Patients’ and providers’ perspectives on medication relatedness and potential preventability of hospital readmissions within 30 days of discharge

Abstract Background Hospital readmissions are increasingly used as an indicator of quality in health care. One potential risk factor of readmissions is polypharmacy. No studies have explored the patients’ perspectives on the medication relatedness and potential preventability of their readmissions. Objective To compare the patients’ perspectives on the medication relatedness and potential preventability of their readmissions with the providers’ perspectives. Methods Patients unplanned readmitted within 30 days after discharge at one of the participating departments of OLVG Hospital in Amsterdam were interviewed during their readmission. Patients’ perspectives regarding medication relatedness of their readmissions, the potential preventability, possible preventable interventions, and satisfaction with medication information were examined. Health‐care providers also reviewed files of these readmitted patients. Primary outcome was the percentage of medication‐related and potentially preventable readmissions according to the patient vs the provider. Descriptive data analysis was used. Results According to patients, 36 of 172 (21%) readmissions were medication‐related, and of these, 21 (58%) were potentially preventable. According to providers, 26 (15%) readmissions were medication‐related and 6 (23%) of these were potentially preventable. Patients and providers agreed on the medication relatedness in 11 of the 172 readmissions, and in two of these, agreement on the potential preventability existed. According to patients, preventive interventions belonged mostly to the hospital level, followed by the primary care level and patient level. Conclusion Patients and providers differ substantially on their perspectives regarding the medication relatedness and preventability of readmissions. Patients were more likely to view medication‐related readmissions as preventable.


| INTRODUC TI ON
Unplanned hospital readmissions within 30 days are increasingly used as an indicator of quality and safety in health care. 1 shows that a median of 21% of readmissions are due to medication and 5%-87% (median 69%) of these readmissions were deemed preventable. 9 The risk for medication-related problems increases with polypharmacy. A review indicates that 18%-38% of patients report medication-related problems after hospital discharge. 10 As the patient is the only constant factor in the care continuum, information from the patient is needed to get insight into medication-related problems occurring between discharge from hospital and readmission. Kari et al 11 show that patient involvement is essential in detecting medication-related problems, because otherwise poor therapy control, non-optimal medication use, or intentional or unintentional non-adherence might be missed.
However, studies investigating patients' perspectives on medication relatedness and preventability of these readmissions are lacking. Consensus between patients and providers with respect to the role of medication as a potential cause of readmissions is necessary to achieve optimal pharmacotherapy. 12 If a readmission is caused by medication according to the patient without being aware that his provider is not convinced of a causal association, a patient could stop independently with the suspicious medication resulting in non-adherence. On the other hand, if a provider believes the readmission is caused by medication but the patient is unaware of the provider's perspective, medication could still be taken by the patient resulting in a repeated readmission.
First, the aim of this study is to describe patients' perspectives on the medication relatedness and potential preventability of their readmissions and compare these with providers' perspectives.
Secondly, we describe the patients' perspectives regarding interventions that could have prevented medication-related readmissions and the patients' satisfaction with information about medication during the index admission.

| Design and setting
The data for this cross-sectional observational study were collected within the context of a larger study on all-cause readmissions. This current study however focuses on medication-related readmissions.
The study was performed at OLVG, a general teaching hospital in Amsterdam, the Netherlands, from July 2016 until May 2017. A list with readmissions within 30 days of discharge was generated within the hospital information system and daily screened by the research coordinator for eligibility.
Patients ≥18 years readmitted within 30 days after an index admission (first admission) to one of the departments of cardiology, gastro-enterology, internal medicine, neurology, psychiatry, pulmonology and surgery were interviewed during their readmission.
Patients were excluded if they were transferred to another hospital or self-discharged, or when it was not the first readmission of the patient and if the readmission was due to attempted suicide or when the patient did not use any medication at all. Furthermore, a readmission was excluded if it was scored by providers (see below) as unrelated to the index admission. This was done to exclude 30-day readmissions that occurred coincidentally. For example, a patient admitted with pneumonia discharged in a good clinical condition and readmitted within 30 days due to a traffic accident. Finally, providers had access to the interviews and registered whether they had used the interview in their review to assess the preventability of a readmission. If a patient interview was used by providers, this interview was excluded as well. The study was approved by the local review board of the hospital (ACWO-MEC, registration number: 16-028).
Patient data were obtained and handled in accordance with privacy regulations.

