Psoriasis and smoking: a systematic review and meta-analysis

Authors


  • Funding sources No external funding.
  • Conflicts of interest None declared.

Summary

Psoriasis is an inflammatory skin disease associated with increased cardiovascular comorbidity. Smoking is associated with an increased risk of cardiovascular disease, and prior studies have suggested that patients with psoriasis are more likely to be active smokers. Smoking may also be a risk factor in the development of psoriasis. We conducted a systematic review and meta-analysis to assess the prevalence of smoking among patients with psoriasis, and we reviewed the contribution of smoking to the incidence of psoriasis. A total of 25 prevalence and three incidence studies were identified. The meta-analysis of prevalence studies included a total of 146 934 patients with psoriasis and 529 111 patients without psoriasis. Random effects meta-analysis found an association between psoriasis and current smoking [pooled odds ratio (OR) 1·78, 95% confidence interval (CI) 1·52–2·06], as well as between psoriasis and former smoking (pooled OR 1·62, 95% CI 1·33–1·99). Meta-regression analysis did not reveal any sources of study heterogeneity, but a funnel plot suggested possible publication bias. A subset of studies also examined the association between moderate-to-severe psoriasis and smoking, with a pooled OR of 1·72 (95% CI 1·33–2·22) for prevalent smoking. The three incidence studies found an association between smoking and incidence of psoriasis, with a possible dose-effect of smoking intensity and duration on psoriasis incidence. These findings suggest that smoking is an independent risk factor for the development of psoriasis, and that patients with established psoriasis continue to smoke more than patients without psoriasis.

Psoriasis is a chronic, immune-mediated skin disease that affects approximately 2–4% of the world population.[1, 2] Psoriasis is associated with a number of cardiovascular risk factors, including hypertension,[3, 4] diabetes,[5] obesity,[6] metabolic syndrome[7] and dyslipidemia.[8] Smoking is a well-established risk factor for the development of cardiovascular diseases such as coronary artery disease, cerebrovascular disease and myocardial infarction.[9] Furthermore, smoking is also associated with certain types of malignancies, gastrointestinal disease and chronic obstructive pulmonary disease.[10] In a 2012 study that assessed the smoking status among 3 billion individuals from 16 different countries, 661 million (22%) were smokers.[11] Previous studies have suggested that smoking is associated with increased severity of psoriasis and diminished treatment responses among patients with psoriasis.[12] In this systematic review and meta-analysis, we characterize the prevalence of smoking among patients with psoriasis as well as the contribution of smoking to psoriasis development.

Methods

We performed a systematic review using the MEDLINE, Embase and Cochrane Central Register databases with the following search criteria: [‘Psoriasis’(MeSH) AND ‘Smoking’(MeSH)]. Our search was limited to English-language, human-subject studies published between 1 January 1980 and 15 June 2013. We also manually searched the references of retrieved articles for any additional studies not identified in the initial searches. To be eligible for study inclusion, original studies needed to fulfil the following criteria: case–control, cross-sectional, cohort or nested case–control design; evaluation of smoking in conjunction with psoriasis; and analyses that compared patients with controls. Specifically, the studies had to evaluate the prevalence or incidence of psoriasis in patients who smoked as defined by either patient self-report or medical chart review.

A total of 128 articles were identified from the initial search, and seven additional studies were found from manual review (Fig. 1). After reviewing all abstracts, 74 full-text articles were further evaluated. Ten of these studies were excluded because they were reviews; five did not include a control group; 15 did not measure the association between psoriasis and smoking; 11 were commentaries, letters or editorials; three lacked adequate data for meta-analysis (such as odds ratio calculations); one study included only palmopustular psoriasis,[13] and one study included only psoriatic arthritis.[14] After these exclusions, 28 studies were included in the meta-analysis (25 prevalence studies and three incidence studies).

Figure 1.

Study selection. PPP, palmoplantar pustulosis; Ps, psoriasis; PsA, psoriatic arthritis.

