Countries participating in the study were recruited from 2 networks of surveillance systems: European Surveillance of Antimicrobial Consumption (20) and European Antimicrobial Resistance Surveillance System (21). A multistage sampling design was used for sample selection in each participating country. Within each country, a region with average prescribed antimicrobial drug consumption was chosen. In those countries where this information was not available (Poland, Czech Republic, Lithuania, Croatia, and Romania), a region was selected that was representative of the country's population in terms of age and sex. In each region, a city (75,000–750,000 inhabitants) and a rural area (5,000–10,000) were selected. By using population registries, including lists of persons in the identified cities and rural areas, persons >18 years of age were selected by simple random sampling (computer-generated random numbers). To calculate the sample size, we needed the standard deviation of the unknown prevalence. As the standard deviation was a function of this unknown prevalence, we took the maximal possible value of the standard deviation (22). To obtain a precision of 0.05, the sample size needed was 400 persons per country. To adjust for possible nonresponse, we selected larger samples; sample sizes in the countries were 1,000–3,000 persons, equally distributed in urban and rural areas. Self-administered questionnaires were mailed between March and July 2003, and reminders with a new questionnaire attached were sent 2–4 weeks later.

We developed an English questionnaire specifically for this survey, translated it into national languages, and back-translated it to ensure consistency.1 The questionnaire was pilot-tested in each country. It could be completed either anonymously or with identifiable details to allow a follow-up study. Questions asked about the respondent's use of antimicrobial drugs during the past 12 months, how they were obtained, how they were stored at home, and whether the respondent would consider using drugs without consulting a physician. Details of the drugs used (name of the medicine, symptom or disease coded with International Classification of Primary Care codes [23], and duration of use) and demographic characteristics of the respondents were included. Antibacterial drugs for systemic use (Anatomical Therapeutic Chemical class J01) (24) were included in the analyses. Medicines erroneously reported as antimicrobial drugs were excluded from the analyses. Ethics or data committee approval for the survey was required in Sweden, Denmark, Belgium, United Kingdom, Ireland, Malta, Czech Republic, Slovenia, Croatia, Romania, and Lithuania and was obtained from the local ethics or data committees of these countries.

Respondents were classified as self-medicating if they reported that they had taken any antimicrobial drugs in the previous 12 months without a prescription from a physician, dentist, or nurse and as prescribed users if antimicrobial drugs had been prescribed. (Physician respondents who reported using nonprescribed drugs were not classified as self-medicating.) Those classified as at-risk for self-medication included those who indicated the intention to self-medicate or store drugs at home. Intended self-medication was defined as answering "yes" or "maybe" to the question, "In general, would you use antimicrobial drugs for yourself without contacting a doctor/nurse/hospital?" Two estimates were used to assess storage of drugs: a maximum estimate, including all respondents who stored antimicrobial drugs, and a conservative estimate that excluded those respondents who stored antimicrobial drugs and had taken the same drugs for a prescribed course in the previous 12 months.

A total of 15,548 respondents completed the questionnaires. The mean response rate of the countries was 40% (Table 1).

Self-medication with antimicrobial drugs occurred in all countries that participated in this survey. We included the data from both countries that had high and low response rates. In most of the countries with low response rates (except Spain), no other information is available about self-medication, an often overlooked issue. The second reason for including these countries was that low response was not a problem of this study only, but a general problem of surveys in these countries (29,30). This finding implies that if we want to include information about these countries, the results may be biased. In addition, debate is growing that low response is less problematic in affecting survey estimates than previously assumed (31). Nevertheless, the prevalence rates of self-medication in countries with low response rates should be considered as a rough estimate and interpreted as an indication that the problem exists.

Antimicrobial drug self-medication prevalence varies widely among different European regions, with the highest rates in eastern and southern countries, and the lowest in northern and western. Besides actual self-medication, intended self-medication is clearly relevant: it is a strong predictor of actual self-medication. Intended self-medication has a much higher prevalence than actual self-medication, indicating that the population at-risk is much larger than those who have actually self-medicated in the previous 12 months. Another risk factor for actual self-medication is the availability of drugs at home; opportunity encourages use. Our findings contribute to the growing evidence that estimates of antimicrobial drug use that are based on prescription data only are likely to underestimate actual consumption in both Europe and the United States (11,32). Our European estimates are low in comparison with those from a recent study in the US Hispanic community that showed that ≈20% of the respondents acknowledged getting drugs without a prescription in the United States (32). The only comparable high rates were found in Spain, Romania, and Lithuania, where they ranged from 9% to 18%. However, these figures should be compared with caution because our estimates refer to acquiring drugs without prescription in the last 12 months and the United States study refers to ever acquiring them.