| Pharmaceutical care during the index admission
In the OLVG Hospital, two different processes are carried out to improve continuity of pharmaceutical care. 13 • On the departments of cardiology, pulmonology, internal medicine, gastroenterology and neurology, our hospital has implemented a Transitional Pharmaceutical Care (TPC) programme. 14 In short, hospital pharmacy teams perform medication reconciliation at hospital admission and discharge using the dispensing history of the community pharmacy and information from the patient/carer himself. Any discrepancies between a patient's actual medication use and the medication prescribed in hospital are discussed with K E Y W O R D S hospital readmissions, medication, patients' perspectives, preventability, providers' perspective the resident. No formal medication review is performed. However, obvious errors in the pharmacotherapy are eliminated, for example lack of a laxative when an opioid is prescribed or no indication for hypnotics at discharge, addressing a stop date for antibiotics or opioids. The reason for medication changes is explained to the patient during discharge counselling, and a written medication summary is provided. The pharmacy team makes a TPC-medication overview that the resident could upload into the discharge letter.
• On the departments of psychiatry and surgery, residents and nurses are responsible for assessing a patient's actual medication use by interviewing patients/carers. If regarded necessary, they can request the hospital pharmacy to obtain a dispensing history from the community pharmacy. At hospital discharge, the resident uploads information from the hospital's prescribing system or types information into the discharge letter to the general practitioner.

| Patients' perspectives
Patients were interviewed during their readmission, or three attempts were made by phone in case the patient was already discharged or when a caregiver (family member or partner) needed to be approached, or in case of a language barrier or when the patient was unable to answer the questions. A structured interview guide was developed based on previous studies on readmissions and expert opinion. [15][16][17][18][19][20] For the purpose of this study, the following main topics were included: patients' perspectives on medication relatedness, patients' perspectives on potential preventability and preventive interventions, and patients' perspectives on medication-related information received during index admission (File S1). Additionally, the following socio-demographic factors were asked: nationality, living situation, educational level and self-experienced health status.
Format of the questions included multiple-choice, yes/no and free text. Interviews were conducted by medical students who received the interview guide and were trained for this. Interviews lasted approximately 30 minutes. During the entire interview period, students were supervised by the coordinating physician-researcher.
Interviewers manually recorded responses on data extraction sheets in Access 2010 (Microsoft).

| Providers' perspectives
Health-care providers who reviewed the readmissions were residents of the participating departments and a pharmacist. First, providers reviewed complete medical records of the readmitted patients to assess whether the readmissions were clinically related to the index admissions. If it was clinically related, the medication relatedness, using the algorithm of Kramer et al, 21 and the preventability, using a modified algorithm of Schumock et al, were assessed. 22 Readmissions that were assessed as potentially preventable by the providers or raised questions after the research coordinator's check were included to be discussed once a month, during a multidisciplinary meeting with the research coordinator, residents and a pharmacist. All readmissions assessed as medication-related by the residents and pharmacist have been reassessed by a senior physician (CS) and a clinical pharmacologist (MJ) to validate the findings.

| Outcomes
Primary outcome was the percentage of medication-related and potentially preventable readmissions according to the patient vs the provider. Secondary outcomes were patients' perspectives regarding interventions that could have potentially prevented the readmission and percentage of patients who were satisfied with information about medication during their index admission.

| RE SULTS
Of 646 readmissions that were screened, 427 (66%) readmissions met the inclusion criteria, and 227 interviews were conducted, of which 172 (76%) were included in the final data analysis (Figure 1).
Main reasons that the interview was not conducted were as follows: failed attempts to get into contact (n = 50), unwillingness to participate (n = 39) and cognitive/physical problems (n = 34). One hundred fifty interviews (87%) were conducted with patients, 4 (2%) with patients and caregivers and 18 (10%) with caregivers. The mean age of the included patients was 62 years (SD 18), 47% were male, and the mean number of the medications at discharge of index admission was 9.2 (SD 5.9) ( Table 1). Table 2 shows patients' and providers' perspectives on medication relatedness and potential preventability in 172 readmissions. According to patients' perspectives, 36 (21%) readmissions were medicationrelated, of which 21 (58%) were potentially preventable (File S2). The causes (n = 23) of the potentially preventable readmissions according to patients were as follows: issues with dosage (n = 8, 35%), for example antibiotic discontinued too soon or too high dosage prescribed; change in medication (n = 6, 26%), for example medication changes that were unclear to the patient; a medication interaction (n = 1, 4%); costs (n = 1, 4%); or adherence (n = 1, 4%). In six readmissions (26%), the patient described an adverse drug reaction as a cause, but in most of those cases, the patient could not pinpoint which medication exactly was responsible for the side-effects. According to providers' perspectives, 26 (15%) readmissions were medication-related, of which 6 (23%) were potentially preventable. In 11 of the 172 readmissions, patients and providers agreed on the medication relatedness, and in two of these, agreement on the potential preventability existed (File S2).