Two reviewers (C.T.H., A.W.A.) independently extracted the data and performed the systematic review, and any disagreements were adjudicated by consensus. The Meta-Analysis of Observational Studies in Epidemiology (MOOSE) guidelines were used to guide analysis.[15] For each study included, we recorded the study year, country in which the study population lived, setting in which the study took place, study design, numbers of case and control subjects, age, sex, whether the results were adjusted for comorbidities, data collection processes (prospective or retrospective), whether the results were a primary or secondary analysis of the publication, and whether psoriasis disease severity was assessed. To measure study quality, we used a previously validated 7-point scale, with values of 0 or 1 assigned to study design, assessment of exposure, assessment of outcome, control for confounding, evidence of bias, and assessment of psoriasis severity. Studies with a score of 0–3 were categorized as lower quality, while studies with scores of 4–6 were categorized as higher quality.[16]

Most of the studies were of case–control design and reported adjusted odds ratios (ORs). Studies that reported prevalence (= 25) were analysed separately from studies that reported incidence (= 3). To estimate the pooled OR and pooled relative risk (RR), we used fixed-effects and random-effects models of DerSimonian and Laird.[17] Multivariate adjusted ORs were used for meta-analysis when available; otherwise, the unadjusted ORs were used in studies where multivariate adjustment had not been performed. All prevalence studies reported current smoking status, while some differentiated prior smoking from current smoking. A subset of studies also differentiated mild from moderate-to-severe psoriasis, and they assessed the association of smoking with psoriasis severity. Study heterogeneity was assessed using the I2 statistic. Due to significant between-study heterogeneity, reported pooled ORs are based on random-effects modelling.

Publication bias was assessed by visual inspection of a funnel plot of the study size vs. standard error, with formal statistical testing performed using the Begg adjusted rank correlation test.[18, 19] To explore sources of study heterogeneity, we performed a meta-regression using prespecified variables and random-effects meta-analysis. Prespecified sources of heterogeneity included study country, subject location (ambulatory or inpatient), multivariate adjustment for confounders, prospective vs. retrospective study design, ascertainment of prevalence vs. incidence, primary vs. secondary analysis, ascertainment of psoriasis disease severity, measure of outcome, and study quality (0–3 vs. 4–6). All analyses were performed using STATA version 11.2 (STATA Corp, College Station, TX, U.S.A.).

Results

A total of 28 studies assessed either the prevalence of smoking among patients with psoriasis or the incidence of psoriasis among smokers.[3, 15, 20-46] Study population characteristics are outlined in Table 1 for prevalence studies and in Table 4 for incidence studies.

Table 1. Psoriasis and prevalence of smoking: study population characteristics
StudyStudy settingStudy designTotal n of patientsMean age (years)
No psoriasis (controls)Psoriasis (cases)No psoriasis (controls)Psoriasis (cases)
  1. NR, not reported; PPP, palmoplantar pustulosis; Ps, psoriasis. aSmoking was the primary study variable.

Armesto, 2012[22]Spain; outpatientCase–control66166147·447·4
aShapiro, 2012[39]Israel; inpatientCase–control1079107968·768·6
Armstrong, 2011[51]U.S.A.; outpatientCase–control241883561·761·5
aAl-Mutairi, 2010[21]Kuwait; outpatientCase–control1835183552·752·3
aGerdes, 2010[26]Germany; inpatientCase–control69421097 (severe Ps)NR49·7
Takahashi, 2010[46]Japan; outpatientCase–control15415157·253·1
aDriessen, 2009[25]Netherlands; outpatientCross-sectional396107 (severe Ps)51·248·5
aJankovic, 2009[29]Montenegro; outpatientCase–control20011046·145·5
aJin, 2009[30]China; outpatientCase–control17817825·426·2
aWolk, 2009[41]Sweden; outpatientCase–control3733734646
aWolkenstein, 2009[42]France; outpatientCase–control106835651·651·6
Xiao, 2009[43]China; outpatientCross-sectional1521309243·6

