Initial depression severity and response to ......et al,5 using patient-level data from six studies,...
6
One of the most controversial studies in psychology, 1 a meta- analysis conducted by Kirsch et al 2 cited 1500 times, found that the efficacy of antidepressant treatment is attributable to decreased responsiveness to placebo among patients who were severely depressed rather than to increased responsiveness to medication. That analysis included data from 35 published and unpublished studies on fluoxetine, venlafaxine, nefazodone and paroxetine conducted between 1987 and 1999. A more recent analysis of the same data-set did not find that initial severity determined drug–placebo differences. 3 Khan et al examined the association of baseline severity and outcome in 45 phase II and III anti- depressant clinical trials. 4 They found that in the active treatment group in trials that included patients with severe depression, more severe depression at baseline was associated with more symptom reduction, yet in the placebo group with less reduction. Fournier et al, 5 using patient-level data from six studies, found a significant interaction between baseline severity and treatment, confirming Kirsch et al ’s conclusion. However, it was only due in small degree to decreased responsiveness to placebo. We retested the hypothesis that the relationship between initial severity and antidepressant efficacy is attributable to decreased responsiveness to placebo among patients with severe depression rather than to increased responsiveness to medication. We conducted both patient-level and meta-analysis of trial-level data. Method We used patient- and trial-level data from 34 randomised placebo-controlled trials (n = 10 737) (1987–2007) of citalopram, duloxetine, escitalopram, quetiapine and sertraline from the NEWMEDS registry 6 (see online supplement DS1 for a list of studies). This included all acute placebo-controlled trials of major depressive disorder in adult populations sponsored or owned by Pfizer (12 studies; active: n = 2455, placebo: n = 888), Lilly (11 studies; active: n = 2425, placebo: n = 1134), AstraZeneca (4 studies; active: n = 1021, placebo: n = 524) and Lundbeck (7 studies; active: n = 1509, placebo n = 781) on these five compounds. In three studies the Hamilton Rating Scale for Depression (HRSD) 7 was estimated based on the Montgomery–A ˚ sberg Depression Rating Scale (MADRS) 8 using equipercentile linking, which gives an equivalent score of one measure on the other measure. It was done using data from 16 studies that included both measures. Analysis of covariance of change from baseline on the HRSD, using last observation carried forward, was examined testing for baseline, a dichotomous variable of placebo v. active treatment and their interaction. A significant interaction of placebo v. active treatment and baseline score would support the hypothesis. To further test the hypothesis, linear and quadratic regression equations of baseline severity and change from baseline to end- point were run separately for placebo and active treatment. A larger regression coefficient for placebo than active treatment would support the hypothesis. Analysis was repeated for citalopram and sertraline only – the two drugs in common with Kirsch et al, 2 and also repeated by study, to examine possible study-level differences. Mixed-models analysis was used to test the data under the assumption that data were missing at random. As an alternative test, we compared effects for those patients with low (522), medium (22–25) and high (above 25) HRSD baseline scores. A significant interaction of baseline group and placebo v. active treatment and greater effects within placebo than active treatment groups would support the hypothesis. In addition, we repeated the above analysis using trial-level data to see whether differences in results obtained might be the result of differences in methodology. Trial-level data were weighted by adjusted inverse variance as done by Kirsch et al. 2 Results Patient-level results did not support our hypothesis. The inter- action of placebo v. active treatment and baseline severity was not significant (F = 1.19, P = 0.28; B = 0.045 (95% CI 70.035 to 0.125), b = 0.059). Linear and quadratic regression results were the same for both models and were almost the same for active treatment and for placebo (R 2 = 0.06, s.e. = 8.1; R 2 = 0.04, 1 Initial depression severity and response to antidepressants v. placebo: patient-level data analysis from 34 randomised controlled trials Jonathan Rabinowitz, Nomi Werbeloff, Francine S. Mandel, Franc ¸ ois Menard, Lauren Marangell and Shitij Kapur Summary Several often-cited meta-analysis have reported that the efficacy of antidepressant medications depends on the severity of depression. They found