Use of serotonin reuptake inhibitors in patients with psoriasis is associated with a decreased need for systemic psoriasis treatment: a population-based cohort study

Authors


Abstract

Objective

To investigate whether psoriasis is affected by the use of serotonin reuptake inhibitors (SSRIs).

Design

A population-based cohort study.

Setting

The general adult population with plaque psoriasis in Sweden between 1997 and 2006.

Subjects

A total of 69 830 patients with plaque psoriasis were identified in the National Patient Register. Whether study subjects were exposed to SSRIs was identified through the Swedish Prescribed Drug Register. The SSRI-exposed subjects (= 1282) had a prescription for SSRIs dispensed twice during 6 months at a Swedish pharmacy between 1 July 2006 and 1 April 2008, with a wash-out period of 1 year or longer. The reference subjects (= 1282), who were not exposed to SSRIs, were matched for age, county of residence, sex, psoriasis severity and seasonal variation.

Main outcome measure

Change in psoriasis severity defined by switching between nonsystemic and systemic psoriasis treatments 6 months after exposure to SSRIs.

Results

The risk of switching from nonsystemic to systemic psoriasis treatments was significantly decreased in the SSRI-exposed group (odds ratio 0.44, 95% confidence interval 0.28–0.68).

Conclusion

SSRI use in patients with psoriasis is associated with a decreased need for systemic psoriasis treatment.

Introduction

Psoriasis is a chronic inflammatory skin disease that affects 2–3% of the adult population in Sweden [1]. In addition, psoriasis is associated with an increased risk of depressive disorders [2]. Furthermore, chronic stress, which has an impact on the serotonergic system, can aggravate psoriasis [3].

Serotonin reuptake inhibitors (SSRIs) are widely used in the treatment for depressive disorders and chronic stress. Case reports of the effect of SSRIs in patients with psoriasis have shown conflicting results; an improvement of psoriasis after use of paroxetine was demonstrated in one study [4], whereas in three studies, a flare-up or worsening of psoriasis followed treatment with fluoxetine or paroxetine [5-7]. SSRIs have been reported to have an anti-inflammatory effect in patients with and animal models of arthritis [8].

We have previously shown that the expression of the target protein for SSRIs, serotonin transporter protein, may be increased in the skin in patients with psoriasis [9]. We also found that this protein colocalized with caspase-3, an apoptosis-regulating protein [9]. These findings suggest that the serotonergic system may be regulating the process of inflammation observed in psoriasis.

However, the possible biological effects of SSRIs on psoriatic inflammation have not been fully investigated. We have conducted a large population-based cohort study to determine the effect of SSRIs in psoriasis and evaluate the hypothesis that SSRI use may decrease the severity of psoriasis.

Methods

We conducted a population-based cohort study in Sweden using nationwide prospectively recorded registries. The study was completed and reported in accordance with the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) recommendations [10].

Ethics

The study was approved by a regional ethics review board in Stockholm, Sweden. Informed patient consent was not required for ethical approval.

Data sources and study participants

Unique personal identification numbers (PINs), assigned to every Swedish resident since 1947 [11], allowed study subject identification and record linkage of data recorded prospectively with very few losses to follow-up, thus minimizing the risk of recall bias. The National Patient Register contains individual-based information on inpatient care, recorded by county since 1964 and nationwide since 1987. Specialized outpatient care data have been recorded since 2001 [12]. Information on diagnoses and surgical procedures is recorded for every hospital discharge or outpatient care visit, according to the International Classification of Diseases (ICD) versions 7–10 [13]. The source population was all Swedish residents ever diagnosed with plaque psoriasis by dermatologists (ICD codes L40.0, L40.5 and L40.8–9) and registered in the National Patient Register (inhospital care during 1997–2006 and outpatient care during 2001–2006) [12]. In total, 69 830 patients with plaque psoriasis were identified.

Serotonin reuptake inhibitors-exposed cohort

The study cohort of patients with psoriasis and also exposure to SSRI treatment was identified by linkage to the Swedish Prescribed Drug Register. Since 1 July 2005, all dispensed drugs are registered with the Swedish PINs [14]. All drugs are classified according to the anatomical therapeutic chemical classification system. Drug utilization is measured in terms of number of prescriptions, defined daily doses (DDDs) and costs. The DDD is the assumed average maintenance dose per day for a drug used for its main indication in adults [15]. The Prescribed Drug Register is complete for the entire Swedish population (patient identity data are missing for <0.3% of all items) [14]. From the database, we obtained information about dispensed substances, date of dispensing and dosage of prescribed drug measured as DDD. The DDDs for each SSRI are shown in Table 1. We defined days of therapy with the DDD at three levels: low (<40% of days with 1 DDD/day), medium (40–90% of days with 1 DDD/day) and high (>90% of days with 1 DDD/day).

