Testing the internal validity of the maintenance/discontinuation study design for antidepressant medications in depression using simulations

Testing the internal validity of the maintenance/discontinuation study design for antidepressant medications in depression using simulations | medRxiv /* */ /* */ <!-- <!-- /*! * yepnope1.5.4 * (c) WTFPL, GPLv2 */ (function(a,b,c){function d(a){return"[object Function]"==o.call(a)}function e(a){return"string"==typeof a}function f(){}function g(a){return!a||"loaded"==a||"complete"==a||"uninitialized"==a}function h(){var a=p.shift();q=1,a?a.t?m(function(){("c"==a.t?B.injectCss:B.injectJs)(a.s,0,a.a,a.x,a.e,1)},0):(a(),h()):q=0}function i(a,c,d,e,f,i,j){function k(b){if(!o&&g(l.readyState)&&(u.r=o=1,!q&&h(),l.onload=l.onreadystatechange=null,b)){"img"!=a&&m(function(){t.removeChild(l)},50);for(var d in y[c])y[c].hasOwnProperty(d)&&y[c][d].onload()}}var j=j||B.errorTimeout,l=b.createElement(a),o=0,r=0,u={t:d,s:c,e:f,a:i,x:j};1===y[c]&&(r=1,y[c]=[]),"object"==a?l.data=c:(l.src=c,l.type=a),l.width=l.height="0",l.onerror=l.onload=l.onreadystatechange=function(){k.call(this,r)},p.splice(e,0,u),"img"!=a&&(r||2===y[c]?(t.insertBefore(l,s?null:n),m(k,j)):y[c].push(l))}function j(a,b,c,d,f){return q=0,b=b||"j",e(a)?i("c"==b?v:u,a,b,this.i++,c,d,f):(p.splice(this.i++,0,a),1==p.length&&h()),this}function k(){var a=B;return a.loader={load:j,i:0},a}var l=b.documentElement,m=a.setTimeout,n=b.getElementsByTagName("script")[0],o={}.toString,p=[],q=0,r="MozAppearance"in l.style,s=r&&!!b.createRange().compareNode,t=s?l:n.parentNode,l=a.opera&&"[object Opera]"==o.call(a.opera),l=!!b.attachEvent&&!l,u=r?"object":l?"script":"img",v=l?"script":u,w=Array.isArray||function(a){return"[object Array]"==o.call(a)},x=[],y={},z={timeout:function(a,b){return b.length&&(a.timeout=b[0]),a}},A,B;B=function(a){function b(a){var a=a.split("!"),b=x.length,c=a.pop(),d=a.length,c={url:c,origUrl:c,prefixes:a},e,f,g;for(f=0;f<d;f++)g=a[f].split("="),(e=z[g.shift()])&&(c=e(c,g));for(f=0;f<b;f++)c=x[f](c);return c}function g(a,e,f,g,h){var i=b(a),j=i.autoCallback;i.url.split(".").pop().split("?").shift(),i.bypass||(e&&(e=d(e)?e:e[a]||e[g]||e[a.split("/").pop().split("?")[0]]),i.instead?i.instead(a,e,f,g,h):(y[i.url]?i.noexec=!0:y[i.url]=1,f.load(i.url,i.forceCSS||!i.forceJS&&"css"==i.url.split(".").pop().split("?").shift()?"c":c,i.noexec,i.attrs,i.timeout),(d(e)||d(j))&&f.load(function(){k(),e&&e(i.origUrl,h,g),j&&j(i.origUrl,h,g),y[i.url]=2})))}function h(a,b){function c(a,c){if(a){if(e(a))c||(j=function(){var a=[].slice.call(arguments);k.apply(this,a),l()}),g(a,j,b,0,h);else if(Object(a)===a)for(n in m=function(){var b=0,c;for(c in a)a.hasOwnProperty(c)&&b++;return b}(),a)a.hasOwnProperty(n)&&(!c&&!--m&&(d(j)?j=function(){var a=[].slice.call(arguments);k.apply(this,a),l()}:j[n]=function(a){return function(){var b=[].slice.call(arguments);a&&a.apply(this,b),l()}}(k[n])),g(a[n],j,b,n,h))}else!c&&l()}var h=!!a.test,i=a.load||a.both,j=a.callback||f,k=j,l=a.complete||f,m,n;c(h?a.yep:a.nope,!!i),i&&c(i)}var i,j,l=this.yepnope.loader;if(e(a))g(a,0,l,0);else if(w(a))for(i=0;i (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0];var j=d.createElement(s);var dl=l!='dataLayer'?'&l='+l:'';j.src='//www.googletagmanager.com/gtm.js?id='+i+dl;j.type='text/javascript';j.async=true;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-P4HH5NV'); Skip to main content Home About Submit ALERTS / RSS Search for this keyword Advanced Search Testing the internal validity of the maintenance/discontinuation study design for antidepressant medications in depression using simulations William U. Meyerson , Tianxi Cai , Jordan W. Smoller doi: https://doi.org/10.1101/2025.10.02.25337182 William U. Meyerson 1 Center for Precision Psychiatry, Massachusetts General Hospital , Boston, MA, USA 2 Department of Psychiatry, Massachusetts General Hospital / Harvard Medical School , Boston, MA, USA MD, PhD Find this author on Google Scholar Find this author on PubMed Search for this author on this site For correspondence: william.ulysses{at}gmail.com Tianxi Cai 3 Department of Biostatistics, Harvard T.H. Chan School of Public Health , Boston, MA, USA 4 Department of Biomedical Informatics, Harvard Medical School , Boston, MA, USA ScD Find this author on Google Scholar Find this author on PubMed Search for this author on this site Jordan W. Smoller 1 Center for Precision Psychiatry, Massachusetts General Hospital , Boston, MA, USA 2 Department of Psychiatry, Massachusetts General Hospital / Harvard Medical School , Boston, MA, USA 5 Psychiatric and Neurodevelopmental Genetics Unit, Center for Genomic Medicine, Massachusetts General Hospital , Boston, MA, USA MD, ScD Find this author on Google Scholar Find this author on PubMed Search for this author on this site Abstract Full Text Info/History Metrics Data/Code Preview PDF Abstract