Introduction
It has been reported that use of non-steroidal anti-inflammatory drugs (NSAIDs) may interfere with the efficacy of antidepressant drugs and significantly contribute to treatment resistance in major depressive disorder (MDD) (Warner-Schmidt et al. Reference Warner-Schmidt, Vanover, Chen, Marshall and Greengard2011). Since antidepressants and NSAIDs are among the most commonly used types of medication, this finding has potentially far-reaching clinical implications. Indeed, Warner-Schmidt and colleagues have proposed that clinicians should balance the need for NSAIDs with their potentially negative impact through compromising the efficacy of antidepressants (Warner-Schmidt et al. Reference Warner-Schmidt, Vanover, Chen, Marshall and Greengard2011). However, such clinical recommendation requires a replication, specification and extension of the reported effect.
The proposal that NSAIDs may antagonize the therapeutic action of antidepressants rests on a series of mouse experiments and an analysis of a human clinical study (Warner-Schmidt et al. Reference Warner-Schmidt, Vanover, Chen, Marshall and Greengard2011). The mouse experiments show that NSAIDs inhibit the molecular and behavioural actions of serotonin-reuptake inhibiting (SRI) antidepressants, but not antidepressants from other classes (e.g. tricyclic antidepressants) in mouse models of depression (Warner-Schmidt et al. Reference Warner-Schmidt, Vanover, Chen, Marshall and Greengard2011). The human clinical study analysis found that subjects who used NSAIDs during treatment with citalopram, a SRI antidepressant, had a significantly lower chance of achieving remission than those who did not use NSAIDs (Warner-Schmidt et al. Reference Warner-Schmidt, Vanover, Chen, Marshall and Greengard2011). While the synergy of rodent and human findings is strongly suggestive, two important questions remain unanswered and will be addressed in the present study.
First, the rodent data suggest that the effect of NSAIDs is relatively specific to SRI antidepressants and that the therapeutic effects of other types of antidepressants (e.g. tricyclics) may not be affected by NSAIDs (Warner-Schmidt et al. Reference Warner-Schmidt, Vanover, Chen, Marshall and Greengard2011). This is most relevant clinically since a differential effect of NSAIDs could be used to personalize the antidepressant prescription by choosing non-SRI antidepressants in patients who require NSAIDs. Here we test the hypothesis that NSAIDs differentially affect the therapeutic efficacy of escitalopram (a SRI antidepressant) and nortriptyline (a tricyclic antidepressant) in a large comparative human study.
Second, the reported clinical analysis was restricted to subjects with available data at week 12 of treatment and did not control for any covariates (Warner-Schmidt et al. Reference Warner-Schmidt, Vanover, Chen, Marshall and Greengard2011). It is possible that those taking NSAIDs were older, more severely depressed, or more or less likely to drop out of the study. Since factors like age and baseline depression severity influence the therapeutic efficacy of antidepressants (Uher et al. Reference Uher, Maier, Hauser, Marusic, Schmael, Mors, Henigsberg, Souery, Placentino, Rietschel, Zobel, Dmitrzak-Weglarz, Petrovic, Jorgensen, Kalember, Giovannini, Barreto, Elkin, Landau, Farmer, Aitchison and McGuffin2009a ; Fountoulakis & Moller, Reference Fountoulakis and Moller2011), it is possible that the reported effects reflect confounding by these variables rather than genuine effect of NSAIDs on antidepressant efficacy. Here we attempt to replicate the effect of NSAIDs on antidepressant efficacy while controlling for relevant clinical variables.
