Abstract
Objective:
Public primary healthcare clinics in Singapore manage a large proportion of elderly patients with chronic diseases. Inappropriate prescribing of medications and polypharmacy in the elderly are associated with adverse outcomes. It is hence important to stop potentially inappropriate medications in this vulnerable group. An approach coined ‘deprescribing’ has been used to describe a patient-centred process of optimising medication regimens. The study aimed to elucidate patients’ attitudes towards the number of medications they were taking and identify factors that might influence acceptance of deprescription.
Method:
A cross-sectional study using the validated Patients’ Attitudes Towards Deprescribing (PATD) questionnaire was performed at two public primary healthcare institutions in Singapore. Participants were on regular follow-up at the clinics for chronic disease management and had at least five regular prescription medications.
Results:
The study found that participants (with a mean age of 68) had an average of four medical conditions and six prescription medications, with the majority (60.3%) expressing that they were taking a large number of medications. Of note, 93.4% of participants were willing to stop one of their medications if advised by the doctor. This was associated with a younger age (<65 years old), not having a discount card for medications and having a higher physician trust score (Wake Forest Physician Trust Score).
Conclusions:
This study showed that majority of the participants were willing to cease a medication that their physician thought was no longer required. Factors were also identified which potentially may be targeted to facilitate deprescription.
Introduction
Polypharmacy (generally defined as five or more medications) may be appropriate for some patients, but there are genuine concerns about its association with inappropriately prescribed medications. 1 The number of medications a person takes is the single factor with the strongest association with inappropriate prescribing and risk of adverse drug events; one report estimated the risk as 38% for four drugs and 82% for seven drugs or more. 1 These have been associated with increased risk of adverse events and death.2,3
A process to systematically cease inappropriate medications and address polypharmacy has been coined ‘deprescribing’. 4 Deprescribing is essentially the optimisation of medication regimens through the cessation or tapering of potentially inappropriate medications. Various algorithms for deprescribing have been proposed,4-6 but none was validated systematically until the recent development of a patient-centred, evidence-based five-step deprescribing process. 7 The patient and healthcare professional (physician or pharmacist) are central in this process. A recent review on the process of deprescribing suggested that deprescription may be facilitated through providing and applying available resources (e.g. practical guidelines on deprescribing in older people 8 ) during each clinical encounter and, at the systems of care level, the importance of having key stakeholders implement strategies that recognise deprescription as an essential part of good clinical care. 9 However, we note that amenability to deprescription is likely to be dependent on socio-cultural conditions, which are expected to differ between countries, and even between care settings in the same country.
Studies have showed that people generally dislike taking medications 10 but may have some reluctance in stopping one of their regular medications.11,12 A majority of patients have reported their willingness to be involved in making decisions about their medications, and therefore it is important for the physician to involve the patient in this process. 13
Recognising that patients’ attitudes and beliefs towards their medications are an important factor in the success of deprescribing, Reeve et al. developed and validated the Patients’ Attitudes Towards Deprescribing (PATD) questionnaire. 14 It explores peoples’ attitudes, beliefs and experiences regarding the number of the medications that they are taking and how they would feel about cessation of one or more of their medications. 14 The PATD questionnaire was developed and first employed in a public teaching hospital in Adelaide, South Australia.14,15 The study showed that 92% of patients were willing to try medication cessation if their doctor thought it was appropriate, and age, number of regular medications and number of medical conditions did not influence this attitude. Interestingly, a significant association was found between the possession of a medication discount card and physician trust scores with patients’ willingness to cease medications. Patients who had a discount card were less likely to be amenable to deprescribing, whereas patient who had higher trust scores in their physicians were more likely to be in favour of deprescribing.
Singapore has a rapidly ageing population, and the number of citizens aged 65 years and above is expected to more than double from the current 440,000 to 900,000 by 2030. 16 Thus the number of elderly patients with multiple comorbidities – and hence polypharmacy – will escalate in the next 15 years. An earlier local study in 2004 at nursing homes has already alluded to this problem; polypharmacy was detected in 58.6% of the 454 elderly residents, with 70.0% of them receiving inappropriate medications. 17 Some 45% of local multi-ethnic Asian patients with chronic diseases are managed in local public primary care clinics (polyclinics) due to subsidised consultation fees and ease of filling prescriptions from their in-house pharmacies. 18 The magnitudes of polypharmacy and inappropriate prescriptions in ambulatory settings are expected to be significant. To mitigate the potential adverse effects of polypharmacy secondary to inappropriate prescribing, it is timely to assess if patients are receptive to deprescribing. We postulated that patients in an ambulatory setting would agree to deprescribing of selected non-essential medication upon the influence of physician, personal and economic factors. Thus, the study aimed to determine the proportion of primary care patients with chronic diseases and prescribed with multiple medications that would agree to deprescribing. The secondary aim was to identify the factors that would influence their attitude towards deprescribing.