Mild Ps: 43·6

Severe Ps: 46·3

aBo, 2008[23]Norway; outpatientCross-sectional17 6031144NRNR
Naldi, 2008[38]Italy; outpatientCase–control690560NRNR
Cohen, 2007[24]Israel; outpatientCase–control664334047·047·7
aNeimann, 2006[36]U.K.; outpatientCross-sectional

Mild Ps: 465 252

Severe Ps: 14 065

Combined: 479 317

Mild Ps: 127 706

Severe Ps: 3854

Combined: 131 560

Mild Ps: 45·7

Severe Ps: 46·3

Mild Ps: 46·4

Severe Ps: 49·8

Sommer, 2006[40]Germany; inpatientCross-sectional104458158·554·4
aHerron, 2005[27]U.S.A.; outpatientCross-sectional408055745·749·8
aNaldi, 2005[33]Italy; outpatientCase–control690560Median age 36Median age 38
aZheng, 2004[45]China; outpatientCross-sectional333189NRNR
aZhang, 2002[44]China; outpatientCase–control789789

Male: 32·2

Female: 29·3

Male: 35·4

Female: 31·2

aNaldi, 1999[35]Italy; outpatientCase–control616404Median age 36Median age 35
aPoikolainen, 1994[37]Finland; outpatient and inpatientCase–control10855NRNR
aMills, 1992[20]Wales; outpatient and inpatientCase–control106106NR43·3
aNaldi, 1992[34]Italy; outpatientCase–control267215NRNR

Prevalence of smoking among patients with psoriasis

A total of 25 studies assessed the prevalence of smoking among 146 934 patients with psoriasis and 529 111 patients without psoriasis (Tables 1 and 2). All but three studies reported a statistically significant relationship between psoriasis and an increased prevalence of smoking. Twenty-two of the studies were conducted among outpatients with psoriasis, and the studies were conducted in diverse locations including Europe, Asia and the Middle East. Fourteen of the studies were deemed high quality (quality score of 4–6), with the major quality difference being the level of covariate adjustment performed within each study. Meta-analysis of these studies revealed a significant association between psoriasis and current smoking, with a pooled OR of 1·78 (95% CI 1·53–2·06, Fig. 2). Meta-regression of current smoking among patients with psoriasis did not reveal any statistically significant sources of heterogeneity. However, studies of higher quality and multivariate adjustment generally reported lower ORs for the association between psoriasis and smoking, suggesting that studies without multivariate adjustment retained some residual confounding (Table 3). Visual inspection of a funnel plot suggested possible publication bias, with an absence of small, negative studies (Fig. 3). Formal assessment of publication bias with the Begg adjusted rank correlation test confirmed likely publication bias (= 0·002).

Table 2. Psoriasis and prevalence of smoking
StudyStudy periodSmoking ascertainmentNumber of patients (%) who smokeMeasure of association (95% CI)
No psoriasis (controls)Psoriasis (cases)
  1. NR, not reported; OR, odds ratio; PASI, Psoriasis Area and Severity Index; Ps, psoriasis. aSmoking was the primary study variable.

Armesto, 2012[22]NRQuestionnaire; manual chart review190 (28·7)232 (35·1)Current smoker: OR 1·34 (1·06–1·70)
aShapiro, 2012[39]1993–2006ICD-9Current: 174 (16·1)Current: 234 (21·7)

Current smoker: OR 1·44 (1·16–1·79)

Adjusted OR 1·38 (1·10–1·73)

Armstrong, 2011[51]2004–2009Manual chart review

Current: 260 (12)

Former: 745 (36)

Current: 113 (14)

Former: 310 (39)

Current smoker: OR 1·38 (1·07–1·77)

Adjusted OR 1·43 (1·01–2·02)

Former smoker: OR 1·59 (1·10–1·58)

Adjusted OR 1·22 (0·95–1·55)

aAl-Mutairi, 2010[21]2003–2007Manual chart reviewCurrent: 371 (32·5)Current: 762 (51·3)Current smoker: OR 2·19 (1·86–2·57)
aGerdes, 2010[26]NRManual chart review2295 (32·2)509 (45·0)