Background Evidence for longer-term use of antidepressant medications (ADM) in major depressive disorder (MDD) derives from maintenance/discontinuation (M/D) trials showing that approximately 20% of patients who remitted on ADMs relapse by 1 year when treatment is maintained vs. approximately 40% if ADMs are discontinued. This treatment effect is larger than that demonstrated in acute-phase trials and has been criticized as “too good to be true.” The enrichment design of M/D trials has been proposed as one explanation for this larger effect size, but it is unclear whether enrichment design alone is a sufficient explanation, raising questions about the internal validity of M/D trials. Objective To test whether the enrichment design of M/D trials is sufficient to account for the larger-than-expected effect of maintenance treatment. Methods We simulated M/D trials by applying the study characteristics including the enrichment phase of 9 real M/D trials to depression trajectories derived from the STAR-D pragmatic trial and depression efficacy data derived from acute-phase trial and compared the resulting relapse rates by arm between the simulated data and real data. Results The simulated ADM average treatment effect increased 1.7-fold after selection on remitters from the enrichment phase from 1.75 Hamilton Depression Rating Scale points in unselected participants to 2.91 HAM-D points in just the participants who pass the enrichment filter from phase 1 to phase 2. Simulated relapse rates and the relative risks between arms had excellent fit on the withheld post-randomization aggregated real trial data. Conclusions Our simulations indicate that the enrichment design of M/D trials is sufficient to explain their larger effect sizes. These results support the internal validity of M/D studies in characterizing the benefits of ADM maintenance treatment. Introduction An estimated 8.8 million Americans have been taking antidepressant medications (ADMs) continuously for 10 or more years 1 . Clinical guidelines recommend that ADMs be continued for at least 6-12 months after remission of a major depressive episode (MDD) to prevent relapse, and for longer periods in individuals with 3 or more lifetime episodes 2 , 3 . The evidence base for this practice does not come from classical acute phase randomized controlled trials (RCTs), because in MDD these are rarely longer than 3 months 1 , 4 . Instead, the recommendations derive from maintenance/discontinuation trials (M/D trials), in which participants are treated with ADM in an open-label fashion in phase 1, after which responders are randomized in phase 2 to continued ADM vs placebo substitution. Meta-analyses of M/D trials report that relapse rates are 37%-45% after 6-12 months in discontinuation arms but only 18%-24% in the continuation arms 5 – 9 . When appropriate for the condition and treatment and appropriately conducted, M/D trials have been praised as an ethical and efficient alternative to classical acute-phase randomized controlled trials, especially when longer follow-up times are desired 10 . However, there is a lack of consensus about whether the M/D design is internally valid in the specific case of ADMs for MDD 11 – 13 . Critics have noted the puzzling discrepancy between the typically modest effect sizes of ADMs in acute phase RCTs (a mere 1.8 points on the 17-item Hamilton Depression Rating Scale per one authoritative meta-analysis) 14 and the frequently larger benefit of ADM treatment in M/D trials. A leading explanation for this discrepancy points to selection effects. Because only phase 1 responders enter phase 2, the latter is enriched for patients who can tolerate ADMs and who receive benefit from them. Because of this enrichment, it would be inappropriate to directly extrapolate the results of M/D trials to patients whose acute response was not sufficient to make them eligible for maintenance treatment Nevertheless, provided the trials are internally valid for the selected cohort, their conclusions remain applicable to clinically similar patients, who comprise a substantial share of real-world practice. While there are good reasons to believe that selection effects operate in M/D trials, it is less clear whether these effects are strong enough to fully explain the large difference in outcomes between arms in M/D trials. For example, because the natural history of depression is highly variable and responsive to placebo effects, it is unclear just how successfully the M/D design enriches for treatment responders. There are other possible explanations for the effect sizes of M/D trials (e.g. withdrawal-relapse confusion 15 , discontinuation-enhanced unblinding 16 , 17 , rebound/”payback” 18 ) which challenge the internal validity of M/D trials. So long as it remains unknown whether selection effects are sufficient to explain the