Methods
Study sites
This study was conducted in Singapore, where subsidised public primary healthcare services are delivered through a network of 18 outpatient polyclinics. Polyclinics are ‘one-stop’ healthcare centres providing primary medical treatment, preventive healthcare and health education.
Participants were recruited from two typical polyclinics in Outram and Queenstown, located in south-central Singapore. Each polyclinic manages an average of 500 patients daily during office hours. The demographic profiles of the patients at these two sites were similar; 37% of them were of age 65 years and older.
Subjects
The target subjects were patients of both genders, of local major ethnic groups and aged between 45 and 84 years who were receiving treatment for more than one chronic disease at the respective polyclinics. They were prescribed with five or more daily medications for their multi-morbidity management. Chronic disease was defined as any disease under the Ministry of Health’s Chronic Disease Management Program (CDMP). 19 Patients with documented history of cognitive impairment or dementia, or who were incapable of understanding and communicating in English were excluded. Older adults were defined as individuals age 65 years and above at the time of data collection.
Sample size calculation
Based on previous literature by Reeves et al., 92% of the participants stated they would be willing to stop one or more of their current medications if possible. To ensure that the 95% confidence interval estimate of the proportion of participants willing to stop medication is within 5% of the true proportion, a sample size of 114 is needed. The investigators decided to increase the sample to 140 participants to allow for incomplete questionnaires with missing data or subject withdrawal.
Instrument
The questionnaire collected demographic data, which included the subjects’ age, gender, ethnicity, highest education level attained, marital status, housing type, living arrangements and eligibility for social and financial support and medical privileges ranging from additional medication subsidies to total waiver of all fees. The medical subsidy schemes include the possession of Civil Service Card, 20 Pioneer Generation Card for senior patients, 21 Community Health Assist Scheme 22 and Medifund. 23
The PATD questionnaire was incorporated into the survey instrument with permission from the original developer, Dr Emily Reeve. 14 The PATD questionnaire comprises 15 questions exploring the attitudes and beliefs of patients towards their medications and willingness to stop them. The first 10 questions measure the responses on a 5-point Likert scale, while the last five are multiple-choice questions.
The Wake Forest Physician Trust Scale (WFPTS) was selected as a measure of patients’ trust in their physicians to enable comparison with related studies. 24 A higher WFPTS score correlates with higher physician trust.
Information pertaining to whether the patient had been seen by the same doctor twice or more frequently in a year at the polyclinic was collected as a surrogate indicator of continuity of care by a regular doctor. This data would be used to understand its association with the WFPTS scores. The instrument also collected back-end data on patients’ medical conditions and their medications. All prescribed long-term oral, inhaled or injectable medications, which were consumed daily by patients, were included. Topical medications were excluded.
Subject recruitment and questionnaire administration
The investigators conducted a briefing to induct the polyclinic staff on the objective and procedures of the study. The polyclinic doctor, nurse or pharmacist screened potential subjects for their eligibility to enrol into the study on a case-encounter basis during their medical consultation at the study sites. Subjects who fulfilled the inclusion and exclusion criteria were directed to a private room where research assistants (nursing students on attachment) provided details of the study protocol and clarified any doubts and queries. They administered the instrument to the enrolled subjects after obtaining their written informed consent. The study completed its subject recruitment over 2 months from June to July 2015.
Data management
The investigators accessed subjects’ electronic medical records to obtain information on their chronic medical conditions and medications. Next, they transcribed the data into a password-protected spreadsheet. The variables were pre-defined in a data dictionary designed by the data management officer in the investigator team. The latter audited and anonymised the data before handing it over to the biostatistician for analysis.
Statistical analysis
Data were coded and analysed using IBM SPSS Statistics version 22.0 for windows. Descriptive data were analysed and reported as median (interquartile range [IQR]) or percentage. Mann–Whitney U statistical test was used to compare non-normal ordinal items among the demographic groups, while analysis on correlation for the outcome on willingness to stop medication and the other 5-point ordinal items were tested using the Spearman’s rank correlation coefficient. Ordinal regression was also performed to adjust for the potential significant factors. p⩽0.05 was considered statistically significant.
The SingHealth Centralized Institutional Review Board (CIRB) granted ethics approval to the study (CIRB Ref: 2015/2106).
