The 801 patients referred a total of 2.622 health problems, with an average of 3.27 (SD 2.22) problems per patient. Females reported 1447 (55.2%) health problems, versus 1.175 among the men. The analysis revealed significant differences in terms of gender (p < 0.001), with a female predominance in pathology of the nervous system and genitourinary tract. A non-significant male predominance in respiratory, skin and sensory organ diseases was observed (table 1).

The 3.634 different drugs consumed by the 801 users were distributed into 12 therapeutic groups. The 5 most consumed groups accounted for 81% of the total, and comprised the digestive apparatus and metabolism, nervous system, cardiovascular system, locomotor apparatus and respiratory system. A total of 54.8% of the drugs were consumed by the 410 females in the sample. Significant differences (p < 0.001) were observed in drug consumption by therapeutic groups and patient gender. These differences were recorded in the therapeutic groups central nervous system and genitourinary tract – with greater female consumption (percentage difference 4.5; 95%CI, 1.8-7.2 and 1.9; 95%CI, 0.9-2.9, respectively), and in respiratory system and systemic anti-infectious therapy, with greater male consumption (percentage difference 5.2; 95%CI 3.3-7.1 and 1.7; 95% CI 0.7-2.8, respectively) (table 2).

By therapeutic subgroups, women showed a greater use of mineral supplements (excluding iron), antianemic agents, antihemorrhoidal and antivaricose drugs, psycholeptics and psychoanaleptics. In comparison, men showed a greater consumption of antacids, antiflatulence and antiulcer drugs, antidiabetic, antigout, and antiparkinson drugs, as well as antiasthmatic, anti-influenza and antitussive medication (table 2).

The mean number of drugs consumed per person was 4.5 (SD 3.2); 54.6% (95%CI, 51.1-58.8) of the patients consumed more than 3 drugs. Drug use increased with age and the female sex (p < 0.001). The relationship was significantly higher in women aged 45-64 years (fig. 1). Significant differences (p < 0.001) were also seen in drug use by educational level, and professional, occupational and marital status. Table 3 shows the mean drug consumptions distributed by predictive variables and gender groups. The number of drugs used was found to decrease with increasing educational level and professional categories. The analysis of differences by gender showed that –with the exception of the illiterate subjects– women consumed more drugs. These differences was statistically significant at the complete primary education level (difference of the means 1.0; 95%CI, 0.2-1.8) and in the lowest professional category, where 76% were involved in domestic activities (difference of the means 2.0; 95%CI, 1.1-2.8). Housewives and inactive subjects were those who consumed most medication. Moreover, widowers were found to consume most drugs, and either married or divorced individuals used more drugs than single subjects (difference of the means 2.0; 95%CI, 0.8-3.2). Married women or widows consumed more drugs than men (difference of the means 0.7; 95%CI, 0.1-1.2). Grouping of the family structure into 2 categories showed significant differences (p < 0.01) in the sense of increased drug use in nucleated families (difference of the means 1.0; 95% CI, 0.5-1.7), and also an increased use of medication among women belonging to a nucleated family (difference of the means 0.7; 95%CI, 0.2-1.2). In turn, significant differences were recorded in drug consumption according to the time elapsed since the last visit to the primary care center (p < 0.001). In effect, drug use was found to increase with the frequency of visits. The analysis of the 4 visiting frequency levels and their distribution by gender revealed increased drug consumption among women, with significant differences in the highest frequency (fig. 2).

Figure 1.

Mean drug consumption, distribution by age and gender.

(0.07MB).

Figure 2.

Mean drug consumption, distributed by frequency of visits (measured as time elapsed since last visit) and gender.

(0.08MB).

Table 4 shows the results of the application of the two models described under Methods' section, where the value of the F-test was found to be significant in both. As a result, the variability accounted for by the equations was significant in the two models. Analysis of Model 1 showed the variability accounted for by the model to be 0.8, and the regression coefficients of the variables «health problems» and «frequency of visits» were significant. In Model 2, the value of the residuals, that of the fitted determination coefficient and the regression coefficients of the variables «health problems» and «frequency of visits» were similar to those of Model 1 (nonconfounding with the rest of variables). The optimum model for explaining the value of gender was Model 1, since it takes into account all the predictive variables studied, without losing valuable accuracy. These data show that: a) the number of health problems and the frequency of visits explain 82.2% of the variability of the variable «number of drugs consumed», and b) for each health problem 1.3 drugs are consumed, fitted for all the predictive variables studied.