that drug–placebo differences increased as a function of initial severity, which was attributed to decreased responsiveness to placebo among patients with severe depression rather than to increased responsiveness to medication. We retested this using patient-level data and also undertaking a meta-analysis of trial-level data from 34 randomised placebo controlled trials (n = 10 737) from the NEWMEDS registry. Although our trial-level data support prevous findings, patient-level data do not; initial severity and outcomes were similarly related in treatment and placebo groups. Declaration of interest J.R. has received research support, and/or consultancy fees and/or travel support from Amgen, Janssen, JNJ, Eli Lilly, Pfizer, BiolineRx, F. Hoffmann-La Roche, Piere-Fabre, Kamada, Spatz Medical, Ethypharm and Avraham Pharmaceuticals and is on the scientific advisory board of MedAvante. F.M. is an employee of H. Lundbeck A/S. F.M. is an employee of Pfizer. L.M. was an employee of Eli Lilly and Company when this work was done. S.K. has received grant support from GSK and has served as a consultant and/or speaker for AstraZeneca, BiolineRx, BMS-Otsuka, Eli Lilly, Janssen (J&J), Lundbeck, Neuro- Search, Pfizer, F. Hoffmann-La Roche, Servier and Solvay Wyeth. Copyright and usage B The Royal College of Psychiatrists 2016. The British Journal of Psychiatry 1–2. doi: 10.1192/bjp.bp.115.173906 Short report
Transcript of Initial depression severity and response to ......et al,5 using patient-level data from six studies,...
One of the most controversial studies in psychology,1 a meta-analysis conducted by Kirsch et al2 cited 1500 times, found thatthe efficacy of antidepressant treatment is attributable to decreasedresponsiveness to placebo among patients who were severelydepressed rather than to increased responsiveness to medication.That analysis included data from 35 published and unpublishedstudies on fluoxetine, venlafaxine, nefazodone and paroxetineconducted between 1987 and 1999. A more recent analysis ofthe same data-set did not find that initial severity determineddrug–placebo differences.3 Khan et al examined the associationof baseline severity and outcome in 45 phase II and III anti-depressant clinical trials.4 They found that in the active treatmentgroup in trials that included patients with severe depression, moresevere depression at baseline was associated with more symptomreduction, yet in the placebo group with less reduction. Fournieret al,5 using patient-level data from six studies, found a significantinteraction between baseline severity and treatment, confirmingKirsch et al’s conclusion. However, it was only due in small degreeto decreased responsiveness to placebo. We retested the hypothesisthat the relationship between initial severity and antidepressantefficacy is attributable to decreased responsiveness to placeboamong patients with severe depression rather than to increasedresponsiveness to medication. We conducted both patient-leveland meta-analysis of trial-level data.
Method
We used patient- and trial-level data from 34 randomisedplacebo-controlled trials (n= 10 737) (1987–2007) of citalopram,duloxetine, escitalopram, quetiapine and sertraline from theNEWMEDS registry6 (see online supplement DS1 for a list ofstudies). This included all acute placebo-controlled trials of majordepressive disorder in adult populations sponsored or owned byPfizer (12 studies; active: n= 2455, placebo: n= 888), Lilly (11 studies;active: n= 2425, placebo: n= 1134), AstraZeneca (4 studies; active:n= 1021, placebo: n= 524) and Lundbeck (7 studies; active:n= 1509, placebo n= 781) on these five compounds. In three
studies the Hamilton Rating Scale for Depression (HRSD)7 wasestimated based on the Montgomery–Asberg Depression RatingScale (MADRS)8 using equipercentile linking, which gives anequivalent score of one measure on the other measure. It was doneusing data from 16 studies that included both measures.
Analysis of covariance of change from baseline on the HRSD,using last observation carried forward, was examined testing forbaseline, a dichotomous variable of placebo v. active treatmentand their interaction. A significant interaction of placebo v. activetreatment and baseline score would support the hypothesis. Tofurther test the hypothesis, linear and quadratic regressionequations of baseline severity and change from baseline to end-point were run separately for placebo and active treatment. Alarger regression coefficient for placebo than active treatmentwould support the hypothesis. Analysis was repeated forcitalopram and sertraline only – the two drugs in common withKirsch et al,2 and also repeated by study, to examine possiblestudy-level differences. Mixed-models analysis was used to testthe data under the assumption that data were missing at random.
