The association between healthcare expenditures and potentially inappropriate medication use in hospitalized older adults in Ethiopia

This prospective observational study was conducted with the primary intention of generating local evidence on the association between the frequency of PIM use and healthcare expenditure in hospitalized older adult patients in Ethiopia. Describing the prevalence of PIM, specific types of PIMs prescribed, specific categories of healthcare expenditure (bed day cost, medication cost, medical supply cost, and investigation cost), and and other factors associated with the total healthcare expenditure were the secondary objectives of the study. Data were collected from September 6, 2021, to December 30, 2022. The study was conducted in the medical wards of Jimma University Medical Centre (JUMC). The JUMC was established in 1930 in Jimma town, 352 km southwest of Addis Ababa. It is the only teaching and referral hospital in the southwestern part of Ethiopia, with a bed capacity of 800. It provides services for approximately 16,000 inpatient and 220,000 outpatient clients a year with a catchment population of approximately 15 million people [28].

All older adult patients admitted to the medical wards of JUMC were considered as the source population, and the study participants were consented older adult patients (60 years and above) admitted to the medical wards (general unit, cardiac unit, stroke unit, and pulmonary unit).

After an in-depth review of the relevant literature, the data collection tool was designed to capture the sociodemographic, behavioural, functional, clinical, medication, healthcare expenditures, and related information of the participants. The tool was translated into two locally predominant languages (Afan Oromo and Amharic) and back-translated into English to check the consistency. The data collectors (two pharmacists with master's degrees in clinical pharmacy and one nurse with a bachelor's degree) were trained on the data collection tool and procedure. A pretest was conducted before the actual data collection. The investigators closely supervised the data collection procedure throughout the study period. All eligible patients were enrolled at admission to the ward and followed strictly during the hospital stay. Patient charts, laboratory results, patient/caregiver interviews, and practitioners in charge were the sources of data for the present study. Body mass index (BMI) was calculated using the standard formula, BMI = weight in kg/ (height in m) ². The weight of the study participants was measured using a beam balance. For taking the weight of the participant, each of them was requested to take off their shoes and heavy clothes, including jackets, jerseys, and belts; the balance was calibrated, and the figure was approximated to the nearest 0.01 kg. For measuring height, a seca vertical height scale was used. Patients were asked to take off their shoes and stand upright in the middle of the board. Thereafter, height was measured to the nearest 0.01 cm by making the participant`s back of the head, the shoulder blades, the buttocks, and the heels touch the measuring board. Upon encountering patients who cannot erect upright because of their illness, their height was estimated from demi-span. In line with a standard recommendation [29], the length from the sternal notch to the finger roots (demi-span) was measured by a flexible validated plastic tape with the patient sitting upright or recumbent depending on the patient’s abilities. Height was estimated as follows: for females: Height in cm = (1.35 × demi-span in cm) + 60–1, for males: Height in cm = (1- 40 × demi-span in cm) + 57.8.

The independent variables were sociodemographic and related variables (age, sex, residence, marital status, educational level, patient’s current working status, current occupation, and financial dependence), behavioural, functional, and related information (alcohol consumption history, cigarette smoking history, khat chewing history, recent traditional medicine use history, cohabitation, activities of daily living, body mass index, kg/m2), and clinical and related information (past medical history, hospitalization in the previous year, psychological condition at admission, current diagnoses, number of diseases diagnosed, Charlson comorbidity index (CCI), length of hospital stay, and medication and related information (past medication history, in-hospital medications, polypharmacy (≥ 5 medications per patient)). The Katz Index of Independence in Activity of Daily Living (ADL) was employed to assess functional health status (disability) at admission [30]. The tool ranks adequacy of performance in six functions (eating, dressing, bathing, transferring, continence and toileting), with each rank assigned a score of 1 or 0. The overall patient ranking is a Katz score of 6 = independent (full function), 3–5 = partially dependent (moderate impairment), and 2 or fewer points = dependent (severe functional impairment) [30, 31]. Psychological condition at admission was objectively assessed using the shortened self-report form of the Geriatric Depression Scale (GDS), which comprises 15 items [32]. Patients respond in a "Yes/No" format. The patients are categorized as having no psychological problem (0 to 4), mild depression/dementia (5 to 9) or severe depression/dementia (10 to 15). Diseases were categorized according to the International Classification of Diseases (ICD)-11 system [33], while MDCalc online was employed to calculate the CCI score [34]. Patients were categorized into mild, with CCI scores of 1–2; moderate, with CCI scores of 3–4; and severe, with CCI scores of ≥ 5 [35]. The Anatomical Therapeutic Chemical (ATC) system is employed to categorize medications [36].

The primary objective of the study is to assess the presence of an association between PIM frequency and total healthcare expenditure. Total healthcare expenditure was operationalized as the sum of medication/prescription cost, bed cost, investigation cost, and cost of medical supplies. Prevalence of PIM, specific types of PIMs prescribed, categories of healthcare expenditure (bed day cost, medication cost, medical supply cost, and investigation cost), and other factors associated with the total healthcare expenditure are the secondary outcomes.

