Combining Pharmacological and Nonpharmacological Interventions in Network Meta-analysis in Psychiatry | Psychiatry | JAMA Psychiatry | JAMA Network
[Skip to Content]
Sign In
Individual Sign In
Create an Account
Institutional Sign In
OpenAthens Shibboleth
[Skip to Content Landing]
1.
Mavridis  D, Giannatsi  M, Cipriani  A, Salanti  G.  A primer on network meta-analysis with emphasis on mental health.  Evid Based Ment Health. 2015;18(2):40-46. doi:10.1136/eb-2015-102088PubMedGoogle ScholarCrossref
2.
Cortese  S, Tomlinson  A, Cipriani  A.  Meta review: network meta-analyses in child and adolescent psychiatry.  J Am Acad Child Adolesc Psychiatry. 2019;58(2):167-179. doi:10.1016/j.jaac.2018.07.891PubMedGoogle ScholarCrossref
3.
Cope  S, Zhang  J, Saletan  S, Smiechowski  B, Jansen  JP, Schmid  P.  A process for assessing the feasibility of a network meta-analysis: a case study of everolimus in combination with hormonal therapy versus chemotherapy for advanced breast cancer.  BMC Med. 2014;12:93. doi:10.1186/1741-7015-12-93PubMedGoogle ScholarCrossref
4.
Welton  NJ, Caldwell  DM, Adamopoulos  E, Vedhara  K.  Mixed treatment comparison meta-analysis of complex interventions: psychological interventions in coronary heart disease.  Am J Epidemiol. 2009;169(9):1158-1165. doi:10.1093/aje/kwp014PubMedGoogle ScholarCrossref
5.
Streiner  DL, Joffe  R.  The adequacy of reporting randomized, controlled trials in the evaluation of antidepressants.  Can J Psychiatry. 1998;43(10):1026-1030. doi:10.1177/070674379804301008PubMedGoogle ScholarCrossref
6.
Tierney  JF, Vale  C, Riley  R,  et al.  Individual participant data (IPD) meta‐analyses of randomised controlled trials: guidance on their use.  PLoS Med. 2015;12(7):e1001855. doi:10.1371/journal.pmed.1001855PubMedGoogle ScholarCrossref
Research Letter
April 17, 2019

Combining Pharmacological and Nonpharmacological Interventions in Network Meta-analysis in Psychiatry

Author Affiliations
  • 1Institute of Primary Health Care (BIHAM), University of Bern, Bern, Switzerland
  • 2Center for Innovation in Mental Health, Academic Unit of Psychology; Clinical and Experimental Sciences (CNS and Psychiatry), Faculty of Medicine, University of Southampton, Southhampton, England
  • 3Department of Psychiatry, University of Oxford, Warneford Hospital, Oxford, England
  • 4Oxford Health National Health Service Foundation Trust, Warneford Hospital, Oxford, England
JAMA Psychiatry. 2019;76(8):867-868. doi:10.1001/jamapsychiatry.2019.0574

Network meta-analyses (NMAs) assess the comparative associations of 2 or more interventions even if they have not been compared in a randomized clinical trial.1 The validity of NMAs is founded on the assumption of transitivity (ie, that effect modifiers do not substantially differ across the included trials).1 The popularity of NMAs on pharmacological or nonpharmacological interventions is increasing in psychiatry.2 Recent NMAs have combined pharmacological and nonpharmacologic interventions in the same network. Although this may be informative for developing guidelines, it is methodologically challenging and could compromise the validity of NMAs. We aimed to evaluate NMAs that combined pharmacological and nonpharmacological interventions and provide guidance on how to conduct them.

Methods

We searched PubMed, PsycINFO, Embase, OVID MEDLINE, biological abstracts, BIOSIS, and Web of Science from inception until August 31, 2018. We appraised NMAs of randomized clinical trials based on the approach proposed by Cope et al,3 focusing on (1) how the control node (or neutral comparator) was defined in the network geometry, (2) differences between pharmacological and nonpharmacological studies with respect to patient characteristics, and (3) the distribution of risk of bias (RoB) in the network. According to the approach of Cope et al,3 we checked if the association of these issues with the results was explored in the retained NMAs (eMethods in the Supplement).

Results

We retrieved 12 NMAs (eMethods in the Supplement). Eight were published between 2017 and 2018: 6 focused on adults, 5 on children/adolescents, and 1 on both. These NMAs covered several psychiatric conditions, including major depressive disorder, anxiety disorders, attention-deficit/hyperactivity disorder, obsessive compulsive disorder, bulimia nervosa, at-risk mental state, and poststroke depression (eMethods in the Supplement).

Five NMAs pooled different types of control conditions (eg, a placebo pill, psychological placebo, or sham intervention) into the same node of the network, assuming that these comparators have similar associations (eMethods in the Supplement). However, this hypothesis should be empirically tested via a meta-regression (when feasible) or subgroup/sensitivity analysis. Only 2 NMAs did so (eMethods in the Supplement).

The existing differences between pharmacological and nonpharmacological studies in patient characteristics for baseline disease severity or previous exposure to treatment were reported in only 3 NMAs and only 1 assessed its association with the results (eMethods in the Supplement). The heterogeneity of patient characteristics was unclear or had not been retrieved from primary studies in most of the NMAs.

