Context: The Romanian healthcare system suffers from critical shortages of healthcare workers, particularly in rural and remote areas. Moreover, the level of health expenditure in Romania is well below the EU average. Current evidence shows that a low level of population access to health professionals is more predominant in areas and communities with an increased social deprivation index. In Romania, community healthcare workers (community nurses and Roma health mediators) emerged in the past 20 years as professional categories essential in decreasing health service-related inequalities, given their ability to reach underserved communities. For this study, we aimed to evaluate if the documented marginalization rates in Romania are predicted by the selected health worker-related variables.
Methods: We analyzed several variables retrieved from a retrospective database, provided by the Ministry of Health, on community nurses and Roma health mediators and its corresponding needs in Romania. The dataset employed for the analysis had a sample of 935. We employed the following variables: NUTS 2 region, terrain type, type of settlement, marginalization rates (primary outcome), three variables on different metrics related to GPs, and two variables on supplementary vacancies for community workers and health mediators. For NUTS2 and terrain type, we conducted a Gaussian regression GLM and conducted bidirectional stepwise selection applying the BIC criterion. The analyses were conducted in R v4.3.3.
Results: Based on the results of the _rst Gaussian GLM, three predictors were kept – the two variables on the supplementary vacancies needed for community health workers and one of the GP-related metrics. Although all _nal predictors were statistically signi_cant (p<0.01), the type of effect varried, with the vacancies needed for community health workers and the GP-related metric having a negative effect (β = -0.049, β = -0.100) while the vacancies needed for health mediators having a positive effect (β = 0.113) The second model kept, in addition to the variables kept in model 1, the variable on settlement type. All predictors were statistically significant, with the subcategories communes and towns having higher negative effects (β = -10.280, β = -7.585) as compared to previous results. As in the _rst model, the vacancies needed for community health workers and the GP-related metric had negative effects on the marginalization rate (β = -0.035, β = -0.069) while the vacancies needed for health mediators had a positive effect (β = 0.123).
Discussion: The results of our analyses show that marginalization rates decreased as the number of community health worker vacancies and the GP-related metric (the number of GPs having an agreement with the District Public Health Authority) decreased as well. The marginalization rates increased as vacancies needed for Roma health mediators increased as well. Additional analyses are needed in order to account for the health status of the NUTS2 regions and settlement types in Romania, together with descriptive analyses on the current number of employed community nurses and Roma health mediators. However, our analysis underscores the importance of adequate access to health professionals in preventing marginalization in the Romanian population.