Methods
Sample
The study described in this paper is a secondary analysis using data from the Genome-based Therapeutic Drugs for Depression (GENDEP) study (Uher et al. Reference Uher, Maier, Hauser, Marusic, Schmael, Mors, Henigsberg, Souery, Placentino, Rietschel, Zobel, Dmitrzak-Weglarz, Petrovic, Jorgensen, Kalember, Giovannini, Barreto, Elkin, Landau, Farmer, Aitchison and McGuffin2009a ). This 12-week, partially randomized, clinical and pharmacogenetic study included 811 patients (297 male, 514 female) with MDD, current depressive episode of at least moderate severity, diagnosed according to DSM-IV/ICD-10 criteria in a semi-structured Schedule for Clinical Assessment in Neuropsychiatry (SCAN) interview (Wing et al. Reference Wing, Sartorius and Ustin1998). Current substance dependence and family or personal history of bipolar depression and schizophrenia were exclusion criteria. Treatment included protocol-guided escitalopram, a serotonin-reuptake inhibitor (dose 10–30 mg daily) or nortriptyline, a tricyclic antidepressant, (dose 50–200 mg daily) for 12 weeks. Participating patients with no contraindications (n=468) were randomly allocated to receive protocol-guided escitalopram (n=233) or nortriptyline (n=235) for 12 weeks. Those with absolute or relative contraindications to one of the study drugs were allocated non-randomly to the other drug, 225 to escitalopram and 118 to nortriptyline. Compliance was confirmed by measuring blood levels of medication at week 8. Of the 811 recruited patients, 628 (77%) completed at least 8 weeks of antidepressant treatment. Severity of depression at baseline and after each week of treatment was measured with the clinician-rated Montgomery–Åsberg Depression Rating Scale (MADRS), the 17-item Hamilton Rating Scale for Depression (HAMD17) and the self-reported Beck Depression Inventory (BDI) with good psychometric properties and high inter-rater reliability for the clinician-rated instruments (Uher et al. Reference Uher, Farmer, Maier, Rietschel, Hauser, Marusic, Mors, Elkin, Williamson, Schmael, Henigsberg, Perez, Mendlewicz, Janzing, Zobel, Skibinska, Kozel, Stamp, Bajs, Placentino, Barreto, McGuffin and Aitchison2008). The MADRS was the primary outcome measure. Nortriptyline and escitalopram led to similar degree of improvement on the three outcome scales (Uher et al. Reference Uher, Maier, Hauser, Marusic, Schmael, Mors, Henigsberg, Souery, Placentino, Rietschel, Zobel, Dmitrzak-Weglarz, Petrovic, Jorgensen, Kalember, Giovannini, Barreto, Elkin, Landau, Farmer, Aitchison and McGuffin2009a ). Details of study design, sample and outcomes are available in previous publications (Uher et al. Reference Uher, Maier, Hauser, Marusic, Schmael, Mors, Henigsberg, Souery, Placentino, Rietschel, Zobel, Dmitrzak-Weglarz, Petrovic, Jorgensen, Kalember, Giovannini, Barreto, Elkin, Landau, Farmer, Aitchison and McGuffin2009a , Reference Uher, Mors, Hauser, Rietschel, Maier, Kozel, Henigsberg, Souery, Placentino, Perroud, Dernovsek, Strohmaier, Larsen, Zobel, Leszczynska-Rodziewicz, Kalember, Pedrini, Linotte, Gunasinghe, Aitchison, McGuffin and Farmer b , Reference Uher, Muthen, Souery, Mors, Jaracz, Placentino, Petrovic, Zobel, Henigsberg, Rietschel, Aitchison, Farmer and McGuffin2010, Reference Uher, Dernovsek, Mors, Hauser, Souery, Zobel, Maier, Henigsberg, Kalember, Rietschel, Placentino, Mendlewicz, Aitchison, McGuffin and Farmer2011). The GENDEP project was approved by the ethics boards in all participating centres and all participants provided written informed consent.
