Association of Early-Stage Psoriasis With Smoking and Male Alcohol Consumption: Evidence From an Italian Case-Control Study

Objective  To analyze the association of psoriasis with smoking habits and alcohol consumption before first diagnosis by a dermatologist.

Design  A multicenter case-control study. Interviews were conducted by trained medical investigators using a structured questionnaire.

Setting  Outpatient services of 10 general and 10 teaching hospitals in northern and southern Italy.

Subjects  Patients with a first diagnosis of psoriasis made by a dermatologist and a history of skin manifestations of no longer than 2 years after the reported disease onset. Patients with new diagnoses of skin diseases other than psoriasis were selected as control subjects. A total of 404 case patients (median age, 35 years) and 616 controls (median age, 36 years) were included in the analysis.

Results  The risk for psoriasis was higher in ex-smokers and in current smokers than in patients who never smoked. The relation with smoking was stronger and more consistent among women than men. In men, a significant association was restricted to the ex-smoker status. Smoking was strongly associated with pustular lesions (29 patients) with an adjusted odds ratio of 10.5 (95% confidence interval, 3.3-33.5) for those smoking more than 15 cigarettes per day. No significant overall association with alcohol consumption was documented after controlling for smoking habits. However, the risk seemed to vary according to sex, with a moderate association being documented in men.

Conclusions  Cigarette smoking and, in men, alcohol consumption are associated with psoriasis. There is a strong association between smoking and pustular lesions.

THE INTERACTION between genetic and environmental factors, ie, multifactorial heredity, seems to play a role in the causes of psoriasis. Several studies exploring environmental factors have considered smoking habits and alcohol consumption, but their results have been discordant. Smoking has been found to be more frequent among patients with psoriasis than among control subjects before the disease onset or the first dermatological diagnosis in some but not all studies. Similar discordant data concern alcohol consumption.^1-8 Variations by sex have been suggested, with smoking being more strongly associated with psoriasis among women and alcohol consumption among men.^4,8 Palmoplantar pustulosis, a disease related to psoriasis, has also been found to correlate strongly with smoking habits, but whether a similar degree of association applies to other varieties of pustular psoriasis remains unclear.⁹ The widespread use of alcohol and cigarettes, the disabling nature of psoriasis, and the lack of effective and safe treatments for the disease justify efforts to further our understanding of these issues.

We herein present a case-control study involving patients with newly diagnosed psoriasis and controls with newly diagnosed dermatological conditions other than psoriasis. Our aims were to analyze the association of psoriasis at an early-stage with common environmental factors, such as smoking and alcohol consumption, and to explore the existence of variations in selected subgroups.

The study was conducted in Italy within the framework of the Italian Group for Epidemiologic Research in Dermatology.⁷ The outpatient services of 10 teaching hospitals and 10 general hospitals in northern and southern Italy participated.

Entry criteria for patients with psoriasis were as follows: a first diagnosis of psoriasis made by a dermatologist and a history of skin manifestations of no longer than 2 years after the reported disease onset. The onset was considered to be the date an individual first became aware of the clinical manifestations attributable to psoriasis. The definition of onset was guided by a thorough inquiry into the timing of relevant signs and symptoms. For inclusion, at least 1 typical plaque of psoriasis, ie, a fixed, well-demarcated erythematous scaly patch, was required; however, most patients presented with diffuse manifestations leading to a straightforward clinical diagnosis by the specialist. Cases of psoriasis guttata and pustular psoriasis were excluded if they were not associated with a typical plaque of psoriasis. Erythrodermic psoriasis was not considered because of its rarity and the diagnostic difficulties. All eligible patients who were seen consecutively during the study period were invited to participate. As controls, we selected subjects with newly diagnosed dermatological conditions other than psoriasis who were seen in the same outpatient services as the case patients. Patients with skin conditions regularly associated with alcohol consumption or liver disorders were excluded (ie, presenting symptoms of porphyria cutanea tarda, cutaneous manifestations of hemochromatosis and cirrhosis, pellagra, and zinc deficiency syndrome). About 1% of the suitable controls were excluded because of these conditions. The controls were the first eligible dermatological patients observed on randomly selected days in the 2 weeks following the selection of case patients. Initially, 2 controls per case patient were selected. Subsequently, however, because of the difficulties experienced in recruitment, inclusion was limited to 1 control per case patient. Controls were age-matched to case patients in decade categories, and the inclusion of case patients and controls was limited to subjects aged 16 through 70 years. Oral informed consent was obtained from each subject, and fewer than 2% of the eligible subjects refused to participate.

