Schizophrenia and tobacco smoking in a Spanish psychiatric hospital $

Schizophrenia Research 60 (2003) 313 317 www.elsevier.com/locate/schres Schizophrenia and tobacco smoking in a Spanish psychiatric hospital $ Adrián LLerena a,b,c, Alfredo de la Rubia b, Eva M. Peñas-Lledó a,c, Francisco J. Diaz d,e, Jose de Leon d, * a Department of Pharmacology and Psychiatry, Medical School, University of Extremadura, Spain b Research and Clinical Psychopharmacology Unit, Merida Psychiatric Hospital, Merida, Badajoz, Spain c Department of Clinical Pharmacology, Beira Interior University Medical School, Covilha, Portugal d Mental Health Research Center at Eastern State Hospital, 627 West Fourth St., Lexington, KY 40508, USA e Department of Statistics, Universidad Nacional, Medellin, Colombia Received 10 May 2000; received in revised form 6 August 2001; accepted 8 August 2001 Abstract This study in a Spanish hospital replicated two US studies suggesting that schizophrenia is associated with smoking when compared with other severe mental illnesses. Neither antipsychotics nor institutionalism could explain this relationship. Seventy of the 100 schizophrenic and 53 of the 100 non-schizophrenic inpatients were current smokers. After correcting for confounding factors, schizophrenia increased the risk of smoking by 2- to 3-fold. Heavy smoking was not associated with schizophrenia. D 2003 Elsevier Science B.V. All rights reserved. Keywords: Schizophrenia; Antipsychotics; Smoking; Tobacco; Logistics regression 1. Introduction The idea that tobacco smoking appears more closely associated with schizophrenia than other severe mental illnesses is supported by studies published worldwide (Table 1). The frequency of current smoking in schizophrenia varied from 64% to 93% according to country, time, out- or in-patient status, inclusion of both genders or only male gender, and sample size. This prevalence was lower in patients with other mental illnesses (30 $ PII of original article S0920-9964(01)00327-9, S0920-9964(02)00507-8. * Corresponding author. Tel.: +1-859-246-7487; fax: +1-859- 246-7019. E-mail address: jdeleon@uky.edu (J. de Leon). 90%). Odds ratios (ORs) comparing prevalence were extraordinarily consistent across studies, ranging from two to three in six of seven studies including both genders (Table 1). Population-based studies suggest that patients with other mental illnesses smoke more than the general population but they do not include a sufficient number of schizophrenic patients (Lasser et al., 2000). Other psychiatric patients may be more appropriate controls for studies of patients with schizophrenia since they are also exposed to factors that could influence smoking such as psychiatric medications or exposure to psychiatric treatment settings where many patients smoke. All the ORs in Table 1 were computed from twoway cross-tabulations and adjustments for such potential confounding variables as gender or age were not 0920-9964/03/$ - see front matter D 2003 Elsevier Science B.V. All rights reserved. doi:10.1016/s0920-9964(02)00508-x