results of M/D trials, the validity of practice guidelines derived from them may be questioned. To address this gap, we performed simulations to examine whether selection effects are sufficient to explain the between-arm differences in relapse rates seen in M/D trials. The simulations are run with parameters sourced from the pre-randomization phase of recent M/D trials, together with information about average treatment effect and estimated distribution of treatment effect from acute antidepressant RCTs. These simulations test whether the results of M/D trials are consistent with what is known about the magnitude of ADM benefits from acute-phase RCTs; the natural history of MDD from longer term, non-placebo-controlled studies; and the unique aspects of the study design of M/D trials. As noted, these simulations have implications for the internal validity of M/D trials of ADMs in MDD and for clinical guidelines about maintenance treatment. Methods Overview of model The simulation model consists of 2 phases and 4 effects. Phase 1 is the single-armed open-label treatment phase. Phase 2 is a two-arm maintenance/discontinuation phase. The four effects driving the potential outcome under the simulation models are 1) the natural history of treated depression as a base layer in all phases and arms, 2) trial-specific placebo effects due to “contextual factors” 19 as an additional layer in all phases and arms, 3) removal of active drug effects in the discontinuation arm of Phase 2, and 4) the loss of open-label certainty once patients enter the randomized phase. In each run of the simulation, a simulated initial cohort meeting eligibility criteria experience a change in depression severity during open-label treatment (phase 1) related to the natural history of treated depression and empirical trial-specific random-effects. Participants who remit then enter phase 2, where some proportion are randomly assigned to continue antidepressant and the rest are switched to placebo. At that point, participants in the discontinuation arm lose whatever benefit they experienced from active drug according to a decay curve, and participants in both arms lose a portion of placebo expectancy effects. Relapse survival curves by arm in phase 2 are calculated according to a relapse rule ( Table 1 ) and compared with relapse rates from real M/D trials. Results were generated across 100 runs. View this table: View inline View popup Download powerpoint Table 1: Characteristics of included studies Identification of Maintenance/Discontinuation Trials to Simulate M/D studies were included in our trial pool if they met the following criteria: 1) they were placebo-controlled M/D studies of antidepressants for MDD included in the most recent systematic review on the topic 6 ; 2) no treatment arm included psychotherapy as a study intervention; and 3) a report of the study was published within 15 years of manuscript preparation (i.e. since 2010). Nine RCTs 20 – 27 met these inclusion criteria with selected characteristics listed in Table 1 (see Table S1 for more detailed characteristics). Model Parameters Model parameters are listed in Table 2 . The number of participants in phase 1, target number of participants in phase 2, duration of phase 1 and 2 in months, the minimum depression score for eligibility, remission threshold, relapse threshold, minimum duration of current episode, maximum duration of current episode, and proportion of participants assigned to continue drug during phase 2 were calculated as the mean value of these variables from the nine included trials, and rounded to the nearest integer where appropriate. The lifetime number of depressive episodes, and maximum age for study inclusion were obtained from the median of these values among the included studies. Most included studies used a compound remission rule, where participants must also meet some less strict cutoff at an intermediate time-point during phase1; on average this cutoff was less strict than the end-phase-1 remission threshold by 2 points. To account for this, we implemented a less-strict cutoff at month 2 in the simulations, which participants had to pass before reaching the more-strict cutoff at month 3. View this table: View inline View popup Table 2: Trial-Averaged Parameter Set Used for Simulation The average drug treatment effect was obtained from a recent meta-analysis of acute-phase RCTs 14 . The supplement describes how we used other data from this source to characterize the distribution of this effect among participants. In the absence of additional information, we assumed that the benefits of active drug wash out after discontinuation according to a decay curve that is the mirror image of the onset curve of antidepressant efficacy, obtained from a meta-analysis of the time-course of onset of ADM effects 29 . The magnitude of the expectancy