The study was conducted during the following 3 separate periods: from January 1, 1988, through August 31, 1990; from January 1 through December 31, 1993; and from April 1, 1996, through October 31, 1997. During the first period, the study was restricted to 6 centers in northern Italy. Subsequently, an additional 14 centers participated, including 6 from the southern part of the country. The results of the 3 phases of the study were comparable, and no important differences among the participating centers were observed. A total of 404 case patients with psoriasis, satisfying our entry criteria, and 616 controls were recruited. The psoriasis cases were classified as ordinary chronic plaque psoriasis (324 [80.2%]), guttata psoriasis associated with chronic plaque psoriasis (30 [7.4%]), pustular psoriasis, mostly represented by so-called psoriasis with pustules (29 [7.2%]), and mainly flexural psoriasis (21 [5.2%]). None of the patients in this last group had evidence of contact allergy. Diagnoses in the control group included eczema (154 [25.0%]), skin cancers (101 [16.4%]), urticaria (86 [13.9%]), skin infections (81 [13.1%]), pityriasis rosea (62 [10.1%]), and a variety of other skin diseases (132 [21.4%]).

Information regarding sociodemographic factors, smoking habits, alcohol consumption, coffee and tea consumption, consumption of selected dietary factors, family history of psoriasis in first-degree relatives (ie, parents and siblings), and personal medical history was obtained from case patients and controls by trained investigators using a structured questionnaire. Anthropometric measures, including height and weight, were also recorded. Information on smoking and alcohol consumption was obtained through a lifetime inquiry, including age at onset, average and maximum daily consumption, kind of consumption, and existence of periods of abstinence. The questionnaire used has been validated in Italian epidemiological studies.¹⁰ The interviewers had biomedical training and experience in medical interviewing. We computed the average consumption of alcohol per day, assuming a comparable pure alcohol content in each type of drink (that is, 120 mL of wine, 330 mL of beer, 40 mL of spirits, or 12-13 g of pure alcohol).¹¹ A period of abstinence of at least 6 months before the date of diagnosis was requested for a patient to be classified as an ex-smoker or ex-drinker.

For each variable, we computed the odds ratios (ORs) of having psoriasis as estimates of relative risks and the corresponding 95% confidence intervals (CIs).

Initially, estimates were obtained from data stratified by age and sex, when appropriate, according to the Mantel-Haenszel method.¹² The significance of the linear trend in risk was assessed using the Mantel test.¹³ Subsequently, to account for the effects of potentially confounding factors, unconditional multiple logistic regression analysis with maximum likelihood fitting was used.¹⁴ The regression equations included terms for age, sex, calendar year at interview, marital status, occupation, history of psoriasis in first-degree relatives, body mass index (BMI; calculated as weight in kilograms divided by the square of height in meters), smoking habits, and alcohol and coffee consumption.