| Patients' perspectives on preventive interventions
Of the readmissions that were medication-related and potentially preventable according to the patient (n = 21), patients reported 23 preventative interventions. Hospital-based interventions were 18 times reported, including performing more diagnostics (33%), improving medication-related information (17%), providing a longer hospital stay (17%), treating symptoms/complaints (17%), providing better aftercare (11%) or reacting faster (6%) ( Table 3). In two cases, patients reported that general practitioner-based interventions could have prevented the readmission, by reacting faster. Two patients reported that he or she could have prevented the readmission by being adherent to therapy. Table 4 shows patients' satisfaction on medication-related information. In readmissions that were medication-related but not preventable according to patients' perspectives (n = 15), patients reported in 93% (n = 14) that they had received as much information as they needed about medicines compared with 67% (n = 14) in readmissions deemed potentially preventable (n = 21). Also, information about side-effects of medicines was more often scored as 'as much information as I needed' in readmissions not preventable according to patients' perspectives compared with potentially preventable readmissions, 87% (n = 13) vs 43% (n = 9), respectively. In 73% (n = 11) of the readmissions scored as not preventable, patients received written instructions, compared with 57% (n = 12) in readmissions scored as potentially preventable.

| D ISCUSS I ON
This study shows that according to patients, readmissions are more often medication-related (21% of readmissions in patients vs 15% in    Improving medication-related information 3 (17) Example patient's answer 'I was confused about my diuretics, one was started and one was stopped. I would get some diuretics upon discharge, however at discharge there was a lot of confusion and I did not get them. Not taking the diuretics could contribute to my rehospitalisation' Longer hospital stay 3 (17) Example caregiver's answer 'My father was discharged too early. The neurologist could not find anything and he thought it was something with the heart. However, the cardiologist refused to examine my father, so there was no follow-up. We thought something was wrong with the medication, but they did not listen to us. Now he is readmitted due to a way too low blood pressure' suggests that more patient engagement is needed not only during hospitalization, but also in the discharge process and the period after hospitalization, especially for pharmaceutical care. This could be achieved by several methods, such as the use of lay language, asking patients what they want to know regarding their medicines, providing written information, repeating information or using the 'teach-back' method, which is a strategy in which patients are asked to restate information that has been presented to them. 29 As previous studies have shown that patients' needs can increase after discharge, also a follow-up phone call after discharge could be helpful to identify and to prevent medication-related problems. [30][31][32] Further research should find out how these interventions could help to reduce medication-related readmissions.
The strength of this study is the description and comparison of the medication relatedness and potential preventability of readmissions according to perspectives of both patients and providers from several hospital departments. However, some limitations need to be discussed. This study is conducted in one hospital, which limits the generalizability. Another limitation is that patients were interviewed about the index admissions during readmission, which could lead to subjectivity and hindsight bias. However, in this way we could obtain information of the period after discharge of the index admission. Some patients could not be interviewed due to severe illness or unwillingness to participate. This could lead to selection bias as healthier or more satisfied patients were more often interviewed, which may have resulted in lower reporting of medication relatedness and preventability.

| CON CLUS ION
Patients and providers differ substantially on their perspectives regarding medication relatedness and potential preventability of hospital readmissions. According to patients, medication-related readmissions occur more often and are more often potentially preventable compared with providers' perspectives. Patients reported most often that actions on the hospital level were possible to potentially prevent the readmission. Further studies need to explore the reasons for the gap between patients' and providers' perspectives.

ACK N OWLED G EM ENTS
We are grateful to the doctors of all departments for their help during the study and the students Najla el Morabet, Jim Hoffmann and Emilie Haitsma for conducting the interviews. Special thanks to Eva Kneepkens for the data collection and support.

DATA AVA I L A B I L I T Y S TAT E M E N T
Data are available on request from the authors.