SOCIO-ECONOMIC DETERMINANTS OF PRESCRIPTION ANTIBIOTIC AND MEDICINE USE AND ITS RELATIONSHIP WITH FAMILY MEDICINE REÇETE EDİLEN ANTİBİYOTİK VE İLAÇ KULLANIMININ SOSYOEKONOMİK BELİRLEYİCİLERİ VE AİLE HEKİMLİĞİ İLE İLİŞKİSİ

Yanlış ve aşırı antibiyotik tüketiminin bireylerin sağlığı üzerinde doğrudan etkileri olduğu gibi, ülkelerin sınırlı sağlık hizmeti bütçelerini etkileyen dolaylı sonuçları da vardır. Gelişmekte olan ülkeler için ilaç ve antibiyotik tüketiminin belirleyicileri hakkında çok az kanıt bulunmaktadır. Aile hekimliği ile hekimlerin reçete ettiği antibiyotik tüketimi arasındaki ilişkiyi yıllar içinde incelemekteyiz. Ulusal temsiliyeti olan Türkiye Sağlık Araştırmalarını 2008-2012 yıllarını kapsayacak şekilde çalışmamızda kullanmaktayız. Reçete edilen antibiyotik ve ilaç kullanımını belirleyen faktörleri anlamak için En Küçük Kareler yöntemini uygulayarak çok değişkenli bir regresyon analizi kullanmaktayız. Bulgular, antibiyotik kullanım eğiliminin 2010'da düştüğünü ve 2008'e kıyasla 2012'de de düşmeye devam ettiğini göstermektedir. Ayrıca aile hekimi ziyaretinin, Sağlıkta Dönüşüm Programının amacı ile tutarsız olacak şekilde, antibiyotik kullanım olasılığını artırdığını gözlemlemekteyiz. Ancak, aile hekimi ziyareti ile antibiyotik kullanımı arasındaki pozitif ilişki yıllar geçtikçe zayıflamaktadır. Türkiye çeşitli önlemler almış ve antibakteriyel direnci bir dereceye kadar azaltmayı başarmış olsa da, antibiyotikler halen daha tedavi edici yöntemler arasında en çok başvurulan ilaç türü olarak karşımıza çıkmaktadır. Anahtar Kelimeler: Reçete Edilmiş Antibiyotik Kullanımı, Reçete Edilmiş Ilaç Kullanımı, Aile Hekimliği, Sosyoekonomik Belirleyenler Mehmet Akif Ersoy İktisadi ve İdari Bilimler Fakültesi Dergisi Mehmet Akif Ersoy University Journal of Economics and Administrative Sciences Faculty Cilt: 7 Sayı: Özel Sayı s.987-1003 Volume: 7 Issue: Special Issue p.987-1003 Aralık 2020 December 988 GENİŞLETİLMİŞ ÖZET