The distribution of case patients and controls according to age, sex, marital status, occupation, family history of psoriasis in first-degree relatives, BMI, and coffee consumption is reported in Table 1. The median age at diagnosis was 35 years in case patients and 36 years in controls. A total of 248 case patients (61.4%) reported onset before 39 years of age. There was a significantly higher proportion of male subjects in the case compared with the control group. No association was documented between marital status or occupation and psoriasis. As expected, the risk for psoriasis was higher in those reporting a family history of the disease. The age- and sex-adjusted OR was 7.9 (95% CI, 4.7-13.3) for subjects who reported a history of the disease in parents, whereas the OR adjusted for age, sex, and number of siblings was 3.7 (95% CI, 2.4-5.8) for subjects reporting a history of the disease in siblings. Body mass index was also associated with psoriasis. Compared with a BMI of lower than 22.0, the age- and sex-adjusted OR was 1.3 (95% CI, 1.0-1.8) for a BMI of 22.0 to 26.0 and 1.9 (95% CI, 1.4-2.6) for a BMI of greater than 26.0. No association was observed between coffee consumption and psoriasis.

Table 2 shows the distribution of case patients and controls according to smoking habits and alcohol consumption. The OR of having psoriasis was higher in previous and current smokers. Compared with people who had never smoked, the Mantel-Haenszel adjusted ORs for ex-smokers was 1.9 (95% CI, 1.3-2.7); for subjects currently smoking up to 15 cigarettes per day, 1.5 (95% CI, 1.1-2.1); for those smoking 16 to 24 cigarettes per day, 1.8 (95% CI, 1.2-2.7); and for those smoking more than 24 cigarettes per day, 2.4 (95% CI, 1.3-4.3). Multivariate ORs were consistent with those adjusted for age and sex. Interestingly, the risk for psoriasis decreased fairly smoothly with increased duration of smoking habits. Moreover, in ex-smokers, the risk tended to decrease with the increase in time since quitting smoking and to increase with the number of cigarettes smoked per day (data not shown).

Case patients tended to report drinking alcohol more frequently than controls. The correlation between alcohol consumption and smoking was weak (Spearman correlation coefficient, 0.24), and after adjusting for smoking and other potential confounders, the risk estimates were closer to 1. Ex-drinkers were not at an increased risk for psoriasis, and no relation between the duration of drinking habits and psoriasis was documented.

Table 3 presents OR estimates for smoking and alcohol consumption stratified by sex. The ex-smoker status was a risk factor among men but not women. As for current smokers, the association was stronger in women than in men, for whom risk estimates appeared even consistent with no effect at all. At variance with these findings, in women no relation of psoriasis with alcohol consumption was documented, whereas an association was documented in men with a dose-response relationship. Table 4 presents OR estimates for smoking and alcohol consumption according to the clinical variety of psoriasis. A particularly strong association was documented between smoking and pustular psoriasis, but not between alcohol consumption and pustular psoriasis. We also examined smoking and alcohol consumption according to age at diagnosis (≤35 vs >35 years) and history of psoriasis in first-degree relatives (present vs absent). No important variations were documented except for somewhat higher estimates of ORs for previous and current smoking status in the younger patients compared with the older ones (data not presented).

Some lines of evidence have suggested relations between psoriasis and smoking habits and alcohol consumption, pointing to a possible sex difference. A number of other diseases that are strongly linked with smoking in the general population, including cardiovascular diseases,^15,16 Crohn disease,¹⁷ lung cancer, tumors of the upper airways,^18,19 and kidney cancer,^18-20 are associated with psoriasis. Some of these associations have been found only in women.^16,20 Also, liver cirrhosis has been linked with psoriasis.^16,21 In the only incidence study of psoriasis we are aware of, the smoking profile for male patients was similar to that from a random sample of the general population, whereas female patients smoked more than women in the general population. Incident cases were defined as cases newly diagnosed at the Mayo Clinic facilities, which cover the whole Rochester, Minn, population.⁵ In a case-control study of palmoplantar pustulosis, a condition related to psoriasis and largely predominant in women, the OR in smokers was around 7 (99% CI, 3.9-13.2), and no association with alcohol consumption was documented.⁹ Other case-control studies of plaque psoriasis have yielded less striking results.^4,6-8 However, in 2 case-control studies from Finland conducted separately in men and women, alcohol consumption before the disease onset was associated with psoriasis among young men, whereas smoking was associated with psoriasis among women.^4,8

Our study confirmed that smoking is associated with psoriasis with a dose-response relationship. The association was stronger and more consistent among women than men and was also particularly strong for pustular lesions. In addition, we found some evidence pointing to an association with alcohol consumption in men.