314 A. LLerena et al. / Schizophrenia Research 60 (2003) 313 317 Table 1 Odds ratios comparing frequencies of current smoking in patients with schizophrenia to patients with other mental illnesses Author Country Type Current smokers OR a 95% CI Schizophrenia No schizophrenia Hughes et al., 1986 USA Outpatients 88% (21/24) 48% (92/193) 7.7 b 2.2 26.6 Carvajal et al., 1989 Chile Inpatients 81% (78/96) 66% (217/331) 2.3 b 1.3 4.0 Calabresi et al., 1991 Italy Outpatients 80% (57/71) 66% (19/29) 2.1 b 0.82 5.6 de Leon et al., 1995 USA Inpatients 85% (201/237) 67% (83/123) 2.7 b 1.6 4.5 Herrán et al., 2000 Spain Outpatients 64% (41/64) 39% (45/115) 2.8 1.5 5.2 de Leon et al., 2002a USA Inpatients 75% (335/449) 55% (70/127) 2.4 1.6 3.6 de Leon et al., 2002b USA Inpatients 83% (55/66) 65% (33/51) 2.7 b 1.1 6.5 Addition of all studies 78% (788/1007) 58% (559/969) 2.6 2.2 3.2 Current smokers in males Schizophrenia No schizophrenia O Farrell et al., 1983 USA Inpatients c 88% (180/204) 71% (55/77) 3.0 b 1.6 5.82 Calabresi et al., 1991 Italy Inpatients 88% (57/65) 90% (18/20) 0.80 b 0.15 4.1 de Leon et al., 1995 USA Inpatients 93% (140/150) 78% (58/74) 3.9 b 1.7 9.0 Diwan et al., 1998 USA Outpatients 86% (54/63) 30% (6/20) 14.0 b 4.3 46.0 de Leon et al., 2002a USA Inpatients 80% (236/294) 66% (49/74) 2.1 b 1.2 3.6 de Leon et al., 2002b USA Inpatients 91% (39/43) 70% (21/30) 4.2 b 1.1 15.2 Addition of all studies 86% (706/819) 71% (207/295) 2.7 1.9 3.7 a All ORs were computed from two-way cross-tabulations. b OR was not described in the article and was calculated for this table. c Includes a small number of females. made. Such adjustments could be made with multivariate analyses, but only two large studies in US hospitals have reported such analyses (de Leon et al., 1995, 2002a). Those two US studies provided significant adjusted odds ratios ranging from 2.3 to 3.8. Two smaller US studies also used logistic regression (de Leon et al., 2002b; Diwan et al., 1998). No multivariate analyses have been reported from other countries. Clinical experience has suggested that schizophrenia could also be associated with heavy smoking among smokers (Lohr and Flynn, 1992). Research results concerning this association are inconsistent. Olincy et al. (1997) showed that 20 patients with schizophrenia had higher urinary cotinine levels when compared with a control group of 20 heavy smokers. However, de Leon et al. (2002a) did not unequivocally replicate the findings of de Leon et al. (1995) that schizophrenia was associated with heavy smoking among smokers. Using a scale for nicotine dependence in current smokers, two studies did not find differences between schizophrenia and mood disorders (Diwan et al., 1998; de Leon et al., 2002b). The aim of this Spanish study is to confirm the conclusion from the multivariate analyses in US hospitals that schizophrenia is associated with current smoking even after adjusting for potential confounding variables. The replication in other countries will help to support the idea that this association is probably explained by biological factors and occurs across different cultures. In current smokers, heavy smoking was also investigated and was defined using the same definition ( z 1.5 packs per day) as in prior studies (de Leon et al., 1995, 2002a). 2. Methods This study was performed in 1994 at Merida Psychiatric Hospital, Spain. The catchment area of this long-term hospital covers 300,000 people. 2.1. Subjects The sample included the first 100 patients with schizophrenia and first 100 patients with other diag-