INTRODUCTION
Antibiotics have saved many lives in treating diseases caused by bacteria since they were first discovered. However, inappropriate use and over-consumption of antibiotics lead to the emergence of antibiotic resistance, which directly impacts patients in the form of more extended illnesses, increased costs, increased mortality, and an extended stay in the hospital (WHO,2017). Besides direct impacts on human life, antimicrobial resistance has some indirect implications on the global economy as well.
Employees will take days off because of sickness for which they cannot use antibiotics due to antibacterial resistance, and this will then reduce productivity and cause a higher cost of treatment (WHO,2017).
In this study, we examine the relationship between the family medicine program and the physician-prescribed antibiotic and medicine utilization between 2008 and 2012. Turkey introduced the HTP in 2003 intending to announce Universal Health Coverage (UHC) to provide its citizens with a more equitable, protective, and quality health care system (Atun et al., 2013: 65) The UHC offers individuals, irrespective of their ability to pay, to obtain the necessary treatment. Both supply-and demand-side reforms led to an increase in access to health care services and pharmaceuticals (Stokes et al., 2015). For example, while only 66.3% of the population was covered by health insurance in 2003, this figure increased significantly to 95% in 2016.
Turkey expanded benefits coverage for public health insurance for the most impoverished population in 2005 by providing outpatient care services and reimbursing outpatient pharmaceutical expenditures, which were not covered under the plan in the past (Tirgil et al., 2018). Also, the reimbursement reform allowed employees to access to the Ministry of Health hospitals and private pharmacies (Atun et al., 2013). All these paved the way to easy access to prescribed drugs, which can cause over-utilization and inappropriate use of antimicrobial medicine Hosoglu, 2008: 1169).
Turkey piloted a family medicine program in Duzce province in 2005. Then, contract-based family medicine scaled up across the country until 2010, which aimed at providing a comprehensive set of primary health care services, including maternal and child health care services, immunization services, antenatal services, and first contact of diseases commonly seen in primary health care, through these centers. The health care services offered by family medicine are free to everyone regardless of their insurance status, and the general government budget meets the cost incurred by these services. In the meantime, easy access to primary care through family physicians may lead to over-consumption and improper use of antibiotics by the public (Ternhag et al., 2014).
Turkey was one of the countries where antibiotics over-used by its population (Karabay et al., 2011). Although antibiotics represent the most substantial proportion (11% of the total market in 2018) by volume among therapeutic groups, there has been a significant decline (33%) in its consumption since 2010 (IEIS, 2018). With the help of the HTP, mainly introducing the UHC, the Ministry of Health buyer of health care services in the market, deciding on which medicinal products will be reimbursed (Ozturk et al., 2019). In this way, reforming the health system allowed the Turkish authorities to have more control over antimicrobial utilization and other prescribed medicinal products.
They indicate that the total utilization of antibiotics increased; specifically, the most substantial increase Another critical study only looked at the socio-economic predictors of prescribed and non-prescribed medicine use in Turkey and found that non-prescribed medicine utilization has been increasing since 2012, while prescribed medicine use has been falling since then (Ozturk et al., 2019).

METHODS
In this study, we focus on two outcome variables. First, we examine the determinants of prescribed medication use. Then, we turn to our focus on the predictors of prescribed antibiotic utilization. The question asked in the survey to define prescribed medicine utilization is, 'Have you taken any prescribed medicine, including herbal drugs, vitamins, and dietary products, within the last two  The datasets cover a wide range of questions concerning the health of respondents, including overall health status, whether the person has/had any health problem expected to last six months or more, to what extent this health problem limited the respondent's daily life, job accidents in the last 12 months, and physical pain in the last four weeks. Besides health-related information, the datasets include information regarding individual-level background characteristics, including age groups, education level, household income, employment status, gender, urbanicity, insurance status, and the respondent's family physician visit.
We use pooled multivariate regression analyses by implementing Ordinary Least Squares (OLS) to analyze the determinants of prescribed medicine and antibiotic use. In this study, we also report the trend for antibiotic utilization after the UHC implementation, which unified different insurance schemes with almost the same benefits package. We cluster our regressions to obtain correct standard errors based on household identifiers that we have available in each dataset.
In our regression analyses, we use 7 age group dummies (15-24 being the reference category, insurance for workers, Bağ-Kur for self-artisans, green card for the poor, and private insurance), and family medicine visit (Yes/No). We also employ 26 statistical area dummies to control for regionspecific effects and survey year dummies in our regression analyses. Table 1 presents descriptive results. We report the summary statistics for prescribed antibiotic and medicine utilization. The sample that we use to shed some light on the determinants of prescribed antibiotic utilization is conditional on whether the respondent utilizes prescribed medicine. That is the reason why we have a smaller sample size for the antibiotic analysis.
The first two columns indicate that younger age cohorts (15-24, 25-34, and 35-44) utilize more antibiotics than their counterparts within the same age groups, and the difference is statistically significant. However, when respondents get older (greater than 55 years old), the mean of antibacterial utilization significantly declines, where more individuals say 'no' to antibiotic utilization. In the meantime, we observe contrary results for prescribed medicine within age groups. Younger cohorts consume less medicinal products, whereas older cohorts (greater than 45 years old) consume more of them.
Besides, we do not observe any statistically significant differences within educational attainments in terms of antibiotic consumption. As for the prescribed medicine utilization, except for illiterate and primary school graduates, middle school, high school, and university or more graduates indicate significantly less utilization of prescribed medicinal products within their groups. Among the Green Card enrollees, a government-funded public health insurance program for the poor, those who said 'yes' to antibiotic consumption is higher than their counterparts who said 'no.' Besides, within the retired government employees, those who say 'yes' to antibiotic utilization is less than those who state 'no.' Although it is significant at the 10% level, respondents state more utilization of antibiotics among employed. While individuals with good health state that they use less antibiotics, those with health problems which limits their daily life seem to use more antibiotics. Besides, respondents who had a work accident and physical pain consume more antibiotics. An interesting outcome is that those with a health problem expected to last six months or longer state that they consume fewer antibiotics than their counterparts, whereas they state that they use more prescribed medicinal products. Figure 1 presents the average prescribed medicine and antibiotic utilization by adults over time.