effect as a function of the proportion of participants assigned to drug was obtained from a meta-analysis of antidepressant RCTs with differing randomization ratios 30 . The contextual factors effect was a constant value added to the depression severity scores of all participants at all follow-up time-points, that was chosen through a parameter grid search to result in the number of phase 1 participants meeting remission to be as similar as possible to the number in the included trials. Natural History of Treated Depression Sequenced Treatment Alternatives to Relieve Depression (STAR-D) was the largest long-term publicly funded trial of antidepressants in MDD and is the field standard for characterizing the natural history of treated depression 31 . Our simulations made use of the depression trajectories of STAR-D participants in months 0 through 12 as a base layer to capture realistic month-to-month fluctuations of depression severity and measurement noise, with other features of the simulation layered on top of this base layer. To ensure simulations represented similar patients as the RCTs, we excluded STAR-D participants with a comorbid psychiatric disorder to better reflect the M/D RCT data. Likewise we only included STAR-D participants who met the composite M/D trial thresholds for To ensure sufficient sample size for matching and to complete missing data, we augmented the STAR-D data with 100 synthetically generated complete cohort replicates designed to preserve relationships among variables. This was achieved through a Bayesian hierarchical local mixture model with local dependence 32 implemented via the multipleSyn function from synMicrodata 33 R package. (v2.1.0), with a burn-in of 100, an interval of 50 between synthesis steps, and otherwise default parameters. As a post-processing step, depression severities in months 7-12 -months with high (69.0%-80.4%) missingness in the original data-were normalized to have a stable mean depression severity concordant with lower-missingness (62.1%) month 6. The synthetic data had high fidelity to the original data. For example, the difference in depression severity between months 0 and 1 had a mean (SD) of 8.3 (7.5) in the original data and 8.3 (9.0) in the synthetic data, and the correlation between the number of prior episodes and depression severity in month 6 was 0.085 in the original data and 0.064 in the synthetic data. Model Scoring We evaluated model fit by comparing simulated outcomes to withheld post-randomization results averaged across the nine included trials. At the end of each post-randomization month (months 1–6), we assessed three outcomes: (1) cumulative relapse frequency in the continuation arm; (2) cumulative relapse frequency in the discontinuation arm; and (3) the relapse relative risk (continuation vs. discontinuation). Among the nine included trials, the mean and standard deviation of these 3 outcome types were calculated at each of these 6 time points, after extracting study-specific relapse rates via plot digitization ( graphreader.com ) (Supplemental Table S3). One study (Shiovitz 2014) lacked data for the 6-month mark, which was imputed using the mice R package (v3.17.0). The goal of this project was to develop a general sense of whether the results of M/D trials are broadly consistent with what we know about MDD and antidepressants from short-term acute phase RCTs, non-placebo-controlled longer studies, and the study design of M/D trials. To support this goal, we used the following pragmatic scoring system. For each outcome and time point, we computed a Z-score as the difference between the simulation mean and the mean across the nine included studies, divided by the standard deviation of that study mean. The mean absolute value of Z-scores across all outcomes and time-points was taken as a summary score. This summary score was interpreted according to rubric in Table 3 . View this table: View inline View popup Download powerpoint Table 3. Interpretation rubric for summary score fit Results Main Model Efficiency of enrichment The mean HAM-D difference between the maintenance and discontinuation arms during post-randomization months 3–6 was 2.91 points, approximately 1.7 times the average treatment effect from acute RCTs used to parameterize Phase 1. This amplification is consistent with enrichment arising from restricting Phase 2 to Phase-1 responders. Relative risk of relapse The relative risk of relapse in the continuation vs discontinuation arm was 0.52 at 3 months and 0.56 at 6 months in the simulations (Figure 2; Table 4 ). This was similar to the corresponding values from the withheld post-randomization phase of the six positive trials (Boulenger 2012), (Durgam 2019), (Goodwin 2013 Study 2), (Rickels 2010), (Rosenthal 2013), and (Thase 2022) but a greater reduction than the negative and/or weak-effects