Significant individual differences were observed among GPs regarding the number of prescriptions, prescribing rates per 100 patients, and the share of prescriptions for UTIs in the total number of antibiotics for systemic use. Years in service and completed specialization in family medicine did not significantly affect either the total number of prescriptions or the prescribing rates per 100 patients.
The percentage of prescriptions for UTIs in the total outpatient consumption varies, according to data from the literature, between 15% and 20%, but with large differences. Several studies of outpatient antimicrobial consumption based on electronic databases recorded 15% (20), 4.6% (21), and 22.7% (22) prescriptions related to UTIs. A study of antimicrobial consumption for genitourinary infections in women in five European countries (UK, Spain, Germany, the Netherlands, and Denmark) (23) recorded proportions between 8% − 16% (UK), 14% (Spain), 18% (Germany), 10% − 22% (Netherlands), and 19% (Denmark). In our study, 22% of prescriptions were issued for uncomplicated and complicated infections in both sexes, which roughly corresponds to the UK study of Dolk and al. (22), and 14.8% for uLUTIs in women over 18 years of age. Antimicrobial utilization showed large fluctuations between the lowest and highest prescribers, verified also by large interquartile ranges in both utilization segments.
To reduce the impact of the list size (varying almost three times), the consumption was expressed in rates per 100 patients. Additionally, the percentages of antibiotics for UTIs compared to all antibiotics for systemic use were analyzed, for each individual GP. The difference between the highest and lowest rates per 100 patients for all UTIs as well as for uLUTIs, as shown in Table 1, was about 6 times, while the proportion of prescriptions for UTIs compared to the total consumption varied slightly over 7 times. Some GPs showed excessively high percentages: team22 used almost half of all antibiotics (48.4%) for urinary infections, while its prescribing rate per 100 (8.08/100) was relatively inconspicuous (compared to the highest rate, 14.6/100). If assessed by only this criterion, we could say that he was an "economical prescriber". As opposed to team22, team10 had the lowest percentage (6.6%), or about 7 times less, while the list size was among the largest (2006 patients), and the second low-prescriber, team12 (9%) had 1335 patients in care. These figures, probably more than any other indicator, illustrate differences in attitudes and knowledge between individual GPs as potentially the most important factor explaining variations. It is difficult to explain six times variations (in rates per 100 patients) in uUTIs because these infections are characterized by recognizable clinical symptoms, absence of major comorbidities, and relatively simple diagnostics. Every GP and his patients form a complex system in the context of prescribing (24), which necessitates consideration of many variables in the analysis of each individual prescribing model.
We additionally compared prescribing in uLUTIs in women over 18 years with the ESAC quality indicators. Out of 3581 prescriptions for uLUTIs, 51% referred to first-line antibiotics, 17.9% to fluoroquinolones, and 30.3% to beta-lactams, which indicates sub-optimal compliance with both the quality indicators and national guidelines (12). Among the first-line antibiotics, there were 31% prescriptions for nitrofurantoin, 12.5% for fosfomycin, and 8% for co-trimoxazole. Relatively poor adherence to guidelines is found in several other studies. In a two-month study by Wellinga et al. the most prescribed drugs were co-amoxiclav (33.1%), followed by trimethoprim (26%), fluoroquinolones (17%), and nitrofurantoin (12%) (25), while a study from Spain reported fosfomycin (47.1%), fluoroquinolones - ciprofloxacin and norfloxacin (21.7%), co-amoxiclav (15.3%), and only 0.3% of nitrofurantoin (26). One of the main reasons for differences in the choice of antibiotics was, according to Llor and Bjerrum, professional attitudes and traditions regarding the treatment of infections in individual countries (27). The proportion of beta-lactams (co-amoxiclav and cephalosporins) in our research was about 30%, similar to the result in a study investigating the appropriateness of treatment based on urine cultures (25). In both studies, GPs equally favored beta-lactams in an indication where they should actually be alternative drugs.
Quality indicators are a helpful tool in assessing the appropriateness of prescriptive practice and are generally in line with the recommendations of the guidelines (19). According to indicator "b" - the share of first-line antibiotics, our result was 51% (desirable range: 80%-100%), while indicator "c" (fluoroquinolones) was 17.9% (desirable range: 5%), thus significantly exceeding the recommended limit. Only one GP reached the level of first-line antibiotics above 80%, while the majority (25 teams or 59.5%) prescribed in the range of 50% − 80%, but still prescribing more than a half of the recommended antibiotics for uLUTIs. About a third of GPs (15 teams) prescribed mostly alternative antibiotics, thus clearly proving their erroneous attitudes that all the three antibiotic groups are much the same regarding efficacy and potential for collateral damage.
The relationship between specialization in family medicine and years in service with the number of prescriptions per 100 patients for uLUTIs was analyzed by Mann-Whitney U test. The lack of statistically significant correlation between the clinical experience and the prescriptions rate per 100 (P = 0.51) indicates that the experience per se is not among the most relevant factors shaping the prescriptive model. "Experienced GPs”, with 15–30 years in service showed the prescribing pattern not much different from colleagues with less experience in family practice. But this conclusion is uncertain because data on the actual number of visits to each GP were not recorded, and thus the number of visits without antibiotics is unknown. The relationship between prescribing rates and specialization also did not reach statistical significance (P = 0.51). Lopez-Vasquez et al. have found a marginal impact of specialization on the extent of antibiotic prescribing (28). GPs with completed specialization did not significantly differ in the proportion of first-line antibiotics and fluoroquinolones compared to GPs without specialization.
Cutting down on antibiotic use in UTIs lead to a decrease in local levels of resistance (29). Given the growing trend of E. coli resistance in Croatia, prescribing nitrofurantoin (after considering contraindications) as the drug of choice for uLUTIs caused by both susceptible and resistant strains of E. coli is of key importance. Fluoroquinolone resistance in Enterobacteriaceae is on the rise globally, exceeding 50% in some parts of the world (30). About one to two-thirds of these strains produce extended-spectrum beta-lactamases, making them resistant to several important antibiotic subgroups: fluoroquinolones, beta-lactams, and co-trimoxazole. Nitrofurantoin has retained efficacy in uLUTIs caused by these bacteria and therefore represents the drug of choice for resistant infections in the future
Strengths and limitations
As our primary goal was to analyze the prescribing pattern in UTIs, diagnostic verification of symptoms (test strip and sediment, urine culture), identification of the causative pathogen, and antimicrobial susceptibility were not included in the follow-up. A too-small sample of doctors and the lack of data to confirm the diagnosis of urinary tract infection by objective methods (test strip, urine culture) are the main disadvantages. Due to the lack of data on bacterial pathogens, we could not determine the impact of resistant microorganisms on the number of prescriptions. The main advantage of our research is in the analysis of prescribing based on routine work in GPs’ offices, as opposed to data from pharmacies or wholesalers.