The use of anti-inflammatory and analgesic drugs
Names, doses and frequency of use of any medication were recorded for each week in the study by the assessing clinician using a purpose-tailored medication form. Drug names were coded into 32 groups by medically trained assessors. In agreement with Warner-Schmidt and colleagues (Reference Warner-Schmidt, Vanover, Chen, Marshall and Greengard2011), we considered NSAIDs and analgesics separately. NSAIDs included ibuprofen, ketoprofen, naproxen, diclofenac, indomethacin, piroxicam, meloxicam, tenoxicam, mefenamic acid, flufenamic acid, tolfenamic acid, celecoxib, rofecoxib, acetylsalicylic acid (aspirin) and salsalate. Analgesics included paracetamol (acetaminophen) and its derivatives but excluded opiates, which were coded separately. Since NSAIDs and analgesics are frequently used as a one-off medication or are used in very small doses for other indications (e.g. low-dose acetylsalicylic acid as an anti-platelet agent), we sought to quantify if the medication was taken at a significant dosage which is likely to have anti-inflammatory/analgesic effect for a significant period of time. Dosage of each drug was coded with reference to the effective dose range recommended by the British National Formulary (BMA/RPS, 2008): dose of at least the minimum recommended therapeutic dose for ⩾1 week was considered as therapeutically significant.
Statistical analysis
The effects of NSAIDs and analgesics on outcomes of treatment with escitalopram and nortriptyline were primarily tested using linear mixed-effect repeated-measures regression analyses. These models allow the use of all available information across repeated measurements, separate effects that generalize across individuals and recruitment centres from those that are individual- or centre-specific and efficiently handle missing data (Gueorguieva & Krystal, Reference Gueorguieva and Krystal2004; Lane, Reference Lane2008). The dependent variables were the total scores on depression rating scales (MADRS, HAMD17, BDI) at up to 12 weekly assessment occasions during the active treatment. All models corrected for sex, age, baseline depression severity, time in the study and centre where the individual was recruited (Uher et al. Reference Uher, Maier, Hauser, Marusic, Schmael, Mors, Henigsberg, Souery, Placentino, Rietschel, Zobel, Dmitrzak-Weglarz, Petrovic, Jorgensen, Kalember, Giovannini, Barreto, Elkin, Landau, Farmer, Aitchison and McGuffin2009a , Reference Uher, Mors, Hauser, Rietschel, Maier, Kozel, Henigsberg, Souery, Placentino, Perroud, Dernovsek, Strohmaier, Larsen, Zobel, Leszczynska-Rodziewicz, Kalember, Pedrini, Linotte, Gunasinghe, Aitchison, McGuffin and Farmer b ).
To achieve the optimal balance between statistical power and complete informative analysis of the dataset, we distinguished between primary and secondary analyses. The primary analysis tested the effect of concurrent medication of NSAID on outcome assessed with the primary outcome measure (MADRS). For this primary analysis, we reserved the nominal significance threshold of p<0.05. Secondary analyses tested the effects of analgesics or NSAIDs and analgesics, taken in any or in at least minimal therapeutically significant doses, on changes in depression severity measured by each of the three outcome scales. For these secondary analyses, we applied a significance threshold corrected for the number of tests performed for each hypothesis (p<0.05/18=0.0028).
In addition, in order to replicate the analysis reported by Warner-Schmidt and colleagues (Reference Warner-Schmidt, Vanover, Chen, Marshall and Greengard2011), we also performed secondary analyses with logistic regressions to test the effects of NSAIDs and analgesics on remission, defined as depression severity score at the last treatment visit below an agreed clinical significance level: MADRS⩽10, HAMD17⩽7, BDI⩽10 (Frank et al. Reference Frank, Prien, Jarrett, Keller, Kupfer, Lavori, Rush and Weissman1991; Moller, Reference Moller2009). To explore the possibility of confounding, we report the odds ratios for remission from uncorrected univariate model (with NSAIDs/analgesics as the only predictor) and after correcting for a progressive range of relevant covariates (baseline severity of depression, sex, age, centre of recruitment).
The hypothesis was concerned primarily with NSAIDs, but a previous report found a similar effect for the non-NSAID non-opiate analgesics (mainly paracetamol) (Warner-Schmidt et al. Reference Warner-Schmidt, Vanover, Chen, Marshall and Greengard2011). Therefore, in each model, we first tested the effect of NSAIDs to provide the most specific test of the hypothesis and, second, we proceeded to test the effect of taking either NSAIDs or analgesics to carry out a less specific test with increased statistical power.