As an alternative test, we compared effects for those patientswith low (522), medium (22–25) and high (above 25) HRSDbaseline scores. A significant interaction of baseline group andplacebo v. active treatment and greater effects within placebo thanactive treatment groups would support the hypothesis. Inaddition, we repeated the above analysis using trial-level data tosee whether differences in results obtained might be the result ofdifferences in methodology. Trial-level data were weighted byadjusted inverse variance as done by Kirsch et al.2
Results
Patient-level results did not support our hypothesis. The inter-action of placebo v. active treatment and baseline severity wasnot significant (F= 1.19, P= 0.28; B= 0.045 (95% CI 70.035 to0.125), b= 0.059). Linear and quadratic regression results werethe same for both models and were almost the same for activetreatment and for placebo (R2 = 0.06, s.e. = 8.1; R2 = 0.04,
1
Initial depression severity and response toantidepressants v. placebo: patient-level dataanalysis from 34 randomised controlled trialsJonathan Rabinowitz, Nomi Werbeloff, Francine S. Mandel, Francois Menard, Lauren Marangelland Shitij Kapur
SummarySeveral often-cited meta-analysis have reported that theefficacy of antidepressant medications depends on theseverity of depression. They found that drug–placebodifferences increased as a function of initial severity, whichwas attributed to decreased responsiveness to placeboamong patients with severe depression rather than toincreased responsiveness to medication. We retested thisusing patient-level data and also undertaking a meta-analysisof trial-level data from 34 randomised placebo controlledtrials (n= 10 737) from the NEWMEDS registry. Although ourtrial-level data support prevous findings, patient-level data donot; initial severity and outcomes were similarly related intreatment and placebo groups.
Declaration of interestJ.R. has received research support, and/or consultancy feesand/or travel support from Amgen, Janssen, JNJ, Eli Lilly,Pfizer, BiolineRx, F. Hoffmann-La Roche, Piere-Fabre, Kamada,Spatz Medical, Ethypharm and Avraham Pharmaceuticals andis on the scientific advisory board of MedAvante. F.M. is anemployee of H. Lundbeck A/S. F.M. is an employee of Pfizer.L.M. was an employee of Eli Lilly and Company when this workwas done. S.K. has received grant support from GSK and hasserved as a consultant and/or speaker for AstraZeneca,BiolineRx, BMS-Otsuka, Eli Lilly, Janssen (J&J), Lundbeck, Neuro-Search, Pfizer, F. Hoffmann-La Roche, Servier and Solvay Wyeth.
Copyright and usageB The Royal College of Psychiatrists 2016.
The British Journal of Psychiatry
1–2. doi: 10.1192/bjp.bp.115.173906
Short report
s.e. = 7.9, respectively) showing a small increase in change frombaseline as a function of baseline severity in both the active andplacebo groups (see online Fig. DS1). The above analysis repeatedby study found that the interaction was not significant in 25 trials(P40.18 and P50.93), trend level in 3/34 trials (P= 0.06, 0.08,0.09) and significant in 6/34 trials. The additional mixed-modelsanalysis found that the interaction of placebo v. active treatmentand baseline severity was significant in 3/34 studies. For thecitalopram and sertraline studies, two compounds overlappingwith Kirsch et al’s study, there was no significant interaction ofplacebo v. active treatment and baseline severity (F= 0.29,P= 0.59). The linear and quadratic regression results were thesame for both regression models and were almost the same foractive treatment and for placebo (R2 = 0.03, s.e. = 8.4; R2 = 0.02,s.e. = 8.1, respectively). Results did not differ using random- andfixed-effects models. The difference in active v. placebo changefrom baseline between the low, medium and high severity groupswas not statistically significant (P= 0.25, Table 1).
