The research report by Hengartner et al. (1) in this journal claims that “antidepressant use during acute inpatient care is associated with an increased risk of psychiatric rehospitalisation over a 12-month follow-up after discharge”. The authors employed a matched pairs comparison of inpatients who were referred with a mixed range of diagnoses to a psychiatric hospital. Patients were matched via nearest neighbor propensity scores (2) and divided into two groups of N = 45 with respect to whether they had been prescribed an antidepressant or not, i.e. whether they were either antidepressant users or non-users. The matching procedure was based on 14 covariates that assessed sociodemographic items, psychosocial impairments, functioning deficits, and illness severity.
While this general methodological approach in naturalistic clinical studies is widely accepted (3), the conclusions made in this particular paper are not warranted. Here, the covariates employed for the matching procedure are, unfortunately, incomplete with regard to the intervention of interest, namely the prescription or non-prescription of antidepressant medication. In fact, none of the 14 matching parameters addressed presence or absence of any reason(s) for clinically meaningful in- or off-label antidepressant use (4, 5), which would be mandatory for the research question under study. Most importantly, assuming a non-random prescription of antidepressants, the patients in the two groups must have been fundamentally different with respect to relevant clinical parameters (6, 7) which, in turn, would be likely to predict outcome as reflected in rehospitalization rate (8). Obviously, these relevant critical factors were not assessed in the analysis. Incidentally, rehospitilization rate is not actually considered to be a marker of treatment quality and is thus a somewhat questionable outcome measure (9).
The paper’s conclusion is thus the ramification of a hidden selection bias due to omission of information of fundamental relevance. To illustrate the pitfalls of this faulty approach, a similar erroneous conclusion would be, for example, that cytostatic drugs are harmful because subjects who use cyotostatic drugs are more likely to finally die from cancer than subjects not using such drugs, even when the two groups had been matched perfectly for age, gender, and a series of more or less useful health parameters. While this association in itself is numerically true, it is logically wrong to imply the causal relationship that cytostatic drugs produce cancer.
In their paper, Hengartner et al. (1) include an extensive section on study limitations, in which they reason that the methodological flaws do not restrict the main conclusion, as underscored in the paper’s title. However, because the results and main messages are based on a fundamental logical error, and because omitting important covariates in propensity matching is a detrimental failure (10), the key message is not justified by the methodology used. Therefore, the paper is certainly misleading and, furthermore, potentially harmful.
Funding
Funding by University of Zürich.
Statements
Author contributions
ES has drafted and written the commentary, with support by experts in methodology.
Conflict of interest
The author declares that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
References
1
HengartnerMPPassalacquaSAndreaeAHeinsiusTHeppURösslerWet al. Antidepressant use during acute inpatient care is associated with an increased risk of psychiatric rehospitalisation over a 12-month follow-up after discharge. Front In Psychiatry (2019) 79(10):1–9. doi: 10.3389/fpsyt.2019.00079
2
RosenbaumPRRubinDB. Constructing a control group using multivariate matched sampling methods that incorporate the propensity score. Am Stat (1985) 39:33–8. doi: 10.2307/2683903
3
MansourniaMAJewellNPGreenlandS. Case–control matching: effects, misconceptions, and recommendations. Eur J Epidemiol (2018) 33:5–14. doi: 10.1007/s10654-017-0325-0
4
ParkLTZarateCA. Depression in the primary care setting. New Engl J Med (2019) 380:559–68. doi: 10.1056/NEJMcp1712493
5
WongJMotulskyAAbrahamowiczMEgualeTBuckeridgeDLTamblynR. Off-label indications for antidepressants in primary care: descriptive study of prescriptions from an indication based electronic prescribing system. Br Med J (2017) 356:j603. doi: 10.1136/bmj.j603
6
Gomez-LumbrerasAFerrerPBallarínESabatéMVidalXAndrettaMet al. Study of antidepressant use in 5 European settings. could economic, sociodemographic and cultural determinants be related to their use? J Affect Disord (2019) 249:278–85. doi: 10.1016/j.jad.2019.01.039
7
FornsJPottegårdAReindersTPoblador-PlouBMorrosRBrandtLet al. Antidepressant use in Denmark, Germany, Spain, and Sweden between 2009 and 2014: Incidence and comorbidities of antidepressant initiators. J Affect Disord (2019) 249:242–52.
8
OlfsonMMechanicDBoyerCAHansellSWalkupJWeidenPJ. Assessing clinical predictions of early rehospitalization in schizophrenia. J Nervous Ment Dis (1999) 187:721–9.
9
ZhangJCarol HarveyCAndrewC. Factors associated with length of stay and the risk of readmission in an acute psychiatric inpatient facility: a retrospective study. Aust New Z J Psychiatry (2011) 45:578–85.
10
LeeJLittleTD. A practical guide to propensity score analysis for applied clinical research. Behav Res Ther (2017) 98:76–90. doi: 10.1016/j.brat.2017.01.005
Summary
Keywords
depression, antidepressant, prognosis, naturalistic study, nearest neighbor propensity score
Citation
Seifritz E (2020) Commentary: Antidepressant Use During Acute Inpatient Care is Associated With an Increased Risk of Psychiatric Rehospitalisation Over a 12-Month Follow-Up after Discharge. Front. Psychiatry 10:990. doi: 10.3389/fpsyt.2019.00990
Received
04 April 2019
Accepted
12 December 2019
Published
05 February 2020
Volume
10 - 2019
Edited by
Danny Horesh, Bar-Ilan University, Israel
Reviewed by
Thorsten Mikoteit, University of Basel, Switzerland
Updates
Copyright
© 2020 Seifritz.
This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.
*Correspondence: Erich Seifritz, Erich.Seifritz@bli.uzh.ch
This article was submitted to Psychopathology, a section of the journal Frontiers in Psychiatry
Disclaimer
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article or claim that may be made by its manufacturer is not guaranteed or endorsed by the publisher.