Current smoker: OR 2·08 (1·81–2·39)

Adjusted OR 2·05 (1·77–2·39)

Takahashi, 2010[46]2006–2008Manual chart reviewNRNRCurrent smoker: OR 1·77 (1·12–2·78)
aDriessen, 2009[25]NRManual chart review

Current: 114 (28·8)

Former: 233 (58·8)

Current: 50 (46·7)

Former: 81 (75·7)

Current smoker: OR 1·73 (1·08–2·75)

Former smoker: OR 1·92 (1·14–3·22)

aJankovic, 2009[29]2007Interview

Current: 72 (36·0)

Former: 34 (17·0)

Current: 42 (38·2)

Former: 20 (18·2)

Smoking (both current and former): OR 1·09 (0·67–1·76)
aJin, 2009[30]1997–2001InterviewSmoking habit: 31 (17·4)Smoking habit: 57 (32·0)

Current smoking: OR 2·31 (1·32–3·80)

Adjusted OR 2·07 (1·12–3·82)

aWolk, 2009[41]2001–2006Questionnaire82 (22·0)138 (37·0)Current smoking: Adjusted OR 1·7 (1·1–2·6)
aWolkenstein, 2009[42]NRQuestionnaire

Current: 197 (18·6)

Former: 309 (29·2)

Current and former smoker: 506 (47·8)

Current: 72 (20·6)

Former: 125 (35·7)

Current and former smoker: 197 (56·3)

Current or former smoker: OR 1·41 (1·10–1·79)

Adjusted OR 1·46 (1·11–1·94)

Xiao, 2009[43]1999–2007Medical code consistent with smoking241 (15·8)

Mild Ps: 309 (19·1)

Severe Ps: 376 (25·5)

Mild Ps: OR 1·35 (1·01–1·80)

Adjusted OR 1·31 (1·04–1·65)

Severe Ps: OR 1·57 (1·20–2·05)

Adjusted OR 1·42 (1·12–1·79)

aBo, 2008[23]2000–2001QuestionnaireNR

Current: 4739

Former: 5581

Men

Current smoker: OR 1·45 (1·15–1·84)

Adjusted OR 1·49 (1·11–2·00)

Former smoker: OR 1·63 (1·31–2·02)

Adjusted OR 1·67 (1·29–2·16)

Women

Current smoker: OR 1·65 (1·35–2·02)

Adjusted OR 1·48 (1·15–1·91)

Former smoker: OR 1·57 (1·29–1·91)

Adjusted OR 1·55 (1·21–1·99)

Naldi, 2008[38]1988–1997QuestionnaireEver smoked: 350 (50·7)Ever smoked: 357 (63·7)Adjusted OR 1·8 (1·3–2·7)
Cohen, 2007[24]NRReported in database1648 (24·8)118 (34·7)OR 1·6 (1·3–2·0)
aNeimann, 2006[36]1987–2002Medical code

Mild Ps: 98 337 (21·1)

Severe Ps: 3157 (22·5)

Mild Ps: 35 762 (28)

Severe Ps: 1158 (30·1)

Mild Ps: OR 1·40 (1·38–1·43)

Adjusted OR 1·31 (1·29–1·34)

Severe Ps: OR 1·31 (1·20–1·44)

Adjusted OR 1·31 (1·17–1·47)

Sommer, 2006[40]1996–2002Manual chart review219 (21·0)264 (45·4)OR 2·96 (2·27–3·84)
aHerron, 2005[27]NRQuestionnaire530 (13·0)205 (37·0)OR 4·02 (3·31–4·88)
aNaldi, 2005[33]1988–1997Interview

Never smoker: 340 (49·3)

Current, cigarettes per day: < 1–10: 108 (15·7)

11–20: 113 (16·4)

≥ 21: 38 (5·5)

Former: 91 (13·2)

Never smoker: 203 (36·3)

Current, cigarettes per day: < 1–10: 99 (17·7)

11–20: 114 (20·4)

≥ 21: 40 (7·1)

Former: 103 (18·4)