To test whether the differences between these results and thoseof Kirsch et al are the result of differences between a meta-analysisthat is limited by having only aggregate-level data on a studyand an analysis of patient-level data, analysis was repeated onaggregate-level data. Unlike the patient-level analysis, whenexamining baseline severity as a continuous variable there was asignificant interaction between placebo v. active treatment andbaseline severity (F= 9.27, P= 0.002). Drug and placebo efficacyincreased as initial severity increased, with baseline severityexplaining more of the variance in the placebo group than inthe active treatment group. The linear and quadratic regressionresults for active treatment were R2 = 0.26, R2 = 0.28 and forplacebo R2 = 0.32, R2 = 0.40 (online Fig. DS2). The difference inactive v. placebo change from baseline between the lower, mediumand high severity groups was not statistically significant (P= 0.99).
Discussion
Baseline severity was not associated with a more pronouncedchange from baseline in the active- v. placebo-treated patients whenusing patient-level data, but was evident to some extent when usingaggregate trial-level data. The patient-level analysis does not supportthe findings of the previous meta-analyses2,4,5 that antidepressantsact at the same magnitude irrespective of initial severity whileplacebo changes as a function of baseline severity. Patient-leveldata are more sensitive than trial-level data in measuring theeffects in question as they allow for adjusting each patient’s changescore by their baseline value and other patient-level characteristics.
The difference between our results and Kirsch et al’s2 appearto be the result of differences in methodology – meta-analysis v.patient-level analysis. This is supported by our finding that whenexamining the same drugs as Kirsch et al using patient-levelanalysis, we did not find the effect that their study did. However, wenote that for the most part our studies did not overlap with thoseincluded in the work of Kirsch et al. Caution is advised when exam-ining positive relationships between baseline severity and symptomimprovement as these may be the result of regression to the mean.
Jonathan Rabinowitz, PhD, Nomi Werbeloff, PhD, School of Social Work, RamatGan, Bar Ilan University, Israel; Francine S. Mandel, PhD, Pfizer DevelopmentOperations, New York, USA; Francois Menard, MD, Lundbec Research Department,Paris, France; Lauren Marangell, MD, Department of Psychiatry, University of TexasHealth Science Center, Houston, Texas, USA; Shitij Kapur, MD, Institute ofPsychiatry, Psychology and Neuroscience, Kings College London, London, UK
Correspondence: Jonathan Rabinowitz, Bar Ilan University, Ramat Gan,Israel 52900. Email: [email protected]
Received 19 Apr 2015, final revision 23 Dec 2015, accepted 27 Feb 2016
Funding
The research leading to these results has received support from the Innovative MedicineInitiative Joint Undertaking under grant agreement number: 115008 of which resourcesare composed of a European Federation of Pharmaceutical Industries and Associations(EFPIA) in-kind contribution and financial contribution from the European Union’s SeventhFramework Programme (FP7/2007-2013). The funding sources were not involved in thecollection, analysis, and interpretation of data; in the writing of the report; and nor inthe decision to submit the paper for publication.
References
1 Jarrett C. The 10 most controversial psychology studies ever published. ResDig: British Psychological Society 2014; 19 September (http://digest.bps.org.uk/2014/09/the-10-most-controversial-psychology.html).
2 Kirsch I, Deacon BJ, Huedo-Medina TB, Scoboria A, Moore TJ, Johnson BT.Initial severity and antidepressant benefits: a meta-analysis of datasubmitted to the Food and Drug Administration. PLoS Med 2008; 5: e45.
3 Fountoulakis KN, Veroniki AA, Siamouli M, Moller HJ. No role for initialseverity on the efficacy of antidepressants: results of a multi-meta-analysis.Ann Gen Psychiatry 2013; 12: 26.
4 Khan A, Leventhal RM, Khan SR, Brown WA. Severity of depression andresponse to antidepressants and placebo: an analysis of the Food and DrugAdministration database. J Clin Psychopharmacol 2002; 22: 40–5
5 Fournier JC, DeRubeis RJ, Hollon SD, Dimidjian S, Amsterdam JD, Shelton RC,et al. Antidepressant drug effects and depression severity: a patient-levelmeta-analysis. JAMA 2010; 303: 47–53.