We found 3 NMAs in which the risk of performance or detection bias was not distributed evenly across pharmacological and nonpharmacological studies (eMethods in the Supplement). Compared with pharmacological trials, those with nonpharmacological interventions were less likely to have participants, caregivers, and outcome assessors masked, which is often an unavoidable limitation as some nonpharmacological treatments cannot always be masked. Four NMAs performed a sensitivity analysis to assess the association of high RoB for lack of masking with the treatment effects, but most of the NMA data were too sparse to draw any conclusion (eMethods in the Supplement).

Discussion

Network meta-analyses that combine pharmacological and nonpharmacological interventions for psychiatric conditions may be prone to violating the transitivity assumption, which may affect their validity. The definition and classification of the control node in the geometry of the network could affect the results of the NMA. A novel approach called component NMA could address this issue, as it explores the treatment effects of interventions with multiple components.4 Furthermore, differences in baseline participants’ characteristics, study RoB, and the level of masking may represent a limitation of NMA in combining pharmacological and nonpharmacological therapies.5 An individual participant data NMA could overcome these limitations, as it allows exploring treatment-patient interactions to check RoB and obtain extra data from trialists.6 Caution is needed when pharmacological and nonpharmacological interventions are combined in an NMA, and the specific potential limitations of this type of NMAs should always be systematically and transparently discussed.

Back to top
Article Information

Corresponding Author: Cinzia Del Giovane, Institute of Primary Health Care (BIHAM), University of Bern, Mittelstrasse 43, 3012 Bern, Switzerland (cinzia.delgiovane@biham.unibe.ch).

Published Online: April 17, 2019. doi:10.1001/jamapsychiatry.2019.0574

Author Contributions: Dr Del Giovane has full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Concept and design: All authors.

Acquisition, analysis, or interpretation of data: Del Giovane, Cortese, Cipriani.

Drafting of the manuscript: Del Giovane, Cortese.

Critical revision of the manuscript for important intellectual content: Cipriani.

Statistical analysis: Del Giovane.

Administrative, technical, or material support: All authors.

Supervision: Cipriani.

Conflict of Interest Disclosures: Dr Cortese declares reimbursement for travel and accommodation expenses from the Association for Child and Adolescent Central Health (ACAMH) from lectures delivered for ACAMH and from Healthcare Convention for educational activity on attention-deficit/hyperactivity disorder. Dr Cipriani is supported by the National Institute for Health Research (NIHR) Oxford Cognitive Health Clinical Research Facility, by an NIHR research professorship (grant RP-2017-08-ST2-006), and by the NIHR Oxford Health Biomedical Research Centre (grant BRC-1215-20005). No other disclosures are reported.

Funding/Support: This study was funded by the NIHR Oxford Health Biomedical Research Centre (grant BRC-1215-20005; Dr Cipriani).

Role of the Funder/Sponsor: The NIHR did not have any role in design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

Disclaimer: The views expressed are those of the authors and not necessarily those of the UK National Health Service, the NIHR, or the UK Department of Health.

Additional Contributions: We thank Kali Tal, PhD, Institute of Primary Health Care (BIHAM), University of Bern, Switzerland, for her editorial suggestions. She was not compensated for her contributions.

References
1.
Mavridis  D, Giannatsi  M, Cipriani  A, Salanti  G.  A primer on network meta-analysis with emphasis on mental health.  Evid Based Ment Health. 2015;18(2):40-46. doi:10.1136/eb-2015-102088PubMedGoogle ScholarCrossref
2.
Cortese  S, Tomlinson  A, Cipriani  A.  Meta review: network meta-analyses in child and adolescent psychiatry.  J Am Acad Child Adolesc Psychiatry. 2019;58(2):167-179. doi:10.1016/j.jaac.2018.07.891PubMedGoogle ScholarCrossref
3.
Cope  S, Zhang  J, Saletan  S, Smiechowski  B, Jansen  JP, Schmid  P.  A process for assessing the feasibility of a network meta-analysis: a case study of everolimus in combination with hormonal therapy versus chemotherapy for advanced breast cancer.  BMC Med. 2014;12:93. doi:10.1186/1741-7015-12-93PubMedGoogle ScholarCrossref
4.
Welton  NJ, Caldwell  DM, Adamopoulos  E, Vedhara  K.  Mixed treatment comparison meta-analysis of complex interventions: psychological interventions in coronary heart disease.  Am J Epidemiol. 2009;169(9):1158-1165. doi:10.1093/aje/kwp014PubMedGoogle ScholarCrossref
5.
Streiner  DL, Joffe  R.  The adequacy of reporting randomized, controlled trials in the evaluation of antidepressants.  Can J Psychiatry. 1998;43(10):1026-1030. doi:10.1177/070674379804301008PubMedGoogle ScholarCrossref
6.
Tierney  JF, Vale  C, Riley  R,  et al.  Individual participant data (IPD) meta‐analyses of randomised controlled trials: guidance on their use.  PLoS Med. 2015;12(7):e1001855. doi:10.1371/journal.pmed.1001855PubMedGoogle ScholarCrossref
×