Greater temporal regularity of primary care visits was associated with reduced hospitalizations and mortality, even after controlling for continuity of care

Previous studies have shown that more temporally regular primary care visits are associated with improved patient outcomes. To examine the association of temporal regularity (TR) of primary care with hospitalizations and mortality in patients with chronic illnesses. Also, to identify threshold values for TR for predicting outcomes. Retrospective cohort study. We used data from the electronic health record of a health maintenance organization in Israel to study primary care visits of 70,095 patients age 40 + with one of three chronic conditions (diabetes mellitus, heart failure, chronic obstructive pulmonary disease). We calculated TR for each patient during a two-year period (2016-2017), and divided patients into quintiles based on TR. Outcomes (hospitalization, death) were observed in 2018-2019. Covariates included the Bice-Boxerman continuity of care score, demographics, and comorbidities. We used multivariable logistic regression to examine TR's association with hospitalization and death, controlling for covariates. Compared to patients receiving the most regular care, patients receiving less regular care had increased odds of hospitalization and mortality, with a dose-response curve observed across quintiles (p for linear trend < 0.001). For example, patients with the least regular care had an adjusted odds ratio of 1.40 for all-cause mortality, compared to patients with the most regular care. Analyses stratified by age, sex, ethnic group, area-level SES, and certain comorbid conditions did not show strong differential associations of TR across groups. We found an association between more temporally regular care in antecedent years and reduced hospitalization and mortality of patients with chronic illness in subsequent years, after controlling for covariates. There was no clear threshold value for temporal regularity; rather, more regular primary care appeared to be better across the entire range of the variable.