trials of (Durgam 2018), (Goodwin 2013 Study 1), and (Shiovitz 2014). Altogether, the relative risk of relapse in the continuation vs discontinuation arm in the simulations represented a greater risk reduction than the nine-trial-average, but because of the spread between the positive trials and negative trials, this discrepancy was less than one standard deviation in magnitude. The relative risks in our simulations were even closer to the relative risks reported in meta-analyses of a larger number of M/D studies; e.g. (Kishi 2023) reports a relative risk of relapse of 0.49 (0.43 to 0.57) at 3 months and 0.51 (0.46 to 0.58) at 6 months 34 . View this table: View inline View popup Table 4: Comparison of relapse rates by arm in simulated and observed trials. Continuation arm relapse rates The absolute rates of relapse in the continuation arm were generally concordant with the relapse rates among the trials. This was particularly true at the 3-month mark, where the 11.9% cumulative relapse rate in the simulations was very similar to the nine-trial-average of 12.7%. The simulated continuation arm relapse rates somewhat undershot the nine-trial-average in months 1 and 2, and overshot them in month 4-6, but the only such month that departed from the nine-trial-average by more than 1 SD was month 1, with a relapse rate in the simulations of 3.3% vs a nine-trial-average of 5.2%; in comparison, (Thase 2022) had a 3.4% relapse rate in the drug arm at month 1. Discontinuation arm relapse rates The absolute rate of relapse in the discontinuation arm were also generally concordant with the relapse rates among the trials. In the simulations, cumulative relapse rates ranged from 7.2% at month 1 to 43.9% at month 6; which paralleled a nine-trial-average that ranged from 8.6% at month 1 to 31.5% at month 6, although tending to overshoot the nine-trial-average in the later months. Model scoring Across the 3 outcomes and 6 time points, the mean departure of the simulation from the nine-trial-average was 0.49 standard deviations, representing excellent fit according to the interpretation schema of Table 3 . Of the 18 outcome-time pairs, 12 met criteria for excellent fit, 6 met criteria for good fit, and none met criteria for fair or poor fit. Download figure Open in new tab Figure 1: Simulated versus observed relapse rates by arm in simulations and real RCTs. Discussion Maintenance/discontinuation trials form the bedrock of evidence for the common practice of continuing antidepressant medications, but it is unclear whether their large effect sizes are consistent with the modest effects of ADMs in acute-phase RCTs. Our simulations demonstrate that typical relapse curves of continuation and discontinuation arms on ADM M/D trials are a straightforward extension of what we already know from acute phase studies, the natural history of depression, and the design of M/D trials. Our simulations had excellent fit to real trial data for all three outcome types: maintenance arm relapse rates, discontinuation relapse rates, and relative risk of relapse between arms. The large risk reduction in the maintenance arm is expected to be a consequence of the 1.7-fold enrichment of the average treatment effect that occurred after selection of remitters during phase 1 of the simulation. These results do not imply that withdrawal-relapse confusion, unblinding, and rebound do not occur in M/D trials, only that these phenomena are not necessary to explain the basic results of the ADM M/D trials. Interestingly, the relative risk of relapse in the simulations hewed even more closely to meta-analyses of a larger number of studies than to the sampling of nine recent studies on which the simulations were based 34 . Some possible explanations for this finding include heterogeneity in patient population at the time of treatment initiation or difference in the details of study procedure or execution contributing to high “random effects” idiosyncratic variance between individual studies that was not captured in our simulations, but that they may be triangulating a more general truth consistent with what is known from ADM acute phase RCTs. One limitation of this work is the inclusion of only M/D trials from the last 15 years. Moreover, the STAR-D data used to model natural history carries a number of limitations including the fact that participants’ treatment was switched, although we believe the impact of this is contained by the fact that no treatments in STAR-D were found to be superior to any other 35 . The simulations were not precise enough to incorporate nuances of remission and relapse rules such as stability between consecutive weeks. Some model parameters carry high uncertainty such as in how antidepressant effects are distributed, with different authors proposing different degrees of heterogeneity 14 , 36 Nonetheless, to the extent