Table 1. DDDs for different SSRI drugs
SSRIDDD
Fluoxetine20 mg
Citalopram20 mg
Paroxetine20 mg
Sertraline50 mg
Fluvoxamine100 mg
Escitalopram10 mg

Inclusion criteria

The cohort of SSRI-exposed subjects (= 1282) had a prescription dispensed for fluoxetine, citalopram, paroxetine, sertraline, fluvoxamine or escitalopram twice during a period of 6 months at a Swedish pharmacy between 1 July 2006 and 1 April 2008. To ensure incident SSRI treatment, a wash-out period of at least 1 year without any dispensed SSRI drugs was required for all patients. Cohort entry was defined as the date of dispensing of a prescription for an SSRI. Psoriasis disease severity was classified according to the last psoriasis treatment dispensed (systemic or nonsystemic) up to 6 months prior to cohort entry. Severe psoriasis was defined as use of systemic treatments (methotrexate, cyclosporine, acitretin and biological agents, i.e. etanercept, infliximab, efalizumab and adalimumab), and mild psoriasis was defined as no use of these treatments. This method of psoriasis classification has been used and validated previously in several studies [2, 16-19].

Reference cohort

The reference cohort included patients from the source population with psoriasis but not exposed to SSRIs (= 1282). Patients in the study and reference cohorts were matched for age, sex, county of residence, psoriasis severity (systemic or nonsystemic psoriasis treatment) and seasonal variation (defined as the date of dispensed psoriasis drug, matched ± 2 months).

Outcome

The main outcome measure was change in psoriasis severity assessed as switching psoriasis therapy from systemic to nonsystemic treatments or vice versa 6 months after exposure to an SSRI. Outcome was measured between 6 and 12 months after cohort entry, defined as either first or absence of psoriasis drug dispensing.

Statistical analysis

As a measure of change in psoriasis severity, we recorded whether or not patients in the two cohorts, exposed and not exposed to SSRI treatment, switched from systemic to nonsystemic psoriasis treatments and vice versa (binary outcome and binary predictor). Conditional logistic regression was used to determine the odds ratio (OR) and 95% confidence interval (CI) of switching psoriasis treatments after exposure to an SSRI. The OR was adjusted for age, sex and county of residence. Age was divided into 10-year strata and counties in Sweden divided into three regions (one rural and two urban). We also performed a sensitivity analysis of the DDD of SSRI drugs at three levels (low, medium and high) (Table 4).

In addition, the use of phototherapy and geographical distribution was examined in both cohorts. Treatments with broadband and narrowband UVB, UVB plus UVA and PUVA were identified from the National Patient Register.

SAS 9.2 (SAS Institute Inc, Cary, NC, USA) was used for statistical analyses.

Results

Overall, 64% of patients were women. The mean age was 55 years amongst both the exposed and the reference cohorts. At study entry, 89% of subjects were classified as having mild psoriasis (Table 2).

Table 2. Characteristics of the two cohorts
CharacteristicsSSRI-exposed groupReference group
= 1282= 1282
  1. IQR, interquartile range.

Sex, n (%)
Female482 (38)482 (38)
Male800 (62)800 (62)
Age group, n (%)
0–19 years28 (2)28 (2)
20–39 years265 (21)265 (21)
40–59 years442 (35)442 (35)
60–79 years421 (33)421 (33)
≥80 years126 (10)126 (10)
Median age, years (IQR)57 (41–69)56 (41–68)
Mean age, years55.054.6
Psoriasis severity, n (%)
Mild (no systemic treatment)1136 (89)1136 (89)
Severe (systemic treatment)146 (11)146 (11)

Amongst patients with mild psoriasis, 29 switched from nonsystemic to systemic treatments after exposure to an SSRI, compared to 64 patients in the reference group (OR 0.44, 95% CI 0.28–0.68; Table 3). Amongst study subjects classified with severe psoriasis, 42 patients switched from systemic to nonsystemic treatment after SSRI use compared to 37 patients in the reference group (OR 1.20, 95% CI 0.71–2.02; Table 3). These findings indicate that SSRI use may have a protective effect in psoriasis.