From a patient perspective, the cost valuation was carried out by a direct allocation method. The study considered healthcare expenditure costs patients spent to gain service and/or items at the inpatient level. Bed days expenditure was calculated by multiplying the bed days of the patient by bed cost/day. Medication (prescription) cost, medical supply cost, and investigation cost were each determined by summing up all costs the patient spent buying medications and medical supplies and undergoing investigations during the whole stay in the medical wards. In the present study, the term investigation expenditure refers to the payment made by the patient to undergo imaging studies (such as X-ray, CT scan, magnetic resonance imaging, ultrasound, and endoscopy) and laboratory tests. For patients who have health insurance coverage, all corresponding healthcare expenditures were calculated by allocating the price of each item (medication, medical supply) and services (bed, laboratory and imaging investigations). For such individuals, the price for services and items was allocated per the centre`s price lists obtained from the pharmacy unit, laboratory/investigation unit, and finance office. Finally, the total healthcare expenditure was determined by summing the bed day cost, medication cost, medical supply cost, and investigation cost in Ethiopian birr (ETB) over the hospital stay. The cost program medications, such as anti-tuberculosis and anti-retroviral medications, were not included in the medication cost estimation. According to the National Bank of Ethiopia`s report, the commercial bank`s weighted average exchange rate of one US dollar to ETB at the start of this study (06 September 2021) was $1 = 45.6011 ETB, while it was $1 = 53.2566 ETB at the end of the study period (30 December 2022). On 01 March 2023, the exchange rate was $1 = 53.7535 ETB [37]. The 2019 updated AGS Beers Criteria® [38] was employed to assess PIM use. This criteria is used as a guide to improve medication selection, educate clinicians and patients, reduce adverse drug events, and evaluate quality of care, cost, and patterns of drug use among older adults. The Beers tool consists of the following criteria: medications that are potentially inappropriate in older adults, those that should typically be avoided in older adults with certain conditions, drugs to use with caution, drug-drug interactions, and drug dose adjustment based on kidney function. Though PIM can also be assessed using tools other than Beers, in the present study, Beers tool was used to incorporate a more extensive list of PIMs as compared to others. The AGS Beers Criteria are also widely used by healthcare providers, researchers, and educators. This tool has been used in most previous studies from Ethiopia [21, 23, 24, 39, 40]. It should be noted that, while using the AGS Beers Criteria®, clinicians should consider patient-specific factors when determining if a medication is to be discontinued, modified, or added.

From the patient registration book review, the number of older adult patients aged 60 years and above admitted to the medical wards of JUMC in a year (2019–2020) just before the study period was 398. Since there were no locally available previous studies on the association between healthcare expenditure and PIM frequency, with the assumption of ensuring adequate power of the study, we planned to include all eligible older adults who will be admitted over the period of one year starting from September 06, 2021. However, during the actual study period (a period during which COVID-19 was a challenge to the healthcare system), there were very limited admissions of older adults to the medical wards of JUMC. Therefore, we could not achieve the number of participants we planned even after extending the study for four months (one year and four months). A total of 176 hospitalized older adult patients were consecutively approached over the study period and 25 were rationally excluded from the study (2 patients with aphasia problem, 12 readmissions, and 11 patients lost to follow-up). Finally, 151 consented older adult patients admitted to the medical wards of JUMC in the study period were considered for analysis.

Data were checked for completeness and accuracy regularly during the data collection period, entry, and before the analysis. Epi data version 4.2.0.0 and STATA V.15.0 were employed for data entry and analysis, respectively. Categorical variables were described using frequency and percentage. Performing the Shapiro–Wilk test, the distribution of continuous variables does not follow a normal (Gaussian) distribution, thus described in median and interquartile (IQ). The Kruskal‒Wallis test was conducted to evaluate the distribution of healthcare expenditures across the frequency of PIMs taken by the patients over the hospital stay. The healthcare expenditure was right skewed in distribution toward larger values. For such kinds of variables with positive scale values with positively-skewed data, it is reasonable to use a generalized linear model (GLM) with a log link and gamma variance. Hence, a GLM with a log link and gamma variance was conducted to determine the association between healthcare expenditure and the number of PIMs. Covariates with a p-value < 0.25 in bivariable regression were used to adjust the association between healthcare expenditure and PIMs in multivariate regression. Multicollinearity was checked using the variance inflation factor (VIF), and all factors included in the multivariate regression had a VIF of < 10 (maximum 2.16). In all the analyses, a p-value of < 0.05 was used to declare statistical significance.

Of the total, at least one PIM use was identified in 111 (73.5%) participants. Furosemide, 77 (43%), was the most frequently used PIM. Clinicians are recommended to strictly monitor sodium level during furosemide initiation and dose alteration in older adults. The weak opioid analgesic, tramadol, 25 (14.5%), was the second most frequently used PIM. In general, cautious use of tramadol is recommended in older patients due to its association with the risk of hyponatremia. On the other hand, tramadol use in older patients with CrCl < 30 ml/min is not recommended because of its CNS adverse effect, and it was used in one patient with severely compromised CrCl (Table 3).

Over their stay in the medical wards, the participants spent a total of 516,636.3 ETB for health care utilization (Fig. 1).

The expenditure was higher in those who took PIMs than in those who did not (385,368.6 ETB versus 131,267.7 ETB). The median expenditures for medical supplies (p = 0.025) and investigations (p = 0.033), as well as the total expenditure (p = 0.023), were significantly different across the number of PIMs (Table 4).

In bivariate regression, a significant association was observed between the total healthcare expenditure of older adults and the number of PIMs prescribed (p = 0.044). However, this association was lost after adjusting for other factors in the multivariate regression analysis. A 30.7% and 19.6% decrease in healthcare expenditure was observed in patients with diseases of the nervous system (p = 0.002) and the respiratory system (p = 0.003), respectively, as compared to those without. On the other hand, each additional number of prescribed in-hospital medications (p < 0.001) and taking blood and blood-forming organ class of medication (p = 0.042) are associated with a 6.8% and 15.1% increase in healthcare expenditure (Table 5).