6 Institute of Medicine. Models of data sharing: THE NEWMEDS CONSORTIUM.Sharing Clinical Research Data: Workshop Summary: 37–38. The NationalAcademies Press, 2013 (http://www.nap.edu/read/18267/chapter/5G39).
7 Hamilton M. A rating scale for depression. J Neurol Neurosurg Psychiatry1960; 23: 56–62.
8 Montgomery SA, Asberg M. A new depression scale designed to be sensitiveto change. Br J Psychiatry 1979; 134: 382–9.
2
Rabinowitz et al
Table 1 Placebo–active differences in Hamilton Rating Scale for Depression (HRSD) scores and drop-out rates by group
HRSD score, mean (s.d.)
Baseline Change from baseline Drug–placebo difference,a mean (95% CI), s.e. Drop-out, %
Treatment group 72.05 (72.38 to 71.72) 0.17
Placebo (n= 3258) 23.0 (4.1) 78.8 (8.1) 35.6
Active (n= 7323) 23.1 (4.2) 710.8 (8.4) 35.0
Baseline severity
Low (less than 22) 72.04 (72.50 to 71.58) 0.24
Placebo (n= 1328) 19.3 (2.7) 77.1 (7.2) 34.8
Active (n= 3046) 19.4 (2.8) 79.1 (7.4) 34.5
Medium (22–25) 71.82 (72.40 to 71.24) 0.30
Placebo (n= 1102) 23.8 (0.8) 79.2 (8.0) 31.9
Active (n= 2345) 23.8 (0.8) 711.0 (8.1) 33.3
High (above 25) 72.41 (73.17 to 71.64) 0.39
Placebo (n= 828) 28.0 (2.4) 710.7 (9.1) 40.0
Active (n= 1932) 28.1 (2.6) 713.1 (9.4) 35.6
a. Largest pairwise difference high 72.41 v. medium 71.82, t= 1.16, d.f. = 6205, P= 0.25.
Data supplement to Rabinowitz et al. Initial severity not related to antidepressant benefits or placebo response: patient-level data analysis from 34 randomised controlled trials. Br J Psychiatry doi: 10.1192/bjp.bp.115.173906
Online supplement DS1
List of studies included
Name / Publication / Registration number
Treatment (active treatment, placebo sample sizes)
Astra Zeneca D1448C00001 [1] NCT00320268 Quetiapine: n=539; Placebo: n=184 D1448C00002 [2] NCT00321490 Quetiapine: n=304; Placebo: n=157 D1448C00003 [3] NCT00326144 Quetiapine: n=154; Placebo: n=156 D1448C00014 [4] NCT00388973 Quetiapine: n=24; Placebo: n=27 Lundbeck 85A Citalopram: n=89; Placebo: n=91 89303 Citalopram: n=134; Placebo: n=66 89306 Citalopram: n=185; Placebo: n=89 91206 Citalopram: n=521; Placebo: n=129 99001 Escitalopram: n=154; Placebo: n=147 99003 Escitalopram + Citalopram: n=157; Placebo: n=117 99024 Escitalopram + Fluoxetine: n=269; Placebo: n=142
Pfizer 101 Sertraline: n=96; Placebo: n=25 103 Sertraline: n=278; Placebo: n=91 104 Sertraline + Amisulpride: n=299; Placebo: n=150 109 Sertraline + Amisulpride: n=211; Placebo: n=104 111 Sertraline + Fluoxetine: n=225; Placebo: n=110 310 Sertraline: n=140; Placebo: n=34 315 Sertraline + Amisulpride: n=162; Placebo: n=80 n-238 Sertraline: n=159; Placebo: n=40 n-240 Sertraline + Desipramine: n=70; Placebo: n=35 n-247 Sertraline + Desipramine: n=88; Placebo: n=42 r-0167 Sertraline + Amisulpride: n=259; Placebo: n=126
A0501075 NCT00636246
Sertraline + Reboxetine: n=468; Placebo: n=51
Lilly F1J-MC-HMAQA Duloxetine + Fluoxetine: n=103; Placebo: n=70 F1J-MC-HMAQB Duloxetine + Fluoxetine: n=119; Placebo: n=74 F1J-MC-HMATA Duloxetine + Paroxetine: n=264; Placebo: n=90 F1J-MC-HMATB Duloxetine + Paroxetine: n=264; Placebo: n=89 F1J-MC-HMAYA Duloxetine + Paroxetine: n=274; Placebo: n=93 F1J-MC-HMAYB Duloxetine + Paroxetine: n=293; Placebo: n=99 F1J-MC-HMBHA Duloxetine: n=123; Placebo: n=122 F1J-MC-HMBHB Duloxetine: n=128; Placebo: n=139 F1J-MC-HQAC Duloxetine: n=34; Placebo: n=35 F1J-US-HMFA [5] NCT00406848 Duloxetine: n=237; Placebo: n=133