Each test was performed separately for escitalopram- and for nortriptyline-treated participants. In addition, differential prediction of outcome of treatment by SRI and tricyclic antidepressants were tested as predictor–treatment interaction. Given the partly randomized design, the latter analysis was restricted to patients who were randomly allocated to treatment with escitalopram or nortriptyline to exclude selection bias.
Power analysis
We aimed to test if concurrent use of NSAIDs clinically significantly influences the efficacy of the SRI antidepressant escitalopram. According to a consensus guideline, a difference in outcome of at least 3 points on the HAMD17 can be considered clinically significant (NICE, 2004). Therefore we calculated statistical power to detect a difference in outcome of ⩾3 points on the HAMD17 between those taking NSAIDs and those not taking NSAIDs. The sample of 458 escitalopram-treated patients, of whom 10% were taking NSAIDs, provided a power of 83% to detect such clinically significant effect at the nominal 0.05 level of statistical significance.
Results
Concurrent medication with NSAIDs and analgesics in the GENDEP sample
Of the 811 patients in the GENDEP study, 78 (10%) were taking NSAIDs and 59 (7%) were taking analgesics during the 12-week study. Of these, 43 took NSAIDs in a significant dose and 43 took analgesics in a significant dose. Either NSAIDs or analgesics were taken by 118 (15%) individuals and 74 (9%) took these in therapeutically significant doses. There were no differences in the rate of concurrent medication with NSAIDs or analgesics between those treated with escitalopram and those treated with nortriptyline (Table 1).
Table 1. Sample description and concurrent medication with non-steroidal anti-inflammatory drugs (NSAIDs) and analgesics
BDI, Beck Depression Inventory, HAMD17, 17-item Hamilton Rating Scale for Depression; MADRS, Montgomery–Åsberg Depression Rating Scale.
The descriptives are stratified by antidepressant (escitalopram or nortriptyline) and NSAID use.
We explored the relationship between concurrent medication and potential confounders, including age, sex, and severity of depression at baseline. We found that older subjects were significantly more likely to take NSAIDs [odds ratio (OR) 1.39, 95% confidence interval (CI) 1.13–1.71, p=0.0016] and analgesics (OR 1.33, 95% CI 1.06–1.68, p=0.0146). Use of analgesics was also significantly associated with higher baseline severity of depression, measured by either MADRS (OR 1.39, 95% CI 1.07–1.81, p=0.0126) or BDI (OR 1.70, 95% CI 1.30–2.22, p=0.0001). Men and women were equally likely to take NSAID and analgesics (all p>0.05).
Effect of concurrent medication with NSAIDs and analgesics on antidepressant efficacy
We first conducted the primary analysis, testing the effect of concurrent medication with NSAIDs on the primary outcome (MADRS) of treatment with the SRI escitalopram in a linear mixed-effect, repeated-measures regression model adjusted for sex, age, baseline depression severity, time in the study and centre of recruitment. This model showed that concurrent medication with NSAIDs had no significant effect on the outcome of escitalopram treatment in the GENDEP study (Table 2). The negative results were confirmed with the two secondary outcome measures: HAMD17 and BDI (Table 2). Concurrent medication with analgesics, or NSAIDs and analgesics combined, did not significantly affect outcome either (Table 2). The results remained essentially unchanged when the prediction was restricted to those taking therapeutically significant doses of NSAIDs and/or analgesics (Table 2). In summary, all 18 tests attempting to detect an effect of NSAIDs and/or analgesics on the outcome of treatment with escitalopram gave negative results.
Table 2. Effect of non-steroidal anti-inflammatory drugs (NSAIDs) and analgesics on outcome of treatment with escitalopram and nortriptyline in the GENDEP study: results of linear mixed-effect repeated-measures analyses
BDI, Beck Depression Inventory; CI, confidence interval; GENDEP, Genome-based Therapeutic Drugs for Depression study; HAMD17, 17-item Hamilton Rating Scale for Depression; MADRS, Montgomery–Åsberg Depression Rating Scale.
