This study used an individual-level health-state transition decision-analytic model to quantify tradeoffs between continuous antidepressant maintenance and discontinuation with active monitoring after remission from major depressive disorder (MDD). The model aimed to inform a preference-sensitive outpatient decision following completion of an 8-month continuation phase after remission. Calibration was performed to randomized maintenance-discontinuation trials and to a longitudinal cohort of first depressive episodes. The analytic time horizon was 5 years.
The modeled population comprised adults in remission from MDD, represented across four clinically anchored relapse-risk strata. These strata ranged from very low relapse risk (for example, a single mild first episode) up to high relapse risk (for example, highly recurrent depression). The clinical context simulated was an outpatient encounter where a decision must be made whether to continue antidepressant treatment indefinitely or to discontinue and implement active monitoring with restart of medication if relapse is detected. The model used a 5-year follow-up period to capture medium-term outcomes of these strategies.
Model parameters and transition probabilities were calibrated to published randomized maintenance-discontinuation trials and to data from a longitudinal first-episode cohort, as reported in the source. The model stratified patients into four relapse-risk groups to reflect clinically relevant heterogeneity in baseline risk of recurrence. The calibrated model then projected illness and medication trajectories under two strategies: continuous maintenance versus discontinuation with active monitoring and treatment restart after relapse detection.
Key outcomes reported by the model were:
These outcomes were used to assess the clinical tradeoff between reducing depression burden and increasing long-term medication exposure.
Across relapse-risk strata, continuous maintenance consistently reduced projected depression burden relative to discontinuation with active monitoring, but at the cost of greater medication exposure. The model reported the medication-years per depression-month averted when choosing maintenance versus active monitoring, with efficiency varying by relapse risk. The range across strata was from a median of 11.8 medication-years per depression-month averted (95% uncertainty interval [UI], 7.8–19.6) in the very low-risk group to 1.5 medication-years per depression-month averted (95% UI, 0.8–3.0) in the high-risk group. These estimates quantify how many additional years on medication are required, on average, to prevent one month of depression for patients in each risk stratum.
The authors evaluated strategy preference across hypothetical patient thresholds that express willingness to accept medication exposure to avoid depression time. At a preference threshold of 3 medication-years per depression-month averted, continuous maintenance was the preferred strategy for moderate- and high-risk patients. At a threshold of 2 medication-years per depression-month averted, maintenance was preferred only for high-risk patients. At a threshold of 1 medication-year per depression-month averted, no risk group favored continuous maintenance over discontinuation with active monitoring. These findings illustrate how individual values about medication burden versus relapse avoidance alter which strategy is favored.
The decision-analytic results indicate that the relative value of continuous antidepressant maintenance versus stopping with active monitoring is strongly influenced by baseline relapse risk and by patient preferences regarding long-term medication exposure. For patients at higher relapse risk, maintenance provides a more favorable tradeoff in terms of medication-years required to avert depression-months. For patients at low relapse risk, the medication burden to prevent relatively small amounts of depression time is substantial. The authors present this model as a quantitative framework to support individualized, preference-sensitive shared decision-making about maintenance antidepressant treatment after remission from MDD.
This work is reported as a preprint on medRxiv and has not undergone peer review. The source declares competing interests for one author and states funding support. The model used aggregate estimates from published literature for calibration; no individual-level participant data were analyzed. Details beyond those reported in the preprint, including model code or additional sensitivity analyses, were not described in the source text provided.