RESULTS
It is important to note here that antibiotic utilization declined dramatically in 2010 and continued declining in 2012. Unifying various insurance schemes into one could be one explanation for this declining trend because having only one buyer of health care services could produce more efficient use of antibiotics in a way that would be used more rationally since there would be more control over prescription antibiotics.
Another likely explanation to clarify why we see a dramatic decline in antibiotic consumption is that the Ministry of Health established the Department of Rational Drug Use to promote rational drug utilization and to raise public and health professionals' awareness on the danger of improper use of antibiotics across the country in 2010 (Akdağ, 2011).
The Ministry of Health also introduced a central system named MEDULA. This new system allows doctors to see their patients' past treatment and medicinal records, which help them avoid prescribing unnecessary drugs or avoid prescribing the same or similar medicinal products. MEDULA helps the Ministry of Health to have control over pharmaceutical expenditures and to track medicinal

Socio-Economic Determinants Of Prescription Antibiotic And Medicine Use And Its
prescriptions. We believe that MEDULA helped reduce prescribed antibiotic utilization throughout the country.
On the other hand, drug prices have declined over time in Turkey, which would increase the quantity demanded of prescribed medicinal products. However, we observe a somewhat constant trend over the studied period, increasing first in 2010, then declining afterward. It appears that the interventions mentioned above have canceled out the effect of the price decline in the quantity demanded of pharmaceutical products.   Figure 3 illustrates some pharmaceutical products in terms of unit scale by years. It is shown that antibiotics products are leading therapeutic groups, among others, by 16,4% in Turkey, which went down to 11% in 2018 among treatment groups. This outcome shows us that the declining trend of antibiotics utilization that resulted from the precautions mentioned above has still been going. However, although its consumption has been declining among therapeutic groups, antibiotics are still the most preferred drug type by 2018 in Turkey.

Figure 3. Treatment Groups On Unit Scale
Our interest lies in two subjects, which are prescribed medicine and prescribed antibiotic consumption. The first column of Table 2 reports the determinants of medicine utilization in the studied period. We show that medicine consumption is higher for older cohorts than the younger cohort aged 15-24. We expect this result because as people grow older, they will have more chronic conditions for which they may need more prescribed medicine.
Although the coefficients are not significantly different from zero for most education levels, those with a high school degree or more seem to consume more prescribed medicine than illiterate individuals. We also observe that employed individuals consume less prescribed medicine, which may result from not being able to go to a doctor during the day. Those who state that they have health problems, limited daily life due to health issues, and feeling physical pain consume more physicianprescribed medicine. In terms of insurance groups, they are all likely to consume more prescribed medicine than the uninsured group as expected because those with formal insurance will not incur the full cost of pharmaceuticals.  Notes: Robust standard errors in parentheses. * p < .1, ** p < .05, *** p < .01. We cluster standard errors at the household level. In all regression models, we control for 26 statistical area dummies. While we control for different insurance groups in all our analyses, we do not report their estimates here. Instead, we show the antibiotic use patterns over time for various insurance groups in Figure 2 in the text.
As for the physician-prescribed antibiotic consumption, older age cohorts consume fewer antibiotics, indicating that doctors may be concerned about the potentially harmful effects of antibiotics to the elderly through antibiotic resistance (Filippini et al., 2006).
Education levels do not produce statistically significant results in antibiotic consumption. It seems that more income yields significant and positive estimates for antibiotic consumption. Besides, we find that income is positively related to more prescribed medicine utilization. What is interesting here that although female respondents state they use more physician-prescribed medicine, they report less consumption of antibiotics than their male counterparts, which can be explained by the fact that doctors do not want to prescribe antibiotics when women are pregnant or expected to be pregnant because this would harm the baby.
Urban residents seem to use more antibiotics than rural ones, which may be an indication of easy access to physicians and pharmacies. Although individuals who report having health problems expected to last six months indicate using more medicine, they seem to consume less of antibiotics, which may be an indication of using other resources than antimicrobial drugs to deal with their health issues. It seems that people who report having good health consume less prescribed medicine and antibiotics.
Another significant result from our analysis is that year dummies show that antibiotic use declines in 2010 and 2012 dramatically compared to 2008. Finally, there is a positive relationship between family medicine visits and antibiotics use, although the aim of establishing a family medicine program was to reduce inappropriately-and over-consumption of antibiotics. However, significant interaction terms indicate that the positive relationship between family medicine visits and antibiotic consumption weakened over the years, indicating that family practitioners are becoming more aware when writing antibiotics.

