Study design
In this descriptive-analytical study, 1068 prescriptions by 48 family physicians in PHC were selected among total 90115 prescriptions from total of 8 rural, 8 urban and 3 urban-rural health centers (sampling frame) from September 2012 to September 2013 in Savojbolagh, Alborz province, Iran.
We selected 1068 prescriptions by simple random sampling and the equal proportion in the all seasons (267 prescriptions in each season). Due to different distribution (coverage) of health insurances in Iran and also the study area, the prescriptions (samples) were selected proportional to the size of population under coverage of each health insurance. Therefore, after estimating the total sample size, the samples were selected as social security health insurance 534 (50%), rural health insurance 428 (40%), therapeutic insurance 83 (8%), and armed forces health insurance 23 (2%).
We had all antibiotic prescriptions list, therefore in the sampling process, the unreadable or prescriptions with poor information were excluded and the next prescriptions were replaced. The sample size was estimated in 1068 by considering the prevalence of antibiotic prescription by family physician (P=0.4), type I error (α=0.05), and the precision (d=0.03) by Cochran formula .
Data collection
Incorrect or mal-prescription of antibiotics was assessed based on four scientific criteria including a) dose per consumption, b) dose per day, c) duration of therapy and d) possible interaction with other antibiotics or drugs. The incorrect prescription was defined if at least one of the items above is not satisfied. Assessment of the prescriptions was performed by a high expert pharmacist (more than 10 years’ experience) who was not involved in the study or analysis of the outcome measure. The assessment was based on Martindale: The Complete Drug Reference. For this purpose, firstly the symptoms and prescribed antibiotics according to the type of diagnosis (severity of the disease) were extracted from the prescriptions and health records by the clinician and the pharmacist.
Based on the standard guidelines, each drug and disease has a standard dose per day, duration of therapy, interaction with other drugs or antibiotics in accordance with various ages, genders, weight and disease. The appropriateness of antibiotic prescriptions were assessed regarding the related guideline and the mentioned four criteria, then the incorrect prescriptions were identified for each prescription.
Same number of prescriptions were selected in each season of the year. The dose of all drugs and antibiotics per each consumption, day, duration, possible interaction based on the type of disease stage, and diagnosis were determined by the Complete Drug Reference. The dose of antibiotics (different format) calculated according to Milligram (mg) for edible (oral) drugs (such as tablet, capsule) and Milliliter (ml) for suspension, ampoule and infusion drugs. The number of drops and the speed of liquid infusion was calculated with this formula: . The drop factor was special for each drug.
A checklist was used for data collection. Variables and information such as age, gender of physicians and patients, number and total price of drug items and the name and type of the prescribed antibiotic, form and usage method of drug, consumption way based on amount of use for each time, duration of treatment course, times of use each day, possible interaction with antibiotics or drugs, rate of combination therapy, type of physicians’ graduation (private or government universities) and also the status of occupation and the working experience (years) were extracted from prescriptions and personnel files.
The government medical universities are classified in 3 levels based on annual scientific report of Iranian Ministry of Health and Medical Education.
Data analysis
SPSS software (version 18.0, Chicago, IL, USA) was used for data analysis. For checking data normality, the Kolmogorov-Smirnov test was used. Chi-square test was used for binary variables and Independent Samples T-test was used for normal quantitative variables. Logistic regression was used to estimate the odds ratio and 95% confidence interval for the possible association between antibiotic prescription and the affecting factors. P-value <0.05 was considered significant in all of the tests.