β, Standardized regression coefficient; CI, confidence interval for the standardized regression coefficient. Positive β means worse outcome in individuals with concurrent medication; negative β reflects better outcome among individuals with concurrent medication. p values in bold denote nominal statistical significance (<0.05).
Analyses testing the effect of concurrent NSAIDs and/or analgesics on outcomes of treatment with the tricyclic antidepressant nortriptyline suggested no consistent effects. One exception was a nominally significant effect of concurrent medication with either NSAID or analgesic on the secondary outcome of self-reported depression severity on BDI. This test suggested a slightly better outcome in those taking NSAIDs or analgesics. However, the findings that only one of the 18 analyses was nominally significant and that the effect was halved when the prediction was restricted to individuals taking clinically significant doses of NSAIDs or analgesics indicate that this is most likely a chance finding (Table 2).
Finally, we aimed to discover whether nortriptyline might be significantly superior to escitalopram among individuals taking concurrent NSAIDs and/or analgesics. This was tested as an interaction between drug and concurrent medication (NSAIDs and/or analgesics) in the reduced samples of individuals who were randomly allocated to receive either escitalopram or nortriptyline. Analysis of the primary outcome measure (MADRS) indicated no significant interaction (Table 2). In one of the secondary outcome measures (the self-reported BDI), there was a nominally significant interaction, suggesting a superiority of nortriptyline over escitalopram among individuals with depression who were also taking analgesics (or NSAIDs and analgesics). However, these results of secondary analyses did not reach the corrected threshold for statistical significance and, as with the effects in the nortriptyline arm (which contribute to this interaction), the effect size of this interaction was halved when the prediction was restricted to individuals taking analgesics (and NSAIDs) at therapeutically effective doses (Table 2).
Concurrent NSAID/analgesics and remission: the effects of covariates
To replicate previously reported analyses, we tested the relationship between the concurrent use of NSAIDs and remission on the primary outcome measure (MADRS ⩽10 at last treatment visit) among escitalopram-treated individuals. Of the 48 individuals using NSAIDs, 20 (42%) achieved remission. Of the 410 individuals who were not taking NSAIDs, 191 (47%) achieved remission with up to 12 weeks of escitalopram treatment. Univariate logistic regression with no covariates revealed no statistically significant effect of NSAIDs on remission during escitalopram (p>0.05). An odds ratio <1 (0.82) suggested that concurrent use of NSAIDs might slightly reduce the chances of remission (Table 3). However, adjustments for age and centre of recruitment have substantially reduced this effect. When all four clinically relevant covariates (age, sex, baseline MADRS score and recruitment centre) were adjusted for, the effect of NSAIDs on remission was reversed with an odds ratio >1 (Table 3). These results suggest that the weak association between NSAIDs and remission is likely to reflect confounding rather than a causal effect of NSAIDs. Similar results were obtained with the two secondary outcome measures (Table 3).
Table 3. Effect of non-steroidal anti-inflammatory drugs (NSAIDs) and analgesics on rates of remission during treatment with escitalopram and nortriptyline in the GENDEP study and effects of potential confounders: results of unadjusted and adjusted logistic regression analyses
BDI, Beck Depression Inventory; CI, confidence interval; GENDEP, Genome-based Therapeutic Drugs for Depression study; HAMD17, 17-item Hamilton Rating Scale for Depression; MADRS, Montgomery–Åsberg Depression Rating Scale.
Odds ratio (OR) <1 means that concurrent NSAIDs reduce the likelihood of remission; OR >1 means that concurrent NSAIDs increase the likelihood of remission. p value <0.05 is given in bold to indicate nominal statistical significance.
