Sample size
The required sample size for this study was calculated assuming a 50% proportion of MRPs, 5% level of precision, 3% error margin, and 5% possible non-response rate, making the minimum sample size 1,121 pregnant women.
Study setting
A facility-based prospective observational study was conducted in the maternity and gynaecology wards of JUMC, a tertiary level public teaching hospital located in Jimma City in southwest Ethiopia, 350 km from the capital city of Addis Ababa. It is the only teaching and tertiary level care hospital in southwest Ethiopia, with a catchment population of approximately 20 million people [18, 19]. Most of the pregnant women referred to the hospital come from rural areas, where many deliveries are attended at home [20, 21]. The Department of Obstetrics and Gynaecology at JUMC provides specialized health services for approximately 7,580 inpatients and 11,590 outpatients each year, with a bed capacity of 265. The department has two wards (gynaecology and maternity/labour), one antenatal care outpatient clinic, one general gynaecological outpatient clinic, and one family planning clinic. Women are treated at the gynaecology inpatient ward before 28 weeks of pregnancy. Most pregnant women admitted to this ward have elective and/or spontaneous abortions, hyperemesis gravidarum (HEG), or other early pregnancy complications. After 28 weeks of pregnancy, women are admitted to the maternity/labour inpatient ward. Women having a vaginal delivery give birth in the labour ward and are transferred to the maternity ward after delivery. If the mother and baby are healthy, they are discharged at the earliest possible time after delivery, usually within 1-2 days. Women having a caesarean delivery are transferred to the maternity ward and usually stay for 72 hours.
Data collection and procedures
Women in the maternity and gynaecology wards at JUMC between February and June 2017 were invited to participate in the study during normal working hours. Patients were informed of the aim and procedures of the study, and written informed consent was obtained from each study participant. Women who were under 18 years of age, too ill to participate, who declined to participate, were hard of hearing, unable to speak or with mental illness, admitted for a brief time (<4 hours), and non-pregnant women admitted to the gynaecology ward were excluded from the study.
The women were followed throughout their stay in the hospital to assess the presence and development of MRPs. Pre-tested data extraction form and an interview-guided structured questionnaire were used to collect the data. Five trained clinical pharmacists (data abstraction and MRP assessment) and four trained nurses (the questionnaire) from JUMC collected the data.
Information on the reason for admission, diagnoses, dosage regimens, discharge medications, maternal and perinatal outcomes, laboratory results, and length of hospital stay was collected by reviewing patients’ medical cards and medication charts. The card and chart reviews were performed for each patient on the first day of admission and repeated on subsequent days. The questionnaire was used to collect maternal socio-demographic characteristics, obstetric history, past medical history and medication experience, social drug use, and medicinal plant use.
MRP identification and assessment
MRPs were classified into eight categories: need for additional drug therapy, unnecessary drug therapy, dose too low, dose too high, ineffective drug, adverse drug reactions, noncompliance [1], and other, subdivided into need for additional laboratory test and/or incomplete drug order (Additional file 1) [3].
MRPs were identified by reviewing patients’ medical cards and medication charts, and patient interviews about medication use while in the hospital. A panel of experts comprised of senior clinical pharmacists and experienced obstetricians/gynaecologists identified MRPs and classified them into categories as recommended by Cipolle et al. [22]. The panel of experts further refined the MRP identification and classification method for the study setting in accordance with Ethiopian standard treatment guidelines and literature reviews (Additional file 1) [3, 23-26].
The clinical significance of each MRP was classified as low (level 1) or moderate/severe (level 2) [3]. Clinical significance was initially assessed by clinical pharmacists recording the MRP, and subsequently discussed and reviewed by the panel of experts. The description of the MRPs, their clinical significance, and the medication(s) involved were recorded using a purpose-built data collection tool.
The classification of medications involved in MRPs was performed per the World Health Organization (WHO) Anatomical Therapeutic Chemical Classification system (ATC) that categorizes medications into 14 main groups [28].
Statistical analysis
Descriptive statistics were used to calculate percentages. The results were presented as medians and ranges. Univariate and multivariate logistic regression analyses were used to calculate odds ratios (ORs) with 95% confidence intervals (CIs) and identify risk factors associated with MRPs. The independent variables were patient-related factors (age, level of education, marital status, occupation, religion, ethnic group, family size, residence place, alcohol use status, and khat chewing), disease-related factors (patient admission ward, i.e., gynaecology or maternity ward; chronic disease; obstetrics category, i.e., caesarean or vaginal delivery; duration of hospital stay), pregnancy related factors (parity, gravidity, gestational age, adverse pregnancy outcome [current and previous], status of anaemia), medicine-related factors (medicines used during admission or prior to admission, ferrous sulphate supplementation, medicinal plant use, concomitant use of medicinal plants), facility-related factors (walking distance to the nearest health facility, and availability of preferred medication for a specific condition). Explanatory variables with p ≤ 0.05 in the univariate analysis were entered into a multivariate logistic regression model to determine independent risk factors of MRPs. As iron was involved in 165 (41.9%) of the MRPs, a post hoc logistic regression analysis was performed excluding iron. All data were analysed using the Statistical Package for the Social Sciences (SPSS) software version 25.0 for Windows (IBM® SPSS® Statistics, Armonk).