A. LLerena et al. / Schizophrenia Research 60 (2003) 313 317 315 Table 2 Frequencies of current smoking and of heavy smoking among current smokers Current smoking Heavy smoking Schizophrenic 70% (70/100) 40% (28/70) Nonschizophrenic 53% (53/100) 43% (23/53) Male schizophrenic 81% (68/84) 40% (27/68) Male nonschizophrenic 64% (44/69) 41% (18/44) Female schizophrenic 13% (2/16) 50% (1/2) Female nonschizophrenic 29% (9/31) 56% (5/9) noses who consented to provide the urine sample needed to take part in the study from approximately 500 hospitalized patients. All patients were Caucasians. Eighty-four percent (84/100) of patients with schizophrenia were male. Sixty-nine percent (69/100) of non-schizophrenic patients were male. The respective mean ages were 55 (SD = 14) and 50 (SD = 14) years for schizophrenic and non-schizophrenic patients, respectively. The respective mean durations of current hospitalization were 17 (SD = 14) and 11 (SD = 12) years. The respective percentages of patients taking antisychotics were 74% (74/100) and 49% (49/100). The respective mean doses in chlorpromazine equivalents of those taking antipsychotics were 326 (SD = 272) and 280 (SD = 304). Clinical diagnoses were taken from the charts. The DSM-IIIR diagnoses among non-schizophrenic patients were organic mental disorders (65%), mood disorders (10%), and others (25%). Study patients were asked about current smoking and number of cigarettes smoked per day. 2.2. Statistics The Statistical Package for Social Sciences (SPSS) was used for calculations. Non-dichotomous variables were dichotomized to calculate ORs as in de Leon et al. (1995; 2002a). The following variables were examined for their possible effect on frequencies of current smoking and heavy smoking: diagnosis (schizophrenia or non-schizophrenia), high neuroleptic dose ( z 500 chlorpromazine equivalents), old age (>55 years), gender, long duration of hospitalization (>2 years), and polydipsia, which was defined as having a morning or afternoon specific gravity of urine less than 1.009 (de Leon et al., 1994). ORs were computed from twoway cross-tabulations for univariate analyses and from logistic regressions for multivariate analyses. In one logistic regression, current smoking was the dependent variable. In the other logistic regression that used only current smokers, heavy smoking was the dependent variable. The 95% confidence intervals (CI) for ORs were calculated. The Hosmer Lemeshow goodnessof-fit test was used to test the fitness of logistic models (a p-value >0.05 indicates good fitness). When examining the association between the variables described above and current or heavy smoking, a multivariate logistic regression computes for each variable an OR that is adjusted for the possible effect of the other variables. Univariate analyses are reported to compare with studies in Table 1. 3. Results 3.1. Frequency of current and heavy smoking The frequencies of current smoking were 70% and 53% among schizophrenic and non-schizophrenic patients, respectively (Table 2). The frequencies of heavy smoking among current smokers were 40% in patients with schizophrenia and 43% in non-schizophrenic patients. Table 3 Odds ratios (and 95% confidence intervals) of significant variables associated with current smoking in univariate and logistic regression analyses (N = 200) Univariate Logistic regression a Schizophrenia 2.1 (1.2 3.7) b 2.3 (1.1 4.6) c High antipsychotic dose Male gender 8.9 (4.2 19.2) d 9.6 (4.2 22.0) e Old age 0.43 (0.24 0.77) f 0.34 (0.17 0.71) g Polydipsia Long hospitalization 0.52 (0.26 1.0) h Only odds ratios that were significant at the 0.05 level are reported in the table. Blank spaces indicate non-significant odds ratios. a Hosmer Lemeshow goodness of fit test: v 2 = 3.3, df =7, p = 0.86. b v 2 = 6.1, df =1, p = 0.01. c Wald v 2 = 5.4, df =1, p = 0.02. d v 2 = 37.7, df =1, p < 0.01. e Wald v 2 = 28.9, df =1, p < 0.01. f v 2 = 8.1, df =1, p < 0.01. g Wald v 2 = 8.2, df =1, p < 0.01. h v 2 = 3.6, df =1, p = 0.06 (borderline significance).

