Estimating Healthcare Transitions:Integrating Logistic Regression and Markov Models to Predict Mortality from Continuity and Discontinuity of Care

Background: Continuity of care (COC) in general practice has been linked to improvedhealth outcomes. However, maintaining COC becomes increasingly challenging over timedue to varying levels of discontinuity. This study conceptualizes COC as a "medical healthsystem technology" with differing degrees of discontinuity (DCOC) across areas, clinics,and subpopulations. The idea is to model the impact of varying effectiveness of thistechnology on patient mortality.Aim: To model likely transitions from COC to increasing levels of DCOC and relateddeath in patients aged 55 and older in general practice. The intension is to estimate degreesof DCOC over time and enable future applications in resource allocation, care disparities,inequality, and health economics analyses.Method: A Markov model with five states (Continuity, Mild Discontinuity, ModerateDiscontinuity, Severe Discontinuity, Death) was used to simulate progression in continuityand cumulative time spent in each discontinuity state towards death. A 12-year cohort of4,571,041 listed patients above 12 years who stayed alive were applied to measure long rundegrees of (D)COC. Annual transition probabilities between states and the absorbing statedeath were estimated using transition rates, logistic regression analysis and marginsanalysis at the end of the analysis period. The plan is to use sensitivity analyses assessedthe robustness of findings.Results: The model shows transitions across care states based on the estimated annualtransition probabilities. This allows for capturing gradual transition and years in continuityof care, mild discontinuity, moderate discontinuity, severe discontinuity and death. Themodel permits long run cohort simulation for elderly patients. Finally, the model can beused to simulate interventions targeting care coordination reduced transitions to degrees ofdiscontinuity.Conclusion: This type of model offers long run insights into care transitions, enablingevidence-based policy decisions to optimize resource allocation, improve care continuity,and enhance patient outcomes.