We should sound a note of caution on some limitations of our study. Our case patients received a first diagnosis of psoriasis by a dermatologist. A history of skin manifestations of no longer than 2 years before the date of the dermatological diagnosis was required. Our aim was to restrict the assessment to recently developed cases that were reliably evaluated by a specialist. To avoid including indeterminate or borderline cases, the presence of a typical plaque of psoriasis was required. However, no satisfactory criteria are available for standardizing the diagnosis of psoriasis in clinical and epidemiological studies.²²

The choice of dermatological controls was dictated mainly by the need to control for the reasons leading to specialist consultation. Usually it is stated that studies with controls with disease at the same organ level as the case patients may result in overmatching.¹⁴ This is a rather protective bias, and we tried to minimize it by excluding skin manifestations regularly associated with alcohol consumption or liver diseases. These represented a small proportion of all the eligible controls. We matched for age but not sex, and there was a difference in the distribution of this variable among case patients and controls. However, we adjusted all our analyses, whenever indicated, for sex and examined risks in stratification by sex.

Our decision to use the index date as the date of the first diagnosis by a specialist may be questionable, a major problem being the inability to separate exposures that precede from those that follow clinical onset. We acknowledge that there are difficulties involved in defining the real onset of a disease like psoriasis, which may slowly evolve from mild nonspecific signs (eg, dandruff) to overt clinical manifestations. It is also difficult to standardize the assessment of the onset of the different skin disorders presented by the control group. We were not interested in acute exposures as triggers of the disease, but in more chronic, predisposing factors. As a consequence, precise onset definition is not such a crucial problem in our analyses.

The retrospective nature of our study is a more general limitation, with its potential for recall and information bias. It was reassuring that the frequency of smoking and alcohol consumption in our control group was similar to the estimates reported for the Italian population during the same period.²³ It was also reassuring that for another exposure frequently reported as a risk factor in humans, namely coffee consumption, no association was documented.

The statistical power of our subgroup analyses may be questioned. This issue is complicated by multiple statistical testing. Since we did not make any correction for multiple tests, we advise readers to give more weight to our overall results than to any specific result in subgroups, which should be taken as mainly explorative. In Italy, women tend to have a remarkably low alcohol consumption,²⁴ and in our study, only 9 female case patients (4.8%) reported consuming more than 2 drinks per day. As a consequence, the analysis for high alcohol consumption lacks substantial statistical power. Recent Italian figures indicate that greater proportions of men than women have given up smoking. The association between previous smoker status and psoriasis in men suggests that, compared with the controls, a greater proportion of male patients with psoriasis gave up smoking, perhaps because of early symptoms of the disease. If selective quitting is implicated, then the risk for current smoker status among men may be underestimated by our study.

The associations we have documented may, in principle, represent the effect of an underlying, unmeasured confounding factor such as psychological stress.²⁵ We did not assess stressful life events. Type A personality and psychological stress have been correlated with smoking and heavy alcohol consumption.²⁶ However, the strong association between smoking and pustular lesions, even if based on a limited number of cases, is more indicative of a genuine biological effect.

More than 4000 substances have been identified in cigarette smoke. Many of the adverse effects of smoke may result from oxidative damage to critical biological substances.²⁷ Such damage could result from oxidants present in cigarette smoke, eg, carbon monoxide, and from the activation of phagocytic cells that generate reactive oxygen species. In 1 study, polymorphonuclear leukocytes from smokers with psoriasis responded to a significantly greater degree to a standard chemotaxin than did polymorphonuclear leukocytes from nonsmokers with psoriasis, control smokers, and control nonsmokers.²⁸ High levels of arachidonic acid metabolites have been documented in psoriatic lesions.²⁹ Peroxidation of arachidonic acid by smoke may lead to the formation of F₂-isoprostanes. Concentrations of these substances are increased dramatically in animal models of oxidant injury and increased in smokers.²⁷ Recently, nicotinic cholinergic receptors have been demonstrated on keratinocytes stimulating calcium influx and accelerating cell differentiation.³⁰ Constant stimulation of these receptors may control keratinocyte adhesion and upward migration in the epidermis. Smoking and nicotine seem to influence inflammatory processes in the skin.^31,32