Current smoker, cigarettes per day: < 1–10: Adjusted OR 1·6 (1·2–2·2)

11–20: Adjusted OR 1·7 (1·1–2·5)

≥ 21: Adjusted OR 1·7 (1·0–3·2)

Former smoker: Adjusted OR 1·9 (1·3–2·7)

aZheng, 2004[45]1997–2001Questionnaire

Male: 49 (28·0)

Female: 2 (1·3)

Male: 53 (50·5)

Female: 0 (0)

Male: OR 2·62 (1·53–4·49)
aZhang, 2002[44]1997–2001Questionnaire

Male: 126 (29·1)

Female: 5 (1·4)

Male: 215 (48·9)

Female: 9 (2·6)

Male: OR 2·33 (1·74–3·11)

Female: OR 1·86 (0·56–6·43)

aNaldi, 1999[35]1988–1997Questionnaire

Never smoker: 318 (51·6)

Current, cigarettes per day: ≤ 15: 130 (21·1)

16–24: 64 (10·4)

≥ 25: 21 (3·4)

Former: 83 (13·5)

Never smoker: 154 (38·1)

Current, cigarettes per day: ≤ 15: 95 (23·5)

16–24: 56 (13·9)

≥ 25: 24 (5·9)

Former: 75 (18·5)

Current smoker, cigarettes per day: ≤ 15: Adjusted OR 1·4 (1·0–2·0)

16–24: Adjusted OR 1·7 (1·1–2·7)

≥ 25: Adjusted OR 2·1 (1·1–3·9)

Former smoker: OR 1·8 (1·2–2·6)

Adjusted OR 1·5 (1·1–2·2)

aPoikolainen, 1994[37]1989–1991QuestionnaireNRNR

Before onset of psoriasis (20 cigarettes per day): OR 1·9 (0·8–4·7)

Adjusted OR 3·3 (1·4–7·9)

After onset of psoriasis (20 cigarettes per day): OR 2·0 (0·8–4·7)

Adjusted OR 2·6 (1·1–6·5)

aMills, 1992[20]NRQuestionnaireCurrent: 25 (23·6)Current: 49 (46·2)Current smoker: OR 2·71 (1·44–5·42)
aNaldi, 1992[34]1988–1990Questionnaire

Never smoker: 133 (49·8)

Current, cigarettes per day: < 15 per day: 66 (24·7)

≥ 15 per day: 30 (11·2)

Former: 38 (14·2)

Never smoker: 79 (36·7)

Current, cigarettes per day: < 15 per day: 54 (25·1)

≥ 15 per day: 47 (21·9)

Former: 33 (15·3)

Current smoker, cigarettes per day: < 15: OR 1·4 (0·8–2·3)

Adjusted OR 1·1 (0·6–1·8)

≥ 15: OR 2·6 (1·5–4·6)

Adjusted OR 2·1 (1·1–4·0)

Former smoker: OR 1·5 (0·8, 2·7)

Adjusted OR 1·1 (0·6–2·0)

Table 3. Potential prespecified sources of heterogeneity explored among the studies reporting an association between psoriasis and prevalence of smoking
Prespecified source of heterogeneityNumber of estimatesStratified random-effects meta-analysis OR (95% CI)Meta-regression P-value for heterogeneity
  1. CI, confidence interval; OR, odds ratio.

Study location
U.S.A.22·42 (0·88–6·67)0·3
Europe141·70 (1·44–2·0)
Other91·73 (1·43–2·08)
Source population
Inpatient32·02 (1·39–2·94)0·4
Outpatient221·74 (1·48–2·04)
Statistical adjustment
Not adjusted102·01 (1·54–2·62)0·1
Adjusted151·63 (1·41–1·88)
Study quality
Lower (0–3)111·90 (1·44–2·51)0·2
Higher (4–6)141·68 (1·43–1·97)
Outcome ascertainment
Billing data111·67 (1·40–2·0)0·4
Chart review141·88 (1·48–2·40)
Analysis of outcome
Primary181·82 (1·50–2·21)0·6
Secondary71·67 (1·33–2·10)
Severity of psoriasis
No distinction201·84 (1·55–2·18)0·4
Mild vs. severe51·57 (1·23–2·01)
Figure 2.