Next, we explored the relationship between concurrent use of NSAIDs and remission on the primary outcome measure (MADRS ⩽10) among nortriptyline-treated individuals. Of the 30 individuals using NSAIDs, seven (23%) achieved remission. Of the 323 individuals who were not taking NSAIDs, 100 (31%) achieved remission with up to 12 weeks of nortriptyline treatment. Univariate logistic regression with no covariates revealed no statistically significant effect of NSAIDs on remission during nortriptyline treatment (p>0.05). As for escitalopram, the unadjusted odds ratio was <1 (0.68) and inclusion of covariates further reduced this non-significant effect (albeit to a smaller degree than for escitalopram; Table 3). Among nortriptyline-treated individuals, the secondary outcome measures gave somewhat different results, with NSAIDs slightly increasing the likelihood of remission. However, only one analysis (BDI, adjusted for all covariates) indicated a nominally statistically significant effect (Table 3).
Discussion
Antidepressants and anti-inflammatory drugs are among the most frequently used types of medication and are often taken concurrently (Vandraas et al. Reference Vandraas, Spigset, Mahic and Slordal2010; Gu et al. Reference Gu, Dillon and Burt2010). Therefore, the recent report suggesting that NSAIDs may interfere with the mechanism of action of the most commonly used type of antidepressant drugs merits attention and scrutiny (Warner-Schmidt et al. Reference Warner-Schmidt, Vanover, Chen, Marshall and Greengard2011). We have attempted to replicate the reported attenuation of SRI antidepressant efficacy with concurrent NSAID use and to extend the investigation to tricyclic antidepressants in a large sample of antidepressant-treated individuals with MDD and prospectively recorded concurrent medication use. Our results show that after adjustment for relevant covariates, antidepressants are equally effective in individuals who use NSAIDs and in those who take no such concurrent medication.
In spite of adequate power to detect clinically meaningful effects, we found no significant effect of concurrent NSAIDs or analgesics on the efficacy of SRI antidepressants. In addition, we have probed the possibility of this drug interaction in several ways. First, we explored jointly and separately the effects of NSAIDs and analgesics on antidepressant efficacy. We found that neither NSAIDs nor analgesics significantly changed the efficacy of the SRI escitalopram. Second, we explored the effect of dose. NSAIDs and analgesics are often used as a one-off medication or are used in small doses for other indications. It is likely that such small or irregular doses could have little effect and may dilute the effect of adequately dosed NSAIDs or analgesics in the overall analysis. Therefore, we have run a sensitivity analysis to explore antidepressant efficacy in individuals who took therapeutically significant amounts of NSAID and/or analgesics. The sensitivity analyses confirmed that NSAIDs and/or analgesics do not affect the efficacy of SRI antidepressants. Third, we controlled for a range of clinical covariates that could confound the relationship between NSAID and antidepressant efficacy. We found that older individuals more often took NSAIDs and analgesics. Older individuals also responded less well to antidepressants (Uher et al. Reference Uher, Maier, Hauser, Marusic, Schmael, Mors, Henigsberg, Souery, Placentino, Rietschel, Zobel, Dmitrzak-Weglarz, Petrovic, Jorgensen, Kalember, Giovannini, Barreto, Elkin, Landau, Farmer, Aitchison and McGuffin2009a ). Consequently, the inclusion of age as a covariate further reduced the small and non-significant effect of NSAIDs and/or analgesics on antidepressant efficacy. From these analyses, we conclude that there is unlikely to be a robust and clinically meaningful effect of NSAIDs or analgesics on the efficacy of SRI antidepressants and that a clinical recommendation to avoid concurrent use of NSAIDs and SRI antidepressants on the basis of efficacy would be premature.
The lack of interaction between NSAIDs and SRI antidepressants in terms of efficacy should not diminish the concerns about the safety of the concurrent use of NSAIDs and antidepressants and the need for gastroprotective measures to minimize the risk of gastrointestinal bleeding (de Jong et al. Reference de Jong, van den Berg, Tobi and de Jong-van den Berg2003; Loke et al. Reference Loke, Trivedi and Singh2008). In the GENDEP study, there were no reports of adverse reactions involving gastrointestinal bleeding, but our sample is too small to assess safety of this combination. Therefore it is important to consider the risk of gastrointestinal bleeding when administering NSAIDs and SRI antidepressants concurrently. This concern does not apply to paracetamol, which does not increase the risk of gastrointestinal bleeding.