Table 3. Number of patients and risk of switching psoriasis treatments before versus after SSRI exposure
Treatment at baselineExposed cohortReference cohortOR (95% CI)
(= 1282)(= 1282)
  1. The risks are shown as odds ratios (ORs) with 95% confidence intervals (CIs), adjusted for age, sex and county of residence.

  2. NS, nonsystemic; S, systemic.

Nonsystemic, n (%)
No switch1107 (97.5)1072 (94.4)Reference
Switch (NS→S)29 (2.6)64 (5.6)0.44 (0.28–0.68)
Total11361136 
Systemic, n (%)
No switch104 (71.2)109 (74.7)Reference
Switch (S→NS)42 (28.8)37 (25.3)1.20 (0.71–2.02)
Total146146 

To assess whether this effect might be dose dependent, we investigated the relation between number of days with the DDD of SSRI therapy and change in psoriasis severity (Table 4). Amongst those with more than 90% of days with 1 DDD/day of prescribed SSRI and mild psoriasis, 23 patients switched treatments after SSRI use, compared to 49 patients in the reference group (OR 0.45, 95% CI 0.27–0.75). The protective effect of SSRIs was also observed amongst those with 40–90% of days with 1 DDD/day of prescribed SSRI; four patients with mild psoriasis switched treatments after SSRI use compared to 13 patients in the reference group (OR 0.30, 95% CI 0.10–0.93). Finally, we also investigated whether this protective effect of SSRIs was gender dependent. When stratified for sex, the OR of switching from nonsystemic to systemic treatment after SSRI use was 0.51 for men (95% CI 0.28–0.96) and 0.37 for women (95% CI 0.19–0.71). We found no major differences between the exposed and reference cohorts in the use of phototherapy or the geographical distribution of patients (data not shown).

Table 4. Number of patients and risk of switching psoriasis treatments before versus after SSRI exposure, stratified according to level of SSRI use (high, medium and low)
Treatment at baselineExposed cohortReference cohortOR (95% CI)
(= 1282)(= 1282)
  1. a

    Only six patients in the SSRI-exposed group and six in the reference group received systemic treatments (all nonswitchers).

  2. The risks are shown as odds ratios (ORs) with 95% confidence intervals (CIs), adjusted for age, sex and county of residence.

  3. NS, nonsystemic; S, systemic.

High level of SSRI use (>90% of days with 1 DDD/day)
Nonsystemic, n (%)
No switch820 (97.3)794 (94.2)Reference
Switch (NS→S)23 (2.7)49 (5.8)0.45 (0.27–0.75)
Total843843 
Systemic, n (%)
No switch75 (70.8)79 (74.5)Reference
Switch (S→NS)31 (29.3)27 (25.5)1.21 (0.66–2.24)
Total106106 
Medium level of SSRI use (40–90% of days with 1 DDD/day)
Nonsystemic, n (%)
No switch234 (98.3)225 (94.5)Reference
Switch (NS→S)4 (1.7)13 (5.5)0.30 (0.10–0.93)
Total238238 
Systemic, n (%)
No switch23 (67.7)24 (70.6)Reference
Switch (S→NS)11 (32.4)10 (29.4)1.41 (0.42–3.12)
Total3434 
Low level of SSRI use (<40% of days with 1 DDD/day) a
Nonsystemic, n (%)
No switch53 (96.4)53 (96.4)Reference
Switch (NS→S)2 (3.6)2 (3.6)1.00 (0.13–7.64)
Total5555 

Discussion

The results of this study demonstrate that patients with mild psoriasis have a decreased need for systemic psoriasis treatment after starting therapy with an SSRI. Furthermore, in patients with severe psoriasis, the need for systemic psoriasis treatment is likely to be reduced with concomitant SSRI use.

The finding that the use of an SSRI results in a decreased need for systemic psoriasis treatment is intriguing. It is possible that treatment of clinical depression could have a direct effect on psoriasis symptoms or an indirect effect whereby improvement in the exposed subjects' mood disorder will increase their ability to comply with nonsystemic psoriasis treatments. Alternatively, a direct biological anti-inflammatory effect of SSRIs, either involving serotonin signalling pathways or independent of the serotonergic system, may explain the present results [8, 20]. Studies in human and animal models have indicated that some SSRIs may have an anti-inflammatory effect [8, 21].

