Abstract
The purpose of this study was to identify predictors of hazardous alcohol consumption in patients with cluster headache (CH). We investigated 246 German CH patients with the Alcohol Use Disorders Identification Test (AUDIT). The average daily alcohol consumption was 6.5 g. Predictors for hazardous drinking (AUDIT ≥ 5 points; 21.5% of patients) were male gender [odds ratio (OR) 4.15, 95% confidence interval (CI) 1.35, 12.71], episodic as opposed to chronic CH (OR 4.8, 95% CI 1.38, 16.67) and a low demanding job as opposed to a high demanding job (OR 2.28, 95% CI 1.15, 4.51). Our data indicate that CH patients drink less alcohol compared with the German population and that CH seems to protect against hazardous alcohol consumption. Moreover, predictors for hazardous alcohol consumption in CH patients are not different from the general population.
Introduction
Increased alcohol consumption in patients with cluster headache (CH) has been reported by some (1, 2), but not all epidemiological studies (3). In particular, potential predictors for increased alcohol consumption in CH patients have never been investigated.
The Alcohol Use Disorders Identification Test (AUDIT) was developed by the World Health Organization for the early identification of hazardous and harmful drinking as well as alcohol dependence, particularly in the preceding 12 months (4, 5). It provides an accurate measure of risk across gender, age and cultures (4, 5). Early studies have used a cut-off ≥8 points in primary care settings as a criterion for hazardous drinking (4), but in the general population a cut-off ≥5 points has been proposed as a better measure (6).
We sought to assess the incidence and predictors of hazardous drinking by using the AUDIT in a cohort of 246 German CH patients.
Methods
Patients
Between April 2002 and March 2004 we prospectively recruited 257 patients from the Headache Clinic at the University of Essen, Germany, nationwide from self-help groups and via internet advertisements on our clinic homepages. Eleven CH patients provided incomplete information and were excluded, leaving 246 for this analysis. Written consent was obtained from all patients and the local Ethics Committee of the University of Essen approved this study. The diagnosis of CH was verified according to the criteria established by the International Headache Society (IHS) in 1988 either by direct standardized history taking (n = 60; 24.4%), a standardized telephone interview by an experienced neurologist (n = 136; 55.3%), or a standardized mailed questionnaire (n = 50; 20.3%). All diagnoses were re-evaluated using the revised IHS criteria from 2004. Of the 246 patients, 191 (77.6%) were men and 55 (22.4%) women. The mean age at first diagnosis was 36.9 years (SD 11.8). Three-quarters (184) of patients had episodic CH, 41 (16.7%) had chronic CH, and 21 (8.5%) were diagnosed after their first bout (newly diagnosed), periodicity thus being undetermined. Ninety percent of the patients experienced CH attacks within the 12 months prior to study enrolment.
The Alcohol Use Disorders Identification Test
The questionnaire consists of 10 questions, each of which can give a maximum of 4 points (total score range 0–40 points). We used a cut-off of ≥5 points from all 10 questions as recommended for the general population (6). Only to approximate alcohol consumption did we use the first two Alcohol Use Disorders Identification Test (AUDIT) questions (7). Accordingly, the answers were scored as follows: question 1 (‘How often do you have a drink containing alcohol?’): never = 0, ≤1/month = 0.25, 2–4/month = 1, 2–3/week = 3, ≥4/week = 5; question 2 (‘How many drinks containing alcohol do you have on a typical day when you are drinking?’): 1–2 drinks = 20 g, 3–4 drinks = 40 g, 5–6 drinks = 70 g, 7–9 drinks = 100 g, ≥10 drinks = 120 g. For comparison we followed the criteria of the British Medical Association (8), which defines hazardous drinking as an average daily consumption of 20 g of alcohol in women and 30 g of alcohol in men.
Statistical analysis
We calculated the average daily consumption of alcohol by multiplying scores from the first two questions divided by seven. We compared the differences in daily consumption between episodic CH and chronic CH patients and between men and women by using the Wilcoxon signed rank test for continuous variables and the χ2 test for categorical variables. All tests were two-tailed, and we considered a P-value of <0.05 as statistically significant.
For all further analyses we used results from all 10 questions. We used logistic regression to identify potential predictors of hazardous alcohol consumption, defined as a cut-off of ≥5 points on the AUDIT. We considered age (quartiles: <35.7, 35.7–43, 43–53, >53 years), gender, education [low (no schooling, elementary/grade school) vs. intermediate (O-levels/six-form high-school, other) vs. high (A-levels/high school, university)], occupation (white collar worker/civil servant vs. others), smoking status (present and former smoker vs. non-smoker), diagnosis (episodic vs. chronic CH), living in a relationship vs. living alone and being treated at our headache clinic vs. not being treated at our clinic as potential predictors.
We calculated univariate and multivariable-adjusted odds ratios (OR) and their corresponding 95% confidence intervals (CIs). We used a stepwise selection procedure to identify significant predictors for hazardous alcohol consumption. Because of the high association between age and alcohol consumption, we forced age into all models. We tested the model fit using the Hoshmer–Lemeshow goodness-of-fit test (9). We evaluated the predictive ability of the model by determining the area under the receiver operating characteristic (ROC) curve (9). All analyses were performed using the SAS system (version 8.2; SAS Institute, Cary, NC, USA).
