Development of a List of Potentially Inappropriate Medications for Patients with Heart Failure (PIMHF) and Study of Association Between PIMHF and Clinical Outcomes

Abstract

Heart failure (HF) is a chronic condition that still has a high mortality and hospitalization. Among causes of in-hospital death and hospitalization, worsening HF was reported to be the leading cause. It can be caused by certain medications that can adversely affect cardiac functions, e.g., causes reduced cardiac contractility and rate. Numerous medications are considered potentially inappropriate for HF patients and called potentially inappropriate medications for HF patients (PIMHF). Several clinical tools for PIMHF detection, including HF guidelines and explicit criteria on HF such as Beers criteria and STOPP criteria, suggested different PIMHF items. Thus, a comprehensive list of PIMHF is still lacking. Also, a PIMHF list should be developed particularly for Thai HF patients. The objectives of this study included (1) develop a PIMHF list for Thai HF patients and (2) examine the association of prescription of any PIMHF with HF-related in-hospital death or hospitalization. The study was divided into 2 phases in accordance with each objective. In phase 1, our list of PIMHF was created with the use of literature reviews of clinical tools on PIMHF, including HF practice guidelines and explicit criteria, and assessed using a three-round Delphi survey of a panel of HF experts. The invited participants consisted of Thai experts in HF who had ≥5 years of experience on caring for HF patients in Thailand. The PIMHF-likely medications were collected to create an initial list. A survey was carried out starting from January 1, 2019 and ended on September 30, 2019. In Delphi round 1, medication items that ≥60% of our participants considered as ‘not a PIMHF’ were discarded. An agreement on PIMHF was assessed in the next two rounds using the questionnaire with a five-point Likert scale, from strongly disagree (1) to strongly agree (5). A consensus was defined using the pre-specified criteria, including convergence (the median 3.5 and interquartile range 1.5), and stability (the difference in medians between round 2 and round 3 <15%). In a Delphi survey, the eligible participants consisted of 10 cardiologists, 4 academic pharmacists, and 3 hospital pharmacists. One hundred medication items gathered from relevant sources were included in the initial list. From round 1, the number of excluded medication items was 8, and no additional PIMHF items suggested by our participants were added to the list. Thus, 92 medication items were then assessed in the next two rounds. After the surveys had been completed, the consensual 47 medication items were included in the final list. The example medications or medication classes were oral corticosteroids (e.g., prednisolone, dexamethasone, and hydrocortisone), NSAIDs and COX-2 inhibitors (e.g., diclofenac, ibuprofen, piroxicam, celecoxib, and etoricoxib), pioglitazone, salbutamol, ergotamine plus caffeine, and cancer drugs (e.g., cyclophosphamide, docetaxel, and fluorouracil) In phase 2, a 1:1 matched case-control study was conducted using data on HF patients obtained from the electronic medical database of the two study hospitals (a secondary-care hospital and a tertiary-care hospital). Case group included the patients who had a history of HF-related in-hospital death and first hospitalization during 2017-2019; control group were recruited from the remaining patients who had no study outcome but had outpatient department visit or non-HF hospitalization for the first time during the same period. One case was matched with 1 control using hospital site, gender, and index year (2017, 2018 or 2019). Last prescription of any PIMHF item within one year until the index date was identified using the hospital medication codes related to 47 PIMHF items in the Thailand list of PIMHF. The adjusted odds ratio of HF-related in-hospital death or hospitalization associated with PIMHF prescription was calculated using a conditional logistic regression (CLR) model. After matching, each case and control group consisted of 1,603 patients. Twenty-one of 47 PIMHF were found to be prescribed. Two hundred and sixty-seven (8.33%) patients, including 153 (9.54%) cases and 114 (7.11%) controls, were found to be prescribed with any PIMHF. Compared with the referent group of patients who were not prescribed with any PIMHF, those prescribed with any PIMHF had an adjusted odds ratio of 1.39 (95% confidence interval [CI]: 0.96 to 2.00, P-value = 0.077). In a subgroup analysis, the medication classes found to be positively associated with HF-related in-hospital death or hospitalization included NSAIDs and COX-2 inhibitors (adjusted OR = 2.51, 95%CI = 1.25 – 5.06, P-value = 0.010) and beta2 agonists (adjusted OR = 4.89, 95%CI = 1.18 – 20.39, P-value = 0.029). Additionally, the medications that can elevate blood pressure, including NSAIDs and COX-2 inhibitors, pseudoephedrine, and ergotamine, were positively associated with HF-related in-hospital death or hospitalization (adjusted OR = 2.25, 95%CI = 1.15 – 4.39, P-value = 0.018). In this study, the most often prescribed PIMHF was oral corticosteroids (2.90%), followed by pioglitazone (2.81%), and NSAIDs/Cox-2 inhibitors (2.65%). In conclusion, this Delphi survey proposed the first list of PIMHF used as a screening tool for medication use in Thai HF patients. The risk of HF-related in-hospital death or hospitalization increases as the following PIMHF items were prescribed: NSAIDs and COX-2 inhibitors, Beta2 agonists, and the medications that increase blood pressure. Thus, a prescription and a review of medication use should be done carefully and entirely among HF patients. The often prescribed PIMHF should be substituted by safer alternative medication to reduce the probability of adverse clinical outcomes.