that these simulations capture the essential features of ADMs and MDD, these results increase our confidence that M/D trials are internally valid and therefore increases the rigor of clinical guidelines that draw on these trials when discussing maintenance treatment. Data Availability No new data were generated for this study. The STAR*D data can be obtained from the NIMH Data Archive, with more information in the Acknowledgments section. The RCT summary data are publicly available from the cited trials. Funding and Disclosure Funding/Support Dr. William Meyerson was supported by National Library of Medicine/NIH grant T15LM007092. Role of the Funder The National Library of Medicine did not participate in the design of the study, the analysis and interpretation of the data, or the preparation and submission of the manuscript for publication. Conflicts of Interest Disclosures Smoller is a member of the Scientific Advisory Board of Sensorium Therapeutics (with options), and has received grant support from Biogen, Inc. Author Contributions Dr Meyerson had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Concept and design: Meyerson Acquisition, analysis, or interpretation of data: Meyerson Drafting of the manuscript: Meyerson Critical review of the manuscript for important intellectual content: All authors. Statistical analysis: Meyerson Obtained funding: Meyerson, Smoller Administrative, technical, or material support: Smoller Supervision: Cai, Smoller Acknowledgment Data and/or research tools used in the preparation of this manuscript were obtained from the National Institute of Mental Health (NIMH) Data Archive (NDA). NDA is a collaborative informatics system created by the National Institutes of Health to provide a national resource to support and accelerate research in mental health. Dataset identifier(s): 10.15154/xx52-jp36 and 10.15154/zk9m-5n40. This manuscript reflects the views of the authors and may not reflect the opinions or views of the NIH or of the Submitters submitting original data to NDA. References 1. ↵ Ward , W. , Haslam , A. & Prasad , V. Antidepressant Trial Duration Versus Duration of Real-World Use: A Systematic Analysis . Am J Med S0002-9343(25)00286–4 ( 2025 ) doi: 10.1016/j.amjmed.2025.04.037 . OpenUrl CrossRef 2. ↵ Gelenberg , A. J. et al. American Psychiatric Association Practice Guideline for the Treatment of Patients With Major Depressive Disorder, Third Edition . Am J Psychiatry 167 (suppl), 1 – 152 ( 2010 ). OpenUrl CrossRef PubMed Web of Science 3. ↵ Kennedy , S. H. et al. Canadian Network for Mood and Anxiety Treatments (CANMAT) 2016 Clinical Guidelines for the Management of Adults with Major Depressive Disorder: Section 3. Pharmacological Treatments . Can J Psychiatry 61 , 540 – 560 ( 2016 ). OpenUrl CrossRef PubMed 4. ↵ Cipriani , A. et al. Comparative efficacy and acceptability of 21 antidepressant drugs for the acute treatment of adults with major depressive disorder: a systematic review and network meta-analysis . The Lancet 391 , 1357 – 1366 ( 2018 ). OpenUrl 5. ↵ Borges , S. et al. Review of Maintenance Trials for Major Depressive Disorder: A 25-Year Perspective From the US Food and Drug Administration . Psychiatrist.com https://www.psychiatrist.com/jcp/review-maintenance-trials-major-depressive-disorder/ . 6. ↵ Hu , Y. et al. Association between duration of antidepressant treatment for major depressive disorder and relapse rate after discontinuation: A meta-analysis . Psychiatry Research 337 , 115926 ( 2024 ). OpenUrl PubMed 7. Kishi , T. et al. Relapse and its modifiers in major depressive disorder after antidepressant discontinuation: meta-analysis and meta-regression . Mol Psychiatry 28 , 974 – 976 ( 2023 ). OpenUrl PubMed 8. Zhou , D. et al. Effects of antidepressant medicines on preventing relapse of unipolar depression: a pooled analysis of parametric survival curves . Psychological Medicine 52 , 48 – 56 ( 2022 ). OpenUrl PubMed 9. ↵ Kato , M. et al. Discontinuation of antidepressants after remission with antidepressant medication in major depressive disorder: a systematic review and meta-analysis . Mol Psychiatry 26 , 118 – 133 ( 2021 ). OpenUrl PubMed 10. ↵ Kopec , J. A. , Abrahamowicz , M. & Esdaile , J. M. Randomized discontinuation trials: Utility and efficiency . Journal of Clinical Epidemiology 46 , 959 – 971 ( 1993 ). OpenUrl CrossRef PubMed Web of Science 11. ↵ Ghaemi , S. N. & Selker , H. P. Maintenance efficacy designs in psychiatry: Randomized discontinuation trials – enriched but not better . Journal of Clinical and Translational Science 1 , 198 – 204 ( 2017 ). OpenUrl 12. Hengartner , M. P. How effective are antidepressants for depression over the long term? A critical review of relapse prevention trials and the issue of withdrawal confounding . Therapeutic Advances in Psychopharmacology 10 , 2045125320921694 ( 2020 ). OpenUrl