316 3.2. Univariate analyses A. LLerena et al. / Schizophrenia Research 60 (2003) 313 317 4.3. Limitations Schizophrenia was significantly associated with current smoking (Table 3). Other variables associated with current smoking were male gender, lack of old age, and possibly short periods of hospitalization. High neuroleptic dose and polydipsia were not significantly associated with current smoking. In addition, schizophrenia was not associated with heavy smoking among current smokers. Only lack of old age was significantly associated with heavy smoking (OR = 0.43, CI = 0.19 0.96). 3.3. Logistic regressions Schizophrenia was significantly associated with current smoking (Table 3). Male gender and lack of old age were also significantly associated with current smoking. No variable was significantly associated with heavy smoking. 4. Discussion 4.1. Univariate analyses Without adjusting for other variables, the odds of currently smoking in patients with schizophrenia was 2.1 times higher than in non-schizophrenic patients (Table 3). This odds ratio is not significantly different from worldwide studies that included both genders (Table 1), as can be seen from overlapping confidence intervals. However, comparisons must be performed with caution, because variable definitions, controlled variables, and criteria for sample selection may vary across studies. 4.2. Logistic regressions When adjusting for other variables, the OR for schizophrenia was 2.3 (Table 3) which was not significantly different from US multivariate studies. In the logistic regression, schizophrenia, male gender, and lack of old age were associated with current smoking, similarly to US multivariate analyses (de Leon et al., 1995, 2002a). In this Spanish study, none of the variables were associated with heavy smoking among current smokers. Unfortunately, no questions on retrospective information on smoking, such as smoking initiation or quitting, were asked. The use of biological measures may have improved the determination of heavy smoking. The inclusion of only the most cooperative patients (100 from each group) is a limitation. However, from the 300 patients not included in this report, we were able to obtain data from the chart including current smoking (but not number of cigarettes per day, polydipsia or antipsychotic dose). When analyses for current smoking were conducted in these 300 patients, they provided remarkably similar ORs for schizophrenia. The OR in the univariate analysis was 2.0 (CI 1.3 3.3) versus 2.1 in the reported sample (Table 3). The corrected OR in the logistic regression was 2.4 (CI 1.4 4.3) versus 2.3 in the reported sample (Table 3). As patients had very long duration of hospitalization, these results may not be generalizable to outpatients. However, if Table 1 is reviewed, two of the three outpatient studies including both genders provided similar odds of currently smoking in patients with schizophrenia (7.7, 2.1 and 2.8). 4.4. Schizophrenia and smoking The remarkable similarities among numerical results from different cultures suggest that a biological explanation for the association between schizophrenia and smoking may be likely although other more complex explanations may also be possible. A prior US study comparing controls and patients with schizophrenia and mood disorders suggested that differences in smoking behavior between controls and patients with mood disorders may be explained by differences in education levels. However, the differences between controls and schizophrenic patients persisted after controlling for education level and other demographic factors (de Leon et al., 2002b). The higher frequency of current smoking in patients with schizophrenia is a combination of a lower quitting rate and a higher initiation rate (de Leon, 1996, de Leon et al., 2002b). Since most patients with schizophrenia start to smoke before the onset of the illness, a vulnerability to schizophrenia

