Geographic equity in primary health care performance in Finland: from individual socioeconomic position into the blind spot of the service system

Primary health care (PHC) has been shown to promote socioeconomic equity in health. However, geographic equity in health outcomes related to PHC performance remains little studied in Finland. This gap mirrors both the limited number of suitable indicators for comprehensive PHC performance evaluation and the complexity of measuring PHC performance at the population level. This thesis applied hospitalisations for ambulatory care sensitive conditions (ACSCs) and mortality potentially preventable by health policy and care (amenable mortality) to measure PHC as well as overall health care performance in Finland. These internationally utilized, population-level proxy health outcomes are suggested to be preventable by timely care. The aims of this thesis were to analyse what individual and area-level variables over time explained geographic variation in ACSCs in Finland – and what kind of over time developmental paths would emerge among areas providing public PHC in Finland when these areas were clustered by their age-standardised ACSC subgroup rates. Moreover, this thesis also aimed to analyse how geographic disparities in amenable mortality developed over time between three geographic areas in Finland – and whether this development disfavoured the capital City of Helsinki due to its increasing residential differentiation. ACSCs were obtained and identified from the Finnish Care Register for Health Care for the total Finnish adult population – and amenable mortality respectively from the Causes of Death statistics for all Finns aged 25-74. These outcomes were then linked to individual sociodemographic data and allocated into areas providing public PHC by individual area of residence. The associations between several selected variables and geographic variation in ACSCs were analysed with three-level nested Poisson regression models (individuals nested in areas providing PHC, nested in areas providing hospital care). The proportion of variance explained by each variable was quantified at three time points in 2011-2017 and in two separate datasets (all ACSCs and ACSC emergency admissions). In the null model, variances between areas providing hospital care were up to twice that between providers of PHC. While individual incomes and comorbidities explained up to third of the variances at both area-levels, area-level disease burden and arrangement and usage of hospital care explained an additional 14-16% and 32-33% of these variances in all ACSCs – and 7-15% and 28-33% in ACSC emergency admissions. After these adjustments the remaining variances in the two area-levels emerged to be nearly alike. To identify geographic disparities over time in the level and development of ACSCs in Finland, a group-based multi-trajectory model was applied. This model clustered areas providing PHC by their annual age-standardised acute, chronic and vaccine-preventable ACSC rates in 1996-2013. Moreover, it was tested which within-cluster values of area-level variables differed between the clusters over time. Three clusters emerged – each of them having a distinct level and development of ACSC rates. In these clusters, chronic ACSC rates over time halved, acute ACSC rates stagnated and vaccine-preventable ACSC rates increased slightly. The northern cluster had constantly the highest ACSC rates. While between-cluster absolute disparities in chronic ACSCs diminished over time, the respective relative disparities stagnated. Moreover, both of these disparities increased in acute and vaccine-preventable ACSC rates disfavouring the northern cluster. However, areas within the rural northern cluster shared the highest disease burden and usage of GP led inpatient wards – as well as the lowest education level and use of private health and dental care. Over time the development of amenable mortality was assessed both within and between three geographic areas in Finland: the city of Helsinki, the nine next most populated municipalities and the rest of Finland. Within these areas, development of geographic disparities in amenable mortality were quantified with Gini coefficients – and development of socioeconomic disparities with concentration indices. Finally, both the levels and over time changes of these disparities were compared between the three geographic areas. Over time geographic disparities in amenable mortality within the three geographic areas remained stable, but the socioeconomic ones slightly increased in the favour of the affluent population. The increase in socioeconomic disparities seemed to mirror both stagnating mortality rates in the lowest income quintile and otherwise consistent gradient in decreasing mortality rates among those with higher incomes. However, over time development in both of these within-area disparities were similar between the three geographic areas – and no hypothesised effect for increasing residential differentiation on mortality was found. To conclude, if age- and gender-adjusted ACSCs are applied to compare PHC performance between local providers in Finland, these values are not only confounded by individual socioeconomic position and health status but also by areas’ disease burden and variables related to hospital care. Indeed, when assessing over time geographic disparities in ACSCs, rural northern Finland seemed to be lagging behind the other parts of the country – possibly due to both high usage of GP led inpatient wards (low-threshold basic level hospital care) and excess disease burden in northern Finland. Either way, this finding emphasizes the need to strengthen health care in rural northern Finland. Finally, despite increasing residential differentiation in Helsinki, disparities in its health care performance did not diverge from those observed elsewhere in Finland – which might mirror the effects of policies of positive discrimination and social mixing applied in Helsinki.