CrossRef PubMed 13. ↵ Greenhouse , J. B. , Stangl , D. , Kupfer , D. J. & Prien , R. F. Methodologic issues in maintenance therapy clinical trials . Arch Gen Psychiatry 48 , 313 – 318 ( 1991 ). OpenUrl CrossRef PubMed Web of Science 14. ↵ Stone , M. B. et al. Response to acute monotherapy for major depressive disorder in randomized, placebo controlled trials submitted to the US Food and Drug Administration: individual participant data analysis . BMJ 378 , e067606 ( 2022 ). OpenUrl Abstract / FREE Full Text 15. ↵ Baldwin , D. S. , Montgomery , S. A. , Nil , R. & Lader , M. Discontinuation symptoms in depression and anxiety disorders . International Journal of Neuropsychopharmacology 10 , 73 – 84 ( 2007 ). OpenUrl CrossRef PubMed 16. ↵ Lin , Y.-H. et al. Assessment of blinding in randomized controlled trials of antidepressants for depressive disorders 2000–2020: A systematic review and meta-analysis . eClinicalMedicine 50 , ( 2022 ). 17. ↵ Scott , A. J. , Sharpe , L. & Colagiuri , B. A systematic review and meta-analysis of the success of blinding in antidepressant RCTs . Psychiatry Research 307 , 114297 ( 2022 ). OpenUrl CrossRef PubMed 18. ↵ Henssler , J. , Heinz , A. , Brandt , L. & Bschor , T. Antidepressant Withdrawal and Rebound Phenomena . Dtsch Arztebl Int 116 , 355 – 361 ( 2019 ). OpenUrl CrossRef PubMed 19. ↵ Hafliðadóttir , S. H. et al. Placebo response and effect in randomized clinical trials: meta-research with focus on contextual effects . Trials 22 , 493 ( 2021 ). OpenUrl CrossRef PubMed 20. ↵ Thase , M. E. et al. Vortioxetine 5, 10, and 20 mg significantly reduces the risk of relapse compared with placebo in patients with remitted major depressive disorder: The RESET study . J Affect Disord 303 , 123 – 130 ( 2022 ). OpenUrl PubMed 21. Durgam , S. , Chen , C. , Migliore , R. , Prakash , C. & Thase , M. E. Relapse prevention with levomilnacipran ER in adults with major depressive disorder: A multicenter, randomized, double-blind, placebo-controlled study . Depress Anxiety 36 , 225 – 234 ( 2019 ). OpenUrl PubMed 22. Durgam , S. et al. Relapse prevention in adults with major depressive disorder treated with vilazodone: a randomized, double-blind, placebo-controlled trial . Int Clin Psychopharmacol 33 , 304 – 311 ( 2018 ). OpenUrl PubMed 23. Shiovitz , T. , Greenberg , W. M. , Chen , C. , Forero , G. & Gommoll , C. P. A Randomized, Double-blind, Placebo-controlled Trial of the Efficacy and Safety of Levomilnacipran ER 40-120mg/day for Prevention of Relapse in Patients with Major Depressive Disorder . Innov Clin Neurosci 11 , 10 – 22 ( 2014 ). OpenUrl 24. Rosenthal , J. Z. , Boyer , P. , Vialet , C. , Hwang , E. & Tourian , K. A. Efficacy and safety of desvenlafaxine 50 mg/d for prevention of relapse in major depressive disorder:a randomized controlled trial . J Clin Psychiatry 74 , 158 – 166 ( 2013 ). OpenUrl CrossRef PubMed 25. Goodwin , G. M. , Boyer , P. , Emsley , R. , Rouillon , F. & de Bodinat , C. Is it time to shift to better characterization of patients in trials assessing novel antidepressants? An example of two relapse prevention studies with agomelatine . Int Clin Psychopharmacol 28 , 20 – 28 ( 2013 ). OpenUrl CrossRef PubMed 26. Boulenger , J.-P. , Loft , H. & Florea , I. A randomized clinical study of Lu AA21004 in the prevention of relapse in patients with major depressive disorder . J Psychopharmacol 26 , 1408 – 1416 ( 2012 ). OpenUrl CrossRef PubMed 27. ↵ Rickels , K. et al. Desvenlafaxine for the prevention of relapse in major depressive disorder: results of a randomized trial . J Clin Psychopharmacol 30 , 18 – 24 ( 2010 ). OpenUrl CrossRef PubMed 28. Goodwin , G. M. , Emsley , R. , Rembry , S. , Rouillon , F. , & Agomelatine Study Group . Agomelatine prevents relapse in patients with major depressive disorder without evidence of a discontinuation syndrome: a 24-week randomized, double-blind, placebo-controlled trial . J Clin Psychiatry 70 , 1128 – 1137 ( 2009 ). OpenUrl CrossRef PubMed Web of Science 29. ↵ Henssler , J. , Kurschus , M. , Franklin , J. , Bschor , T. & Baethge , C. Trajectories of Acute Antidepressant Efficacy: How Long to Wait for Response? A Systematic Review and Meta-Analysis of Long-Term, Placebo-Controlled Acute Treatment Trials . J. Clin. Psychiatry 79 , ( 2018 ). 30. ↵ Salanti , G. et al. Impact of placebo arms on outcomes in antidepressant trials: systematic review and meta-regression analysis . Int J Epidemiol 47 , 1454 – 1464 ( 2018 ). OpenUrl CrossRef PubMed 31. ↵ Rush , A. J. et al. Sequenced treatment alternatives to relieve depression (STAR*D): rationale and design . Controlled Clinical Trials 25 , 119 – 142 ( 2004 ). OpenUrl CrossRef PubMed Web of Science 32. ↵ Murray , J. S. & and Reiter , J. P. Multiple Imputation of Missing Categorical and Continuous Values via Bayesian Mixture Models With Local Dependence . Journal of the American Statistical Association 111 , 1466 – 1479 ( 2016 ). OpenUrl 33. ↵ Kim , H. J. , Lee , J. Kim , Y.