We further extended the examination of concurrent NSAIDs and analgesics effect to the efficacy of nortriptyline, a tricyclic antidepressant. Most analyses indicated that nortriptyline is similarly effective in subjects taking NSAIDs and/or analgesics as in those without such concurrent medication. However, secondary analyses of self-reported outcomes indicated that concurrent analgesics and/or NSAIDs may slightly improve the efficacy of nortriptyline. Since this was a small effect and only emerged in secondary analyses, it should be treated with caution unless replicated. The possibility that anti-inflammatory agents could selectively enhance the efficacy of one type of antidepressant but attenuate the efficacy of other types of antidepressants is intriguing, but there is little support for such differential effects: only one secondary outcome measure suggested that there is an interaction between concurrent NSAIDs or analgesics and antidepressant type. Such findings have no clinical implication at present but raise a challenge for future studies to explore the interaction of different types of antidepressants, analgesics and anti-inflammatory drugs.
It has been proposed that inflammation and inflammatory cytokines play important part in the causation and maintenance of depression (Miller et al. Reference Miller, Maletic and Raison2009; Hayley, Reference Hayley2011; Li et al. Reference Li, Soczynska and Kennedy2011). The previously reported negative effect of anti-inflammatory drugs on antidepressant efficacy (Warner-Schmidt et al. Reference Warner-Schmidt, Vanover, Chen, Marshall and Greengard2011) would appear to contradict the role of inflammation in the maintenance of depression. Our results suggest that the relationship between NSAID use and antidepressant treatment outcome is due to confounding by age and other factors rather than a direct effect of NSAIDs on antidepressant efficacy. This finding is consistent with the cytokine/inflammation hypothesis of depression in that older age and physical illness and not anti-inflammatory treatment per se are related to poor outcomes. An effect in the opposite direction found among nortriptyline-treated individuals suggests that NSAIDs may slightly augment antidepressant action.
The present findings have to be evaluated with respect to several limitations. First, since the relatively low rates of NSAID use among GENDEP study participants limit the statistical power of this investigation, we are not able to rule out effects of small size. However, the escitalopram-treated group in the GENDEP study is large enough to detect clinically meaningful effects. Second, while we were able to explore the effect of dose and a number of meaningful clinical covariates, other potentially relevant factors have not been measured. For example, chronic painful symptoms could influence both NSAID use and treatment outcome (Leuchter et al. Reference Leuchter, Husain, Cook, Trivedi, Wisniewski, Gilmer, Luther, Fava and Rush2010), but the lack of a specific measure of these symptoms precluded a full exploration in the GENDEP sample. The interplay between NSAID use and these additional factors may require a separate investigation.
We conclude that the SRI antidepressant escitalopram and the tricyclic antidepressant nortriptyline are as effective in treating depression in individuals who take concurrent NSAIDs or analgesics as they are in other individuals with depression. There is no need to avoid the combination of NSAIDs and antidepressants for reasons of efficacy.
Acknowledgements
The GENDEP project was funded by the European Commission Framework 6 grant, EC Contract Ref.: LSHB-CT-2003-503428. Lundbeck provided nortriptyline and escitalopram for the GENDEP study. GlaxoSmithKline and the UK National Institute for Health Research of the Department of Health contributed to the funding of the sample collection at the Institute of Psychiatry, London. The sponsors had no role in the design and conduct of the study, in data collection, analysis, interpretation or writing the report. Dr Uher is supported by a grant from the Innovative Medicines Initiative of the European Commission (Grant Agreement no. 115008).
Declaration of Interest
N. Henigsberg has participated in clinical trials sponsored by pharmaceutical companies including GlaxoSmithKline and Lundbeck, and has received honoraria for participating in expert panels from pharmaceutical companies including Lundbeck. D. Souery is a member of national advisory boards for AstraZeneca, Bristol–Myers Squibb, Eli Lilly and Lundbeck.