Our results show 80% of the medication consumed to correspond to five therapeutic groups. Although the order may vary, the most frequently consumed therapeutic groups reported in the literature essentially coincide with our own findings1,10,20,22-25. By therapeutic subgroups, our observation that women consume more psycholeptics and psychoanaleptics coincides with the published reports1,6,18,20,23-25. Our results agree with those reporting female consumption to be greater in relation to psychotherapeutic agents, with no gender differences in relation to cardiovascular drugs and diuretics1,24. However, other studies have reported differences in relation to these last therapeutics groups, central nervous system, nutritional or hematological medication23,25.

In the descriptive analysis, the present results show that women in global terms consume more medication than men, and that drug use increases with age. These findings coincide with the literature, though in our case significant differences by gender were only observed for the 45-64 years age group2-4,6,8,10,16,20,26,27. In contrast, Kotzan et al2, in a study of the Medicaid population, found no gender differences in the number of drugs prescribed per patient. These authors explain this result in terms of a poor health status among the males of the study population involved. Our results coincide in reflecting increased drug use at the lower educational and social levels6,8,22,26,27, and are consistent with those of Payne et al10, who found education to be a predictor of medication use. Moreover, women were seen to consume more drugs at all social levels, though the differences only reached statistical significance in the category fundamentally comprising housewives. The higher drug use observed among inactive individuals and housewives was probably related to patient age, since 91% of the housewives were over age 45, and 76% of the inactive individuals were over 65 years old. However, housewives consumed more drugs than the inactive users despite being younger – an observation that could be related to gender (99% of the users engaged in domestic activities corresponding to women, while 74% of the inactive subjects were males), the social role of women involved in domestic activities, and to morbidity. These considerations would be in line with the results of Rohlfs et al28, indicating that both young and mature housewives consume more medication than women who are either actively employed or unemployed. In agreement with other studies6,22, widows and widowers consume most drugs, though we believe this could be linked to both patient age (80% over age 65) and morbidity (more health problems than in the average population).

The frequency of visits to primary care centers has been associated to increased drug use both in our study and elsewhere in the literature6,7,10,22,26. The fact that women make use of health services more often than men has already been documented6,10,20,26,28. However, in coincidence with the description of Payne et al10 regarding the proportion of Canadian women reporting medication use increments with the number of physician visits, on analyzing consumption in our series by fitting to the frequency of visits, we found that women continue to use more drugs than males in all frequency categories. Drug consumption is obviously related to the existence of health problems. The mean number of problems for which our patients consumed drugs was again greater among women than in men – in agreement with other authors6,10,20,22. In our case, the number of health problems recorded was higher than in the study by López-Torres et al22, who only considered chronic health problems in elderly patients, and not perceived morbidity as in both our case and in the Health Survey of the Valencian Community20.

In the study by Roe et al23 using a database with pharmacy claim and eligibility information, gender differences in medication are reported in several classes of medications, including antidepressants and antianxiety and pain medication. However, the authors did not investigate the possible causes of such differences, and moreover claimed that future research should be conducted to afford better understanding of the role of socioeconomic status, ethnicity and geographic region. In this sense, the increased drug consumption associated with the female gender, advanced age, a low educational level, lower professional categories or housewife status, and inactivity or unemployment –documented both in our study and elsewhere in the literature– fails to sustain statistical significance when taking into account the number of health problems and the frequency of visits in the multivariate analysis. In this way, the Women's Health Surveillance Report from the Canadian Institute for Health Information reported that medication use increases with a greater number of chronic diseases, physician visits, and age10. In contrast, Van der Waals et al29 concluded that the gender difference among benzodiazepine users seems to be due to general practitioners being less stringent when prescribing these drugs for women. We agree with Eggen6, who considered that health problems (in terms of number) are the best predictor of drug use. Only the more frequent utilization of health services, controlling for the rest of variables (including the number of health problems), conditions increased drug consumption. If as suggested by our results and those of Payne et al10, health problems are effectively the main factor underlying increased drug consumption among women, then historical factors (e.g., social discrimination) could cause women to become ill more often or feel more ill – thus leading to increased drug consumption28. If this were not the case, then patient gender bias could manifest through the tendency of physicians to overdiagnose women and/or underdiagnose males –thereby prescribing more drugs among the former– but not through a greater or lesser accessibility to medication on the part of women29. The gender of patients and/or doctors may condition the diagnosis of illness, though once the patients are diagnosed, health problems are the main factor underlying drug consumption among both women and men.