The increased risk for psoriasis in women who are smokers, together with the higher prevalence of psoriasis among men compared with women documented in several population studies,³³ points to the possible role of hormonal and reproductive factors, since smoking has a well-defined antiestrogenic effect.³⁴ Heavy alcohol consumption in men may increase the risk for infection or mechanical trauma that, in turn, may affect psoriasis. Moreover, an effect of alcohol on lymphocyte transformation has been suggested.³⁵ If smoking and, possibly limited to men, alcohol consumption cause psoriasis, the impact on public health would be far from negligible. The population-attributable fraction for smoking calculated from our study was about 0.2, indicating that about 1 in 5 cases of psoriasis may be related to smoking. Independent of their causal role, the associations we have documented could improve our understanding of the comorbidity related to psoriasis and should be taken into consideration when providing care for the disease. In addition, our data concerning pustular psoriasis suggest that chronic exposure to selected environmental factors may influence the clinical expression of psoriasis. The effect of environmental factors on the extension, distribution (eg, acral lesions), clinical variety, and response to treatment of psoriasis should be evaluated further in prospective studies.

Accepted for publication June 24, 1999.

The study was partly supported by the following pharmaceutical companies: Janssen-Cilag, Cologno Monzese; Roche, Milano; Schering, Segrate; Boehringer Ingelheim, Firenze; Glaxo-Wellcome, Verona; and Novartis, Origgio, Italy.

We wish to thank Gillian Jarvis for her editorial assistance and the dermatologists from the participating centers for making this study possible.

Steering Committee

T. Cainelli, MD; A. Rebora, MD; A. Peserico, MD; G. Tognoni, MD.

Regional Coordinators

Catania Clinica Università,Catania: A. Di Prima, MD. Padova Clinica Università, Padua: A. Belloni Fortina, MD; A. Peserico, MD. Napoli Clinica I Università, Naples: N. Balato, MD. Ferrara Clinica Università, Ferrara: A. R. Virgili, MD. Terni Clinica Università, Terni: P. L. Bruni, MD. Ospedale Militare, Taranto: V. Ingordo, MD. Arcispedale S. Maria Nuova, Reggio Emilia: G. Lo Scocco, MD. Torino Clinica Università, Turin: C. Solaroli, MD; P. Puiatti, MD. Verona Clinica Università, Verona: D. Schena, MD; A. Barba, MD; C. Chieregato, MD. Ospedale Riuniti, Bergamo: A. Di Landro, MD. Ospedale Civile, Cremona: E. Pezzarossa, MD. Bologna Clinica Università, Bologna: S. Marzaduri, MD. Ospedale M. Bufalini, Cesena: R. Morelli, MD. Milano Clinica Università, Ospedale S. Paolo, Milan: R. Vergani, MD; C. Pazzini, MD. Firenze Clinica Università, Florence: P. Carli, MD; W. Volpi, MD. Spedali Civili, Brescia: P. G. Calzavara-Pinton, MD. Napoli Clinica II Università, Naples: F. Grimaldi-Filioli, MD; A. Lo Schiavo, MD. Casa Sollievo della Sofferenza, S. Giovanni Rotondo: M. Iannantuono, MD. Ospedale S. Paolo, Savona: A. Farris, MD; C. Micalizzi, MD. Ospedale Nuovo, Monza: E. Rossi, MD.

Reprints: Luigi Naldi, Clinica Dermatologica, Ospedali Riuniti, L go Barozzi 1, 24100 Bergamo, Italy (e-mail: gised@uninetcom.it).