Results
In total, 157 patients were approached to participate in the study. Of these, 20 refused or did not meet the criteria, resulting in 137 participants who completed the questionnaire. Out of these, one withdrew from the study. A total of 136 participants were included in the final analysis.
The study population comprised 41.2% females, of Chinese (69.9%), Malay (8.8%) and Indian (15.4%) ethnicity. The median age of the participants was 68 years. They had a median of four medical conditions and six prescription medications. The three most common medical conditions were hypertension, hyperlipidaemia and type 2 diabetes mellitus. Some 75.7% of participants were on follow-up with the same doctor, and 72.8% of patients had additional medical subsidies for their visits. The demographic characteristics of the study population are shown in Table 1.
Demographic characteristics of the study population.
The higher the Wake Forest Physician Trust Score, the higher the trust between patient and physician. 24
No significant association was found between increasing age and the number of medical conditions (p=0.07) or medications (p=0.41). There was also no significant difference in the number of medications and number of chronic diseases between those participants who were elderly (⩾65 years) compared with the younger participants. The participants who were ⩾65 years had a median of four medical conditions and took a median of six chronic medications. No significant association was noted between seeing the same doctor two or more times and the physician trust score.
The majority (60.3%) of the participants expressed that they were taking a large number of medications (Table 2). If advised by the doctor, 93.4% of participants were willing to stop one of their medications. Of the elderly participants, 91.6% were willing. Although a significant proportion of participants were comfortable with the number of medications that they were taking (82.4%) and believed that all their medications were necessary (88.8%), 72.8% still had the desire to reduce the number of medications they were taking. A quarter of participants (25.0%) felt that they might be taking one or more medications that they no longer needed, and 30% felt that one or more of their medications were giving them side effects.
Patients’ attitudes towards deprescribing based on the PATD questionnaire#.
# Adopted from Patients’ Attitudes Towards Deprescribing Questionnaire. 14
Patients’ willingness to stop a medication was positively associated with the six factors shown in Table 3 but had a strong association only with the patient’s desire to take fewer medications (Question 5). They remained willing to stop a medication even when they were comfortable with the number of medications and had the belief that all the medications were necessary.
Spearman Correlation of responses within the PATD questionnaire.
Significant results.
The patients’ desire to reduce medications was positively associated with the three factors in the Table 3, especially when the doctor indicated possibility of stopping the medication (Question 4).
The patients generally felt comfortable taking a large number of medications despite some of them perceiving side effects from one or more medications. Those who felt side effects seemed more likely to perceive that not all the medications were necessary. It appeared that the willingness and desire to stop a medication were not associated with the perception of side effects.
Individual characteristics that were in favour of deprescribing included a younger age (<65 years old), not having a medication discount card and having a higher physician trust score (WFPTS) (Table 4). After adjusting for the factors in Table 4, only age and a higher physician trust score remained as significant factors. The association between possession of any medical subsidies and willingness to stop a medication became non-significant.
Ordinal regression of factors influencing willingness to stop medication.
Significant results.
Discussion
The results showed that the majority of the participants were willing to stop a regular medication if their physician thought it was no longer required. A high physician trust score and a younger age group were significant factors influencing this attitude. Nonetheless, the vast majority of elderly participants (91.6%) were also willing to cease a medication. There is no reported cut-off value in the literature for what is considered a high score for the WFPTS, but the median score obtained in our study was similar to Reeve et al.’s paper. 15 It would be interesting to compare the median WFPTS score if it is employed in other future local studies.
It has long been recognised that trust plays a central role in patient–physician relationships. Similar to the findings of this study, Reeve et al. also found that high physician trust scores were associated with greater willingness to stop a medication. 15 Beyond deprescribing, there is also evidence to support the relationship between trust and a patient following the recommendations of their physician. A study looking at colorectal cancer screening in low-income patients showed that greater trust in the primary care provider was associated with colorectal cancer screening completion. 25 Higher trust scores were independently correlated with better adherence to treatment and satisfaction with the physician. 26
This study alludes to the importance of establishing trust in the physician as a pre-requisite measure in order to effectively influence change in attitudes and practices of patients towards optimal health outcomes. Current literature suggests that physician trust can be enhanced via continuity of care by the same physician, physician personality and behaviour and patients’ perceived freedom to select their physician of choice. 27 Hence, physicians and administrators need to critically examine current care delivery models and systematically introduce measures to catalyse the development of trust between physician and patient. Ensuring adequate time to allow better physician–patient communication during consultation and arranging subsequent reviews by the same physician are care processes which can be implemented to build trust in physician.