Results
Daily alcohol consumption
The overall average daily consumption of alcohol was 6.5 g, with episodic CH patients consuming significantly more alcohol compared with chronic CH patients and men consuming significantly more alcohol than women. Likewise, men with episodic CH consumed significantly more alcohol each day compared with men with chronic CH. There was no significant difference in daily alcohol consumption between women with episodic CH and women with chronic CH. We also did not find a significant difference between the number of patients with episodic CH and chronic CH who drank more than 20 g (women) or 30 g (men) daily (Table 1).
Daily alcohol consumption
ECH, Episodic cluster headache; CCH, chronic cluster headache.
Average: values are in grams ± SD; approximated according to (7).
Categories: values are in percentage.
Calculated according to (10).
Reference in (10).
Predictors for hazardous alcohol consumption
A total of 53 (21.5%) patients scored ≥5 points on the AUDIT and were classified as hazardous alcohol consumers. Table 2 summarizes the univariate predictors for hazardous drinking.
Univariate predictors for the Alcohol Use Disorders Identification Test (AUDIT) cut-off ≥5 in patients with cluster headache (CH)
Low education: no schooling, elementary/grade school; intermediate education: O-levels/six-form high-school, other; high education: A-levels/high school, university.
Low demand occupation: blue collar worker, retired, self-employed, student, apprentice, house-wife, unemployed, helping family member.
CB, Clinic-based; NCB, non-clinic-based.
The stepwise selection procedure that evaluated all potential predictors (Table 2) identified three main risk factors: male gender, episodic as opposed to chronic CH and blue-collar/other workers compared with white-collar workers (Table 3). Use of a backward elimination procedure yielded the same final model. The goodness-of-fit test (P = 0.16) indicated that the final model fitted reasonably well. The area under the ROC curve was 0.71, indicating a moderate discrimination of the final model.
Multivariable∗ predictors for the Alcohol Use Disorders Identification Test (AUDIT) cut-off ≥5 in patients with cluster headache (CH)
Low demand occupation: blue collar worker, retired, self-employed, student, apprentice, house-wife, unemployed, helping family member.
Adjusted for all other variables in the model.
Discussion
The average daily alcohol consumption in our CH patients was 6.5 g. Using a cut-off score ≥5 on the AUDIT we identified male gender, episodic as opposed to chronic CH and having a low demanding compared with having a high demanding job as predictors for hazardous alcohol consumption.
Data about the significance of alcohol consumption in CH are conflicting (1–3) and potential predictors of hazardous alcohol consumption among CH patients have not been investigated. In our cohort, the average daily alcohol consumption was 6.5 g compared with 22.9 g in the German general population (10). Male gender and a less demanding job are predictors of alcohol consumption not only in our cohort of CH patients but also in the general population. Our data are in contrast with a recent Italian study (2), which may be due to (i) differences in the study population, (ii) a time trend and (iii) differences in life style. The Italian study investigated only male patients from a referral centre while our cohort consisted of a largely population-based sample with 22.4% women and 75% of patients not treated at our centre. Furthermore, data collection in the Italian cohort was between 1975 and 1995 but between 2002 and 2004 in our cohort, and habits such as drinking and smoking probably are different between the two countries. Our finding, that CH patients drink less alcohol compared with the German general population probably reflects that alcohol triggers attacks. This is plausible for episodic CH patients in our cohort, 90% of whom had had attacks during the preceding 12 months, and even more for chronic CH patients, who are in a permanent episode. This also explains why chronic CH patients drink even less alcohol than episodic CH patients. We cannot conclude that alcohol influences the course of the disease. However, our data indicate that CH does not predispose to alcohol indulgence, but rather seems to be protective against hazardous alcohol consumption. Furthermore, the predictors for increased alcohol consumption in CH patients are not different from the general population.
The strengths of our study include the large cohort, the motivation of CH patients to give accurate answers to elucidate conditions underlying their disease and the use of the AUDIT as a validated tool (4, 5). A cut-off score ≥5 rather than ≥8 yields a better trade-off between sensitivity and specificity for the German population (6). Furthermore, when using a cut-off ≥8 the univariate analysis yields similar results.
Among the limitations are the lack of a direct comparison group from the general population or patients with other headaches. However, the AUDIT relies on absolute rather than comparative measurement. Moreover, the alcohol consumption in our cohort was just approximated and compared with federal statistics (10). Yet (i) the method for the approximation described is established (7) and (ii) it was not the primary aim of our study to investigate absolute amounts of alcohol consumption. Finally, our cohort does not represent all CH patients in Germany, thus obviating a generalization to other populations. Future studies must be awaited to discover if other predictors for hazardous alcohol consumption among CH patients are identified.
Footnotes
Acknowledgements
The authors thank the ‘Cluster-Kopfschmerz Selbsthilfe Gruppen Deutschland’ (CSG) for their continuing support and all patients who participated in this study.