F1J-US-HMFS NCT00536471 Duloxetine: n=586; Placebo: n=190
Additional References 1. Weisler, R., et al., Extended release quetiapine fumarate monotherapy for major depressive
disorder: results of a double-blind, randomized, placebo-controlled study. CNS Spectr, 2009. 14(6): p. 299-313.
2. Cutler, A.J., et al., Extended release quetiapine fumarate monotherapy in major depressive disorder: a placebo- and duloxetine-controlled study. J Clin Psychiatry, 2009. 70(4): p. 526-39.
3. Bortnick, B., et al., Efficacy and tolerability of extended release quetiapine fumarate (quetiapine XR) monotherapy in major depressive disorder: a placebo-controlled, randomized study. J Affect Disord, 2011. 128(1-2): p. 83-94.
4. Katila, H., et al., Randomized, double-blind study of the efficacy and tolerability of extended release quetiapine fumarate (quetiapine XR) monotherapy in elderly patients with major depressive disorder. Am J Geriatr Psychiatry, 2013. 21(8): p. 769-84.
5. Oakes, T.M., et al., Safety and tolerability of duloxetine in elderly patients with major depressive disorder: a pooled analysis of two placebo-controlled studies. Int Clin Psychopharmacol, 2013. 28(1): p. 1-11.
Figure DS1. Change on Hamilton Rating Scale for Depression (HRSD, also known as the HAMD) to end-point by baseline severity – patient level data
Figure DS2. Change on Hamilton Rating Scale for Depression (HRSD, also known as the HAMD) to end-point by baseline severity – trial level meta-analysis
10.1192/bjp.bp.115.173906 published online May 19, 2016 Access the most recent version at DOI: BJP KapurJonathan Rabinowitz, Nomi Werbeloff, Francine S. Mandel, François Menard, Lauren Marangell and Shitijcontrolled trialsplacebo: patient-level data analysis from 34 randomised Initial depression severity and response to antidepressants v.
MaterialSupplementary
http://bjp.rcpsych.org/content/suppl/2016/05/09/bjp.bp.115.173906.DC1Supplementary material can be found at:
Referenceshttp://bjp.rcpsych.org/content/early/2016/05/03/bjp.bp.115.173906#BIBLThis article cites 0 articles, 0 of which you can access for free at:
permissionsReprints/
[email protected] to To obtain reprints or permission to reproduce material from this paper, please
P<P Published online 2016-05-19T00:05:14-07:00 in advance of the print journal.
to this article atYou can respond /letters/submit/bjprcpsych;bjp.bp.115.173906v1
from Downloaded
The Royal College of PsychiatristsPublished by on December 14, 2017http://bjp.rcpsych.org/
digital object identifier (DOIs) and date of initial publication. theindexed by PubMed from initial publication. Citations to Advance online articles must include
final publication). Advance online articles are citable and establish publication priority; they areappeared in the paper journal (edited, typeset versions may be posted when available prior to Advance online articles have been peer reviewed and accepted for publication but have not yet
http://bjp.rcpsych.org/site/subscriptions/ go to: The British Journal of PsychiatryTo subscribe to