Meta-analysis of psoriasis and prevalence of current smoking. CI, confidence interval; OR, odds ratio.

Figure 3.

Funnel plot of psoriasis and prevalence of current smoking. The absence of studies in the lower left corner suggests possible publication bias, with missing studies of smaller cohort and effect size. OR, odds ratio.

The association between psoriasis severity and the prevalence of smoking was addressed in five studies. Patients with moderate-to-severe psoriasis had a pooled OR of 1·72 (95% CI 1·33–2·22) for prevalent smoking, which did not differ significantly from the overall meta-analysis results (Fig. 4). Seven studies also differentiated former smoking from current smoking among patients with psoriasis. Meta-analysis of the association between psoriasis and a history of former smoking found a pooled odds ratio of 1·62 (95% CI 1·33–1·98) for former smoking (Fig. 5).

Figure 4.

Meta-analysis of moderate/severe psoriasis and prevalence of smoking. CI, confidence interval; OR, odds ratio.

Figure 5.

Meta-analysis of psoriasis and former smoking. CI, confidence interval; OR, odds ratio.

Number of cigarettes per day

A total of seven studies quantified the number of cigarettes smoked per day among patients with psoriasis compared with control subjects. One study of 178 patients with psoriasis and 178 controls reported a dose-effect relationship: patients with psoriasis had an OR of 1·81 (95% CI 0·84–3·90) for smoking 1–10 cigarettes per day and an OR of 2·23 (95% CI 1·04–4·75) for smoking > 10 cigarettes per day compared with controls.[30] A self-report questionnaire-based study from Norway found that men who smoked ≥ 20 cigarettes per day were 1·4 times more likely (95% CI 1·06–1·85) to report having psoriasis, compared with those that smoked < 20 cigarettes per day, whereas women who smoked ≥ 20 cigarettes per day were 1·12 times more likely to report having psoriasis, in comparison with those who smoked less frequently.[23] Other studies found similar associations between psoriasis and number of cigarettes smoked per day[20, 33-35, 37] (Table 2). Because these studies used different cut-off points for the number of cigarettes per day (e.g. 10, 15 or 20), meta-analysis could not be performed on the association between psoriasis and the number of cigarettes smoked per day. However, these studies generally support an increased number of cigarettes per day smoked by patients with psoriasis when compared with patients without psoriasis.

Duration and pack-years of smoking

Four studies reported associations between psoriasis and the total number of pack-years smoked. In one study, patients with psoriasis were 1·42 times more likely to have smoked for 1–10 years, compared with those without psoriasis (95% CI 0·67–3·01), and nearly three times more likely to have smoked ≥ 10 years, compared with patients without psoriasis (adjusted OR 2·96, 95% CI 1·30–6·72).[30] Other studies did not find robust differences in smoking duration between patients with and without psoriasis[33-35] (Table 2). In the same study that examined duration of smoking, patients with psoriasis were also 1·73 times more likely to have 1–5 pack-years and 2·32 times more likely to have ≥ 5 pack-years of smoking history, compared with patients without psoriasis (95% CI 0·79–3·83; 95% CI 1·08–5·00, respectively)[30] (Table 2).

Smoking and incident psoriasis

A total of three publications assessed smoking and incident psoriasis (Tables 4 and 5). One study reported results from three populations of patients [the Nurses' Health Study (NHS), the Nurses' Health Study II (NHS II), and the Health Professionals' Follow-Up Study (HPFS)];[31] another publication was an earlier analysis of the NHS II and therefore included the same population of patients.[32] The third publication assessed incident psoriasis in the General Practice Research Database (GPRD), a U.K.-based database of patients from the general community.[28] The pooled data from the NHS, NHS II and HPFS included 185 836 participants with 2410 incident cases of psoriasis, while the GPRD population included 3994 cases of incident psoriasis and 10 000 control subjects.