-M. & Murray , J. synMicrodata: Synthetic Microdata Generator . ( 2024 ). 34. ↵ Kishi , T. et al. Relapse and its modifiers in major depressive disorder after antidepressant discontinuation: meta-analysis and meta-regression . Mol Psychiatry 28 , 974 – 976 ( 2023 ). OpenUrl PubMed 35. ↵ Warden , D. , Rush , A. J. , Trivedi , M. H. , Fava , M. & Wisniewski , S. R. The STAR*D project results: A comprehensive review of findings . Curr Psychiatry Rep 9 , 449 – 459 ( 2007 ). OpenUrl CrossRef PubMed 36. ↵ Meyerson , W. U. , Pieper , C. F. & Hoyle , R. H. Use of Quantile Treatment Effects Analysis to Describe Antidepressant Response in Randomized Clinical Trials Submitted to the US Food and Drug Administration: A Secondary Analysis of Pooled Trial Data . JAMA Network Open 6 , e2317714 ( 2023 ). OpenUrl View the discussion thread. Back to top Previous Next Posted October 05, 2025. Download PDF Data/Code Email Thank you for your interest in spreading the word about medRxiv. NOTE: Your email address is requested solely to identify you as the sender of this article. Your Email * Your Name * Send To * Enter multiple addresses on separate lines or separate them with commas. You are going to email the following Testing the internal validity of the maintenance/discontinuation study design for antidepressant medications in depression using simulations Message Subject (Your Name) has forwarded a page to you from medRxiv Message Body (Your Name) thought you would like to see this page from the medRxiv website. Your Personal Message CAPTCHA This question is for testing whether or not you are a human visitor and to prevent automated spam submissions. Share Testing the internal validity of the maintenance/discontinuation study design for antidepressant medications in depression using simulations William U. Meyerson , Tianxi Cai , Jordan W. Smoller medRxiv 2025.10.02.25337182; doi: https://doi.org/10.1101/2025.10.02.25337182 Share This Article: Copy Citation Tools Testing the internal validity of the maintenance/discontinuation study design for antidepressant medications in depression using simulations William U. Meyerson , Tianxi Cai , Jordan W. Smoller medRxiv 2025.10.02.25337182; doi: https://doi.org/10.1101/2025.10.02.25337182 Citation Manager Formats BibTeX Bookends EasyBib EndNote (tagged) EndNote 8 (xml) Medlars Mendeley Papers RefWorks Tagged Ref Manager RIS Zotero Tweet Widget Facebook Like Google Plus One Subject Area Psychiatry and Clinical Psychology Subject Areas All Articles Addiction Medicine (568) Allergy and Immunology (863) Anesthesia (300) Cardiovascular Medicine (4435) Dentistry and Oral Medicine (444) Dermatology (382) Emergency Medicine (608) Endocrinology (including Diabetes Mellitus and Metabolic Disease) (1509) Epidemiology (15229) Forensic Medicine (30) Gastroenterology (1124) Genetic and Genomic Medicine (6600) Geriatric Medicine (668) Health Economics (997) Health Informatics (4536) Health Policy (1368) Health Systems and Quality Improvement (1613) Hematology (541) HIV/AIDS (1264) Infectious Diseases (except HIV/AIDS) (15916) Intensive Care and Critical Care Medicine (1103) Medical Education (623) Medical Ethics (146) Nephrology (667) Neurology (6599) Nursing (346) Nutrition (998) Obstetrics and Gynecology (1144) Occupational and Environmental Health (957) Oncology (3332) Ophthalmology (974) Orthopedics (369) Otolaryngology (420) Pain Medicine (436) Palliative Medicine (130) Pathology (663) Pediatrics (1693) Pharmacology and Therapeutics (691) Primary Care Research (711) Psychiatry and Clinical Psychology (5447) Public and Global Health (9232) Radiology and Imaging (2198) Rehabilitation Medicine and Physical Therapy (1370) Respiratory Medicine (1196) Rheumatology (593) Sexual and Reproductive Health (712) Sports Medicine (530) Surgery (712) Toxicology (99) Transplantation (289) Urology (265) (function(){function c(){var b=a.contentDocument||a.contentWindow.document;if(b){var d=b.createElement('script');d.innerHTML="window.__CF$cv$params={r:'a00be9d55fa4df88',t:'MTc3OTYyMTY1MA=='};var a=document.createElement('script');a.src='/cdn-cgi/challenge-platform/scripts/jsd/main.js';document.getElementsByTagName('head')[0].appendChild(a);";b.getElementsByTagName('head')[0].appendChild(d)}}if(document.body){var a=document.createElement('iframe');a.height=1;a.width=1;a.style.position='absolute';a.style.top=0;a.style.left=0;a.style.border='none';a.style.visibility='hidden';document.body.appendChild(a);if('loading'!==document.readyState)c();else if(window.addEventListener)document.addEventListener('DOMContentLoaded',c);else{var e=document.onreadystatechange||function(){};document.onreadystatechange=function(b){e(b);'loading'!==document.readyState&&(document.onreadystatechange=e,c())}}}})();