We found that many persons used antimicrobial drug leftovers from previous prescriptions, as was the case in reports from the United States (8–10,12). Drugs could be left over because extra tablets were dispensed (in many countries pharmacies dispense drugs per package, not exact number of tablets) or because of patient noncompliance. Noncompliance may result in 2 inappropriate courses if the patient does not take the amount of medication prescribed and self-medicates later. Earlier findings indicated lower compliance in Italy and Spain than in Belgium, France, and the United Kingdom (33). In Italy, 41% of the interviewees who had taken drugs in the previous 12 months saved part of the course for future use, whereas only 4% of British interviewees reported this behavior (33).

In general, respondents' self-diagnosed disorders were self-limiting and antimicrobial drugs would not have been indicated. In contrast to studies in developing countries, this study identified few cases of self-medication for sexually transmitted diseases (34,35). Only 2 respondents in Lithuania reported self-medication for "gynecological infection" that might have been a sexually transmitted infection.

In this survey, persons who were more prone to self-medicate with antimicrobial drugs were younger persons, more educated, and had chronic diseases. This finding corresponds to those of studies conducted in the United States and Greece, which also found that higher educational status is associated with misuse of drugs (8,17). This relationship cannot be directly attributed to educational status. The interpretation of symptoms is also relevant. Previously, a study in the United States showed that persons with a higher education level tended to believe that antimicrobial drugs were less effective for upper respiratory infections with clear discharge but more effective with discolored discharge (36).

Antimicrobial drug self-medication is a cause for concern because it may contribute to the spread of antimicrobial drug resistance. Self-treatment with a drug that is ineffective against the causative organism or with an inappropriate dosage may increase the risk of selection of resistant organisms that are difficult to eradicate. These resistant organisms may then be transferred into the community. Our findings illustrate that adverse effects are aggravated by self-medication when unnecessary drugs, such as chloramphenicol, tetracycline, and aminoglycosides, are taken. Other problems related to self-medication include drug interactions, masked diagnoses, and superinfection.

Our results are comparable to those of other studies such as the Eurobarometer study (14,19,27). A study on antimicrobial drug storage among Spanish households showed that 42% of Spanish households had drugs at home, including those currently used (14). This finding is comparable to the prevalence of drugs stored (50%) in our study. In Malta, a higher prevalence (19%) of self-medication was found (19) than in our study, perhaps because the study included self-medication in the previous 2 years, while our study included the previous 12 months. Furthermore, the pattern of the prescribed use of drugs in different regions of Europe in our study is similar to that found in the study by Goossens et al., which was based on information from national databases (4).

A strength of our study is that we used the same methods and comparable samples in all countries, which facilitated an overview of the European situation. The low response rate in some countries is a limitation of our study, however. Although we calculated the prevalence rates adjusted for nonresponse, they are based on the assumption that respondents who replied after the reminder most resemble nonrespondents.

As with all self-reported data, results of this survey have the potential for recall bias, underreporting, or overreporting. We attached the list of the most commonly used antimicrobial drugs in each country to the questionnaires to reduce recall problems. To discourage underreporting of self-medication, the questions about drug use were formulated in a neutral way in which the source of the drug could be chosen from 6 predefined sources or "other source."

Substantial variation in the prevalence rates of antimicrobial drug self-medication among the European regions suggests that cultural (37) and socioeconomic factors play a role, as do disparities in health care systems such as reimbursement policies, access to health care, and drug dispensing policies. Another factor is the acquisition of antimicrobial drugs from pharmacies without prescription, which occurred most frequently in eastern European countries. Although over-the-counter sale of antimicrobial drugs is illegal in all participating countries, there is clearly a need to enforce the law in some countries.

Antimicrobial drug self-medication is a cause for concern in Europe. Even the lowest prevalence, 1 person per 1,000 respondents, implies that 10,000 persons in a population of 10,000,000 are self-medicating annually. Our study indicates a high prevalence of self-medication in countries that reported high resistance levels (southern and eastern countries). Even in the countries with low actual self-medication, substantial intended self-medication and drug storage occurs. Efforts to reduce inappropriate use of antimicrobial drugs should include the issue of self-medication and should involve prescribers, pharmacists, and the general public. The number of tablets dispensed in pharmacies should be limited, and patients should be instructed to discard their leftover drugs. Large-scale public campaigns, such as those recently launched in the United States, Canada, Belgium, and Australia (38), should include detailed instructions and emphasize the potential risks of using antimicrobial drugs without medical guidance.