Table 4. Smoking and incident psoriasis: study population characteristics
StudyStudy settingStudy designTotal number of patientsMean age
Nonsmokers (controls)Smokers (cases)Nonsmokers (controls)Smokers (cases)
  1. NR, not reported. aSmoking was the primary study variable. bThe odds ratio is an unbiased estimator of the incidence ratio with this study design.

aLi, 2012[31]U.S.A.; outpatientsProspective cohortNRNR49·2

Current: 48·7

Former: 50·2

a,bHuerta, 2007[28]U.K.; outpatientProspective cohort, nested case–control analysis10 000 nonpsoriasis controls3994 psoriasis casesNRNR
aSetty, 2007[32]U.S.A.; female outpatientsProspective cohort51 779

Current: 9023

Former: 17 730

35·3

Current: 36·1

Former: 36·6

Table 5. Smoking and incident psoriasis: study outcomes
StudyStudy periodSmoking ascertainmentNonsmokers that developed psoriasis (control) (%)Smokers that developed psoriasis (%)Measure of association (95% CI)
  1. NR, not reported; OR, odds ratio; PsA, psoriatic arthritis; RR, relative risk. aSmoking was the primary study variable. bThe odds ratio is an unbiased estimator of the incidence ratio with this study design.

aLi, 2012[31]1996–2008 (NHS); 1991–2005 (NHS II); 1986–2006 (HPFS)Questionnaire1124

Current: 309

Former: 977

Current smoker: Age-adjusted RR 1·95 (1·68–2·27)

Multivariate-adjusted RR 1·94 (1·64–2·28)

Former smoker: Age-adjusted RR 1·43 (1·31–1·56)

Multivariate-adjusted RR 1·39 (1·27–1·52)

a,bHuerta, 2007[28]1996–1997OXMIS/ReadPatients without psoriasis who smoke: 2008 (20·1)Patients with psoriasis who smoke: 1013 (25·4)

OR 1·37 (1·25–1·50)

Adjusted OR 1·45 (1·31–1·59)

aSetty, 2007[32]1991–2005Questionnaire494

Current: 131

Former: 262

Current smoker: Age-adjusted RR 1·82 (1·50–2·21)

Multivariate-adjusted RR 1·78 (1·46–2·16)

Former smoker: Age-adjusted RR 1·40 (1·20–1·62)

Multivariate-adjusted RR 1·37 (1·17–1·59)

In the pooled analysis of the NHS, NHS II and HPFS, individuals with a history of smoking were 1·39 times more likely to develop incident psoriasis, after multivariate adjustment for confounders (95% CI 1·27–1·52), and current smokers were nearly twice as likely to develop psoriasis, compared with nonsmokers (adjusted OR 1·94, 95% CI 1·64–2·28).[31] Similar results were found in the GPRD study, where active smokers had an increased risk of developing psoriasis (adjusted OR 1·45, 95% CI 1·31–1·59).

Pooled analysis of the NHS, NHS II and HPFS also revealed a positive association between smoking intensity, smoking duration and risk of incident psoriasis. Those who smoked 1–14 cigarettes per day were 1·81 times more likely to develop psoriasis, compared with nonsmokers (95% CI 1·38–2·36), and those who smoked ≥ 25 cigarettes per day were over twice as likely to develop psoriasis, compared with nonsmokers (adjusted OR 2·29, 95% CI 1·74–3·01).[31] Subjects who had smoked < 10 years had close to the same risk of developing psoriasis as nonsmokers (adjusted OR 1·07, 95% CI 0·83–1·39), whereas those who had smoked ≥ 30 years were nearly twice as likely to develop psoriasis (adjusted OR 1·99, 95% CI 1·75–2·25).[31] The overall population-attributable risk of smoking to incident psoriasis ranged from 15% in the NHS II to 20% in the NHS, suggesting that up to one in five cases of psoriasis is attributable to smoking.

