This population-based survey examined randomly selected adults applying a rigorous sampling procedure. The 90% of response rate indicates that the sample is reasonably representative of adult people in Yogyakarta City. The involvement of the head of neighborhoods in knocking on doors, as expected, improved the response rate.
This study identified a 4-week period prevalence of 7.3% for using non-prescribed antibiotics and highlighted patterns of use where amoxicillin, ampicillin, fradiomicyn-gramicidin, ciprofloxacin and tetracycline were the self-medicated antibiotics and were purchased mainly in pharmacies. These antibiotics were used to treat a variety of minor symptoms, such as the common-cold, cough, sore throat, and fever, for mostly less than five days of use. Such practices were based on reasons of previous successful experience, saving time and money, and information obtained from health professionals, lay people, and printed materials.
We acknowledge some limitations in this study. Although this study uses the population of Yogyakarta as its sampling frame, generalization to the whole population of Indonesia should be done carefully. The area of this study, Yogyakarta City, is an urban with a high population density and a mostly literate population. Thus, results may be readily generalizable to other urban areas of Indonesia. For example, the pattern of SMA found in two big cities in Java Island is similar with those in this current study . The situation may well differ in rural areas, however. Further, the use of antibiotics in this study is self reported, and therefore there are some issues regarding subjectivity and imprecision. However, such issues were minimized by using a four-week recall period, which has been extensively used by other similar studies [5, 6]. The sample size calculation was based on an estimated period prevalence of self medication with antibiotics of 50%, the midpoint of estimation in the literature. This approach had short comings, however, as the figure is very dependent on the length of the recall period. Unfortunately, our calculation did not account for the variation in these periods. Moreover, because of limited resources a larger precision (d = 0.05) was used to calculate the sample size and a probability proportionate to size (30% and 5% of sub-districts and neighborhoods, respectively) was applied to limit the numbers of clusters in the second and third stages. However, the involvement of all clusters (districts) of the first stage could minimize sample bias . In addition, a quarter of the respondents did not mention their current employment thus this question may be sensitive in the local socio-cultural context. People would perhaps be unwilling to be honest when they are unemployed or non-economically active or employees of informal/casual sectors. Despite such limitations, as a consequence of representativeness of the urban population following the inclusion of a random sample, information yielded from this study is invaluable in describing the use of non-prescribed antibiotics.
A one-month period prevalence of self medication with antibiotics in this study is comparable to that of a household survey in Jordanian population (a one-month period prevalence 9%) . Higher period prevalence has been reported in other developing regions, for example, Mexico (2-week period prevalence 18%)  and Sudan (4-week period prevalence 48%) . A four- week period prevalence of 7.3% found in this present study is higher than the previous Indonesian study (a 4-week period prevalence of 3%) . This finding indicates a potential increase in self medication with antibiotics among Indonesians, particularly in urban areas. Further, a four-week period prevalence of 7.3% indicates that an even larger proportion of the urban population in Indonesia would self medicate with antibiotics annually.
Generally, the patterns and the reasons of using non-prescribed antibiotics found in this study are similar to those in other less developed countries [1, 2, 6, 26] and in some European countries . Such patterns, for example, the use of non-prescribed antibiotics for the common-cold, indicate inappropriate use. They also indicate overprescribing of antibiotics as people do tend to duplicate their previous prescriptions for self medication [28, 29]. Although in Indonesia and elsewhere prescription scripts are retained in pharmacies, people commonly make notes of prescribed medicines for future reference if similar medical problems reoccur [28, 29]. In Indonesia there is a promising intervention program to improve rational prescription of antibiotics particularly for URTI (Upper Respiratory Tracts Infection) known as Monitoring, Training and Planning (MTP) conducted by the Centre for Clinical Pharmacology and Drug Policy Studies. The program has been implemented among health professionals in hospitals in Yogyakarta and has been successfully decreased antibiotics prescription for URTI. Such programs should be applied extensively with the involvement of health professionals in both public and private health service; and private practices.
The patterns of use also indicate inadequate information given to patient as well as to the public. In most developing countries, drug information given by health providers in both primary health care centers and hospitals is not yet optimal . In addition, pharmacy personnel tend to be businessmen rather than professional. When antibiotics are requested by consumers requests are neither refused nor questioned . Since antibiotics are legally grouped as prescription-only medicines, information about such medicines is, unfortunately, not readily available to the general public, but only to health practitioners. It is different with the over-the-counter (OTC) medicines or general sales, where information about the product can be easily accessed through the packed inserts and mass media advertisements. Currently, there is an initiative to locate Indonesian pharmacists in primary health care centers. Such an initiative is promising, particularly to provide adequate information related to medicines including antibiotics and to improve the quality of health care service in general.
Economic considerations are an important factor influencing SMA behavior . There is mixed evidence on the association between lower income and SMA behavior [1, 11–15, 31]. In this study household income level is not associated with either intent or actual SMA. However, this issue is significant for further investigation in Indonesia because a substantial proportion of the whole Indonesian population is poor . Further, as found in this study, having no health insurance is a significant factor associated with self medicating with antibiotics, which is consistent with a previous Indonesian study . Unfortunately, only approximately 30% of the whole population is covered by health insurance .