DISCUSSION
In this study, we explore the determinants and consumption patterns of prescribed antibiotics and medicine in Turkey by using large and nationally representative Turkey's Health Surveys from 2008 to 2012. We also examine how the relationship between family medicine visits and antibiotics evolve over the years. To the best of our knowledge, this is the first study investigating the determinants of antibacterial medicine utilization before and after the implementation of the UHC, using nationally representative datasets from Turkey. We uniquely provide essential information on the relationship between antibiotics use and family medicine visits in Turkey.
Antimicrobial-drug resistance has indirect implications as well as direct consequences. One of the indirect adverse effects includes the constraint on the limited health care budgets of countries, which is caused by over-use and improper utilization of antimicrobial medicines (McGowan, 2001: 286).
Although Turkey has made significant progress in taking precautions against antibiotic consumption as part of the HTP, improper use of antibiotics still poses a threat to the health system budget because

999
Turkey is a country where antibiotic consumption and antimicrobial drug resistance is still high. In line with this, a report indicates Turkey will incur an economic loss between $220 billion and $1.4 trillion by 2050 because of antibiotics' antimicrobial resistance and over-consumption (TEPAV, 2017).
Findings indicate that those who visit family medicine are more likely to use antibiotics, which is consistent with the literature by Ternhag and colleagues (Ternhag et al., 2014). They report that easy access to primary care health centers leads to more antibiotic prescriptions. Another critical study indicates that family physicians write the most prescriptions containing antibiotics compared to specialists in other areas such as pediatrics and internal medicine (Mollahaliloğlu et al., 2013: 281).
However, the positive relationship between antibiotic use and family practitioners has weakened, indicating that Turkey has promoted rational drug use over the years through the means of family medicine. Besides, Turkey still needs to promote rational drug use awareness and effectively take some policy actions through the Department of Rational Drug Use to reduce inappropriate use of antibiotics and antimicrobial resistance. For this purpose, policymakers can implement policies such as legislative regulations, including rational drug use in primary education curricula, medical and pharmacy faculties.
Family physicians are the first contact of most individuals when they get sick. Along the way, family physicians were educated on the danger of antibiotic utilization. However, it still seems a significant issue in terms of prescribing antibiotics. Therefore, special attention needs to be given to these family medicines to reduce inappropriate and over-consumption of antibacterial.
Our estimations report that antibacterial drug utilization declined over time between 2008 and 2012. The dramatic decline we observe was in 2010, and the decline continued afterward. We could explain this result by seeing patients' records on drug utilization through the MEDULA system helped doctors avoid prescribing unnecessary antibacterial. MEDULA also allowed the Ministry of Health to observe physicians' attitudes towards prescribing antimicrobial medicine, and if needed, they could intervene with the process to reduce antibacterial resistance. Another likely explanation is that different insurance schemes, including Social Insurance Organisation, Bağ-Kur, and the General Employees Retirement Fund, were unified under Social Security Institution, which was established as a single organization for financial pooling and purchasing (Atun et al. 2013). Introducing rational drug use programs throughout the country was also beneficial to reduce antibacterial resistance.
Income level increases the prescribed antibiotic consumption, which contradicts with the literature (Filippini et al., 2006, Covvey et al., 2014. Our result suggests that wealthier households in Turkey spend more on antibiotics compared to lower-income households. A likely channel to explain this result is that as high-income individuals will not worry about the high cost of antibiotics, doctors may tend to prescribe them with more antibacterial drugs. Therefore, policy action is needed to intervene in these situations, where inappropriate use of antibiotics occurs.
Although most of the coefficients show the expected sign, education levels do not have a significant relationship with the prescribed antibacterial utilization. Finding may imply that educated people in Turkey are not aware of the perverse effects of antibiotic resistance. These finding urges policymakers to provide students early in life with more information on the danger of antibiotic utilization.