Discussion

This systematic review and meta-analysis supports and quantifies the association between psoriasis and smoking. We examined both the prevalence of smoking among patients with psoriasis and the association between smoking and incident psoriasis. From the studies included in this systematic review, both current and former smokers were more likely to develop incident psoriasis, compared with nonsmokers. Additional factors associated with increased odds of developing psoriasis included a greater number of cigarettes smoked per day, longer durations of smoking habits, and greater pack-years of smoking. Similarly, among prevalence studies, patients with psoriasis were more likely to be either current or former smokers compared with those without psoriasis.

A number of pathophysiological mechanisms likely underlie the associations between smoking and psoriasis.[47] Smoking induces oxidative stress and free radical damage, reduces concentrations of antioxidants in the plasma, increases vascular endothelial dysfunction, and increases plasma viscosity.[48] Cigarette smoke contains approximately 1017 free radicals per puff.[49] This increased free radical exposure has the potential to trigger a cascade of systemic consequences, including development of psoriasis. A study conducted of patients with plaque psoriasis, for example, showed significantly higher Psoriasis Area and Severity Index (PASI) scores in patients with psoriasis who are smokers (= 28), compared with nonsmoking patients with psoriasis (= 26; = 0·014).[50] Oxidants, including superoxide (inline image) and hydrogen peroxide (H2O2), are elevated in both lesional and nonlesional skin of patients with psoriasis, whereas many antioxidant levels are reduced.[51] Nicotine induces an increased secretion of interleukin (IL)-12 from dendritic cells. Numerous other inflammatory cells and cytokines, including tumour necrosis factor, interferon-γ, IL-2 and granulocyte-monocyte colony-stimulating factor, are produced through nicotine-activating pathways.[51] Smokers also show a dysregulated expression of vascular endothelial growth factor, an important element in angiogenesis. Increased, abnormal angiogenesis induced by nicotine is shown to accelerate atherosclerotic plaque formation in mice.[52] This pathological blood vessel formation may, in part, explain the relationship between smoking and both psoriasis and atherosclerosis.[51]

Environment and behaviour also likely interact with the genetic loci that predispose for psoriasis. A two-stage case–control study (= 7223) investigated the relationship between smoking behaviour and nine different single nucleotide polymorphisms (SNPs) that are associated with risk of psoriasis.[53] Two SNPs at the CSMD1 gene (a tumour suppressor gene that influences epithelial cell turnover), rs7007032 and rs10088247, were significantly associated with smoking (P-value of 0·0023 for both SNPs). However, multiple logistic regression analysis did not show a significant association between psoriasis and these two SNPs. CSMD1 may also contribute to psoriasis development by influencing the differentiation of keratinocytes, and to smoking habits by influencing addictive behaviour at a neuronal level. The interaction between these genetic loci and smoking behaviour may play a role in influencing the development of psoriasis but will require further investigation.

This study has several strengths. Firstly, we performed a systematic review and meta-analysis using multiple data sources. Secondly, multiple authors reviewed the data independently and assessed study quality. We also used random effects meta-analysis to arrive at a more conservative estimate of effect size, due to between-study heterogeneity. We also formally assessed possible sources of heterogeneity using meta-regression analysis. The findings of this systematic review and meta-analysis are limited by the quality of the primary data. For example, some studies did not perform multivariable adjustment, thereby possibly overestimating the effect size in some cases. Because most studies were case–control and used ORs as a measure of effect size, it is also possible that the meta-analysis point estimates provide an overestimate of the effect relative to studies of incidence. We also found evidence of possible publication bias, with an absence of small, negative studies. The currently published studies were also limited in their assessment of psoriasis disease severity, making it difficult to draw any conclusions on any possible association between smoking and the severity of psoriasis.

In conclusion, this systematic review and meta-analysis supports a strong association between increased prevalence of smoking among patients with psoriasis and an increased incidence of psoriasis among smokers. Patients who smoke should be informed of the increased risk of developing psoriasis as well as the additional adverse effects that any amount of smoking can have on one's health. Patients with established psoriasis should also be strongly counselled to quit smoking in order to minimize the risk of cardiovascular comorbidities associated with psoriasis.

Ancillary