Strengths and limitations

Particular strengths of this study include its thorough design with a well-defined study base, which minimizes bias whilst maximizing the generalizability of the results. There were two strict criteria for exposure to an SSRI. First, to ensure incident SSRI treatment, only patients who had undergone a prior wash-out period of SSRIs of at least 1 year were included. Secondly, to reflect clinically relevant SSRI treatment, we included only patients with at least two dispensed prescriptions of an SSRI within 6 months, as the biological effect of SSRIs will take approximately 4–6 weeks to develop. Patients who experienced side effects after a few weeks were not included in this study, because the available register information is not adequate to determine whether these patients stopped treatment because of the side effects or for other reasons. Further strengths are that the diagnosis of plaque psoriasis is made by dermatologists, minimizing the possibility of misclassification of disease. Both databases have high internal validity with identity data missing for <0.3% of all items in the Swedish Prescribed Drug Register [14] and 0.5% in the National Patient Register. Furthermore, only around 1% of visits or admittance to in- and outpatient care lack data regarding diagnosis in the National Patient Register [12].

The reference group is strictly matched to allow the effect of SSRI exposure on psoriasis severity to be determined; however, this does not allow any other conclusions to be drawn. Seasonal variation is an important matching factor; not only does it appear from clinical experience in Sweden that psoriasis improves during summer, seasonal variation in the binding potential for serotonin transporter protein has also been shown [22]. However, information on two other important confounders, smoking and alcohol, is lacking.

There are some important limitations to consider. First, using systemic psoriasis treatment as a proxy for severe psoriasis, misclassification of psoriasis severity may be possible. Patients with severe psoriasis may not receive systemic treatment and therefore may be classified incorrectly as having mild psoriasis. The possibility of misclassification is unlikely to differ between study groups and thus may be regarded as nondifferential. A second limitation is that according to the study design, patients who experience side effects of SSRIs and stop treatment after 1–2 months are not included. It would be interesting if this group of patients experienced a flare-up of psoriasis as a side effect; this would cause an overestimation of the protective effect of SSRIs in psoriasis. However, this exclusion criterion can also be considered a study strength as it is clear that the subjects in the two cohorts tolerated SSRIs and are most likely to benefit from any possible treatment effects. Thirdly, we lacked information on comorbidities and concomitant medication in our two cohorts. Because of differences in comorbidities and/or concomitant medication, patients exposed to SSRIs might not receive systemic psoriasis treatment, leading to an overestimation of the protective effect of SSRIs.

Finally, delay by dermatologists in changing psoriasis treatments even though patients may experience a change in psoriasis severity is a further possible limitation to consider. Phototherapy is a fairly common systemic treatment for psoriasis in Sweden. However, this treatment was not included as a marker of severe psoriasis because it is not recorded in the Swedish Prescribed Drug Register. Phototherapy use is included in the National Patient Register, but there may be variations between counties in the reporting of such outpatient treatment procedures. Therefore, we examined the use of phototherapy as well as geographical distribution of patients and found no major differences between the SSRI-exposed and reference groups (data not shown).

In the present study, the protective effect of SSRI use was seen even at low drug doses (40–90% of days with 1 DDD/day of prescribed SSRI). This protection remained and did not improve amongst patients prescribed medium (40–90%) or high doses (>90% of days with 1 DDD/day of prescribed SSRI), indicating a threshold effect of SSRIs rather than a dose-dependent effect.

In conclusion, use of SSRIs in psoriasis is associated with a decreased need for systemic psoriasis treatment. To our knowledge, this finding has not been previously reported. Therefore, we recommend future studies to investigate the potential use of SSRIs as a novel and easily accessible treatment option for psoriasis.

Conflict of interest statement

None of the authors has any conflict of interest to declare.

Acknowledgements

Financial support was received from the Swedish Psoriasis Association. We thank Dr Fredrik Granath (Clinical Epidemiology Unit, Department of Medicine Solna, Karolinska Institutet) for statistical advice and Professor Mona Ståhle (Dermatology and Venereology Unit, Department of Medicine Solna, Karolinska Institutet) for valuable comments on the manuscript.

Ancillary