average negative effects on health care quality for individuals with lower SES, when measured as avoidable hospitalizations in a specific population of older adults. Results are ambiguous as to whether this differed by reimbursement model.

The thesis takes a comprehensive scope on SES differentials in effects from market reform in publicly financed PHC. It thereby contributes with both broad and in-depth empirical evidence to a sparsely researched but highly policy relevant research area. The findings suggest that market reforms do influence socioeconomic differentials in PHC, and that reimbursement models may play an important role in shaping these effects. It concludes that, for individuals with lower SES, in the largest Swedish regions, market reforms appear to have increased health care access, both in terms of geographical proximity and visit volume, but effects on health care content and quality are either unknown, ambiguous or point to negative effects that are specific to individuals with lower SES.

The results call for vigilance and curiosity regarding potential heterogeneities in effects of market models in PHC. Most importantly, they emphasize that policymakers need to closely and continuously monitor provider reimbursement models, as their effects - though potentially modest on average - may disproportionately impact individuals with lower SES. The findings also highlight the need for future studies to employ more granular data on diagnoses and treatments to better understand how market reforms impact the content and quality of care.

List of scientific papers

I. Sveréus S, Kjellsson G, Rehnberg C. Socioeconomic distribution of GP visits following patient choice reform and differences in reimbursement models: Evidence from Sweden. Health Policy. 2018 Sep;122(9):949-956. https://doi.org/10.1016/j.healthpol.2018.07.017

II. Sveréus S, Kjellsson G, Rehnberg C. Market reform and socio-economic segregation in primary care - a counterfactual approach to separating the effects of provider location and patient choice. [Manuscript]

III. Sveréus S, Petzold M, Rehnberg C. Change in avoidable hospitalizations for low-income elders following quasi-market reform in primary care - Evidence from a natural experiment in Sweden. Soc Sci Med. 2024 Apr;346:116711. https://doi.org/10.1016/j.socscimed.2024.116711

IV. Sveréus S, Rehnberg C, Kjellsson G. Heterogeneous effects of part per visit payments in primary care: insights from a natural experiment in Sweden. [Manuscript]