We include age groups to analyze the effects of demography on prescribed medicine and antibiotic use. Our findings report that while older cohorts utilize more prescribed drugs compared to younger cohorts, they consume significantly less of prescribed antibiotics (Marra et al., 2010). A channel to explain this finding could be that doctors may stay away from prescribing antibiotics to older patients since antibacterial resistance may have potentially harmful effects on adults (Filippini et al., 2006: 77).
We can also explain this outcome by the fact that since using multiple medicines is common for older people; doctors think twice when prescribing antibiotics to these people because there may be side effects or complications other than antibacterial resistance.
Another impressive result that came out from Table 2 is that those with a health problem lasted at least six months indicates that while they use more prescribed medicine than their counterparts without a health problem, their utilization of antibiotics is much less. One explanation is that since these people may have chronic conditions, they need to go to a hospital to see a specialist because that is the only place where they can get better treatment for their chronic condition than family physicians.
Specialists may be better aware of the danger of prescribing antibacterial drugs to these people on their health.
Employed individuals seem to be using fewer antibiotics than unemployed; those with good health consume less of antibiotics, those who state that they had a limited daily life due to a health condition are more likely to be using antibiotics than those who state that they did not, having work accident and physical pain correlate with more antibacterial drug utilization.
People in urban areas are likely to use more prescribed antimicrobial drugs. A channel to explain this outcome is that urban dwellers have easy access to pharmacies and health care facilities. We need to keep in mind that since there is no referral system in Turkey, it could be that individuals in urban areas may prefer going to any health institution (public or private) to obtain what they need quickly. In terms of prescribed medicine, urban dwellers are likely to use more of them than rural residents. Turkey also introduced mobile pharmacy services in 2009 to ease access to pharmaceutical products for those who live in a rural neighborhood. However, according to our results, there are significant differences between urban and rural residents in terms of prescribed drug utilization. Therefore, policymakers need to address this problem and maybe strengthen the mobile pharmacy system to have rural people easily reach pharmaceuticals.

CONCLUSION
This study decomposed prescribed medicine utilization into whether the person consumes prescribed antibiotics or not and study its determinants. We have demonstrated a significant and positive association between family medicine visits and antibiotic consumption, which indicates easy access to primary care health centers that are scaled up in 2010 throughout Turkey. However, our results also show that family medicine visits are associated with less prescribed antibiotics consumption over the years. Besides, family medicine visits are related to more prescribed medicine utilization.
Our study has some limitations. For example, not controlling doctors' characteristics may lead to some spurious correlation while examining the relationship between prescribed medicine and antibiotics utilization and individuals' socio-economic characteristics. When data made available, it would be critical to include doctors' characteristics into regression analyses to understand the relationship between our outcome variables and main control variables. Notwithstanding limitations, our results will contribute to the existing literature on health economics, mainly revealing a significant correlation between family medicine programs and antibiotics consumption.
The declining trend in antibiotics use over the studied period indicates that Turkey's interventions helped reduce antibiotics utilization. Specifically, educating family medicine practitioners on the danger of inappropriate and over-use of antimicrobials is of great importance to policymakers in countries with limited health care budgets. Therefore, continuing to implement these precautions and enforcing them will help reduce the improper use of antibiotics. Turkey still needs to continue taking severe actions to reduce inappropriate use and over-consumption of antimicrobial drugs and antibacterial resistance. For future studies, examining what types of antibiotics have been prescribed in different regions of Turkey would be useful for policymaking when the data provides this information, which is currently unavailable.