Deprescribing has been successfully tested as a systematic way to tackle the problem of polypharmacy among patients. Garfinkel et al. conducted a non-randomised deprescribing study in ambulatory elderly adults in the community. 28 It achieved significant reductions in medication use and improved self-reported general health. Two randomised studies focusing on deprescribing in older hospitalised adults reported reduced use of potentially inappropriate medicines using the Screening Tool of Older People’s Prescriptions (STOPP).29,30 One of them also reported non-significant reductions in falls, all-cause mortality, and general practitioner visits during the 6-month follow-up period. 29
Currently, there is no single recommended, most effective strategy of performing deprescribing. The deprescribing process requires time, expertise and involvement of both the physician and patient. Physicians themselves must feel comfortable, competent and confident of deprescribing before they roll out to their patients. It is thus important that the knowledge, attitudes and perceptions of local primary care physicians towards deprescribing be assessed with further research before designing any interventions.
Deprescribing faces inherent challenges that need to be overcome. Among patients with multi-morbidity, the paucity of deprescribing guidelines hampers physicians in making decision on the selection of drug classes suitable for deprescribing without compromising on the health and safety of their patients. 31 The use of an information technology-based decision support system within the electronic prescription framework may be a solution, but it requires further evaluation via research and development. In addition, case discussions on deprescribing involving multidisciplinary team, including geriatricians, have been shown to be effective interventions. 32 There is a need for more local primary care physicians to take on post-graduate training in geriatrics. 33 They will expand the pool of physicians trained to be competent in deprescribing in primary care.
Half of the patients in this study were comfortable with a pharmacist being involved in stopping a medication and providing the follow-up. Pharmacists could thus play a more proactive role in deprescribing within the multidisciplinary team, such as screening and identifying suitable patients, and support the attending physician in effecting the process.
Strength and limitations
This study is the first to be conducted on multi-ethnic Asian population in primary care, providing an unprecedented insight into the views of the patients with multi-morbidities on deprescribing. The favourable response rate of 88% allowed the completion of patient recruitment within 2 months at the study sites.
The study was undertaken in public polyclinics in Singapore, limiting its generalisability of findings to patients who are managed by other primary healthcare providers such as private general practitioners. Nonetheless, polyclinics provide primary healthcare services to 45% of the local patients with chronic diseases. 18
A convenience sampling method was used to recruit the study participants and we do not have information on the total number of patients who would have otherwise fulfilled the inclusion criteria during the study period. Hence, this may limit the representation of the overall patient population in these two polyclinics.
Furthermore, the PATD questionnaire was available only in English, which restricted its administration solely to patients who were competent in this language. Locally validated questionnaires, which are translated to other major languages, will need to be developed to cater to non-English patients to assess the local population’s willingness for deprescribing. Comprehensive coverage of the population is essential to guide policy decisions and implementation in clinical practices.
Conclusion
This paper is, to our knowledge, the first study assessing the attitudes of patients in Singapore towards deprescribing. This study showed that 93.4% of the participants were willing to cease a medication that their physician thought was no longer required. This would be a worthwhile approach to reduce inappropriate polypharmacy to optimise care and reduce harm to patients. A patient’s age and trust in the physician influenced their willingness towards stopping a medication. Further research is required to identify physician factors to facilitate deprescribing.
Footnotes
Acknowledgements
The authors would like to acknowledge the kind approval of Dr Emily Reeve of University of Sydney for the use of the PATD questionnaire. The investigators would like to thank SingHealth Polyclinics for the seed funding; the research assistants and lecturers from Ngee Ann Polytechnic for assisting in the data collection and data entry; Caris Tan for her role as protocol administrator and her assistance in seeking approval from SingHealth Centralized Institution Review Board; Usha Sankari for data management; Nurse Manager Lee Hwee Khim, the polyclinic doctors, nurses and pharmacists at the study sites for their support and help rendered during the period of subject recruitment.
Ng WL, Tan ZWM, Koh YLE, TNC designed the study protocol. Ng WL, Tan ZWM supervised the research assistants during subject recruitment and study execution. Koh YLE analysed the de-identified data in consultation with Ng WL and Tan ZWM. Ng WL, Tan ZWM, Koh YLE, TNC interpreted the results. Ng WL, Tan ZWM drafted the manuscript. All authors reviewed and finalised the paper before journal submission
Declaration of Conflicting Interests
The authors declare that there are no conflicts of interest.
Funding
This work was supported by Singhealth Polyclinic.