A. LLerena et al. / Schizophrenia Research 60 (2003) 313 317 317 may be associated with a high vulnerability to starting smoking (de Leon, 1996, de Leon et al., 2002b). In that sense, Freedman et al. (1997) have described a genetic neurophysiological abnormality in patients with schizophrenia and their relatives, which is temporarily corrected by a high peak of nicotine. Smoking may be accompanied by an increased risk of tardive dyskinesia (Dalack et al., 1998). Smoking may be associated with metabolic induction, so smokers tend to consistently receive higher doses of antipsychotics (de Leon et al., 2002a). 5. Conclusion This study conducted in Spain confirms the conclusion from multivariate analyses in two US hospitals that schizophrenia is strongly associated with a higher frequency of current smoking even after adjusting for potential confounding variables. This association appears to be consistent across different countries. Neither antipsychotics nor institutionalism can explain this association because non-schizophrenic patients at long-term hospitals are also exposed to these factors. An association between schizophrenia and heavy smoking among current smokers was not supported by this study. Acknowledgements The authors are grateful to the patients and staff from the Merida Psychiatric Hospital for their collaboration. This study was possible by a grant awarded to Dr. LLerena from the Government of Extramadura (Junta de Extremadura, Consejeria de Educación Ciencia y Tecnología y Fondo Social Europeo IPRI0036). Dr. de Leon s salary was partly supported by grants MH-51380 from the National Institute of Mental Health. Francisco Diaz is an Associate Instructor in Statistics at Universidad Nacional de Colombia and was supported by a grant from the Kentucky Department of Mental Health and Mental Retardation and the Department of Psychiatry of the University of Kentucky through the University Collaboration Project, and by a fellowship from the Department of Statistics of the University of Kentucky. References Calabresi, M., Casu, G., Dalle Luche, R., 1991. La prevalenza del fumo di tabacco nei pazienti psichiatrici. Minerva Psichiatrica 32, 89 92. Carvajal, C., Passig, V.C., San Martin, R.E., Zuñiga, S.A., 1989. Prevalencia del consumo de cigarrillos en pacientes psiquiatricos. Acta Psiquiatrica y Psicologica de America Latina 35, 145 151. Dalack, G.W., Healy, D., Meador-Woodruff, J.H., 1998. Nicotine dependence in schizophrenia: clinical phenomena and laboratory findings. Am. J. Psychiatry 155, 1490 1501. de Leon, J., 1996. Smoking and vulnerability for schizophrenia. Schizophr. Bull. 22, 405 409. de Leon, J., Verghese, C., Tracy, J., Josiassen, R., Simpson, G.M., 1994. Polydipsia and water intoxication in psychiatric patients: a review of the epidemiological literature. Biol. Psychiatry 35, 408 419. de Leon, J., Dadvand, M., Canuso, C., White, A.O., Stanilla, J., Simpson, G.M., 1995. Schizophrenia and smoking: an epidemiological survey in a state hospital. Am. J. Psychiatry 152, 453 455. de Leon, J., Tracy, J., McCann, E., McGrory, A., Diaz, F.J., 2002a. Schizophrenia and tobacco smoking: a replication study in another US psychiatric hospital. Schizophr. Res. 56, 55 65. deleon,j.,diaz,f.j.,rogers,t.,browne,d.,dinsmore,l., 2002b. Initiation of daily smoking and nicotine dependence in schizophrenia and mood disorders. Schizophr. Res. 56, 47 54. Diwan, A., Castine, M., Pomerlau, C., Meador-Woodruff, J.H., Dalack, G.W., 1998. Different prevalence of cigarette smoking in patients with schizophrenia vs. mood disorders. Schizophr. Res. 33, 113 118. Freedman, R., Coon, H., Myles-Worsley, M., Orr-Urtreger, A., Olincy, A., Davis, A., Polymerpoulos, M., Holik, J., Hopkins, J., Hoff, M., Rosenthal, J., Waldo, M.C., Reimherr, F., Wender, P., Yaw, J., Young, D.A., Breese, C.R., Adams, C., Patterson, D., Adler, L.E., Kruglyak, L., Leonard, S., Byerley, W., 1997. Linkage of a neurophysiological deficit in schizophrenia to a chromosome 15 locus. Proc. Natl. Acad. Sci. U. S. A. 94, 587 592. Herrán, A., de Santiago, A., Sandoya, M., Fernández, M.J., Díez- Manrique, J.F., Vázquez-Barquero, J.L., 2000. Determinants of smoking behavior in outpatients with schizophrenia. Schizophr. Res. 41, 373 381. Hughes, J.F., Hatsukamy, D.K., Mitchell, J.E., Dahlgren, L.A., 1986. Prevalence of smoking among psychiatric outpatients. Am. J. Psychiatry 143, 993 997. Lasser, K., Boyd, J.W., Woolhandler, S., Himmelstein, D.U., Mc- Cormick, D., Bor, D.H., 2000. Smoking and mental illness: a population-based prevalence study. JAMA 284, 2606 2610. Lohr, J.B., Flynn, K., 1992. Smoking and schizophrenia. Schizophr. Res. 8, 93 102. O Farrell, T.J., Connors, G.J., Upper, D., 1983. Addictive behaviors among hospitalized psychiatric patients. Addict. Behav. 8, 329 333. Olincy, A., Young, D.A., Freedman, R., 1997. Increased levels of nicotine metabolite cotinine in schizophrenic smokers compared to other smokers. Biol. Psychiatry 42, 1 5.