Background
Nursing home patients are old and fragile, and over 80% have dementia [
1]. Multimorbidity is common in old age, and cardiovascular diseases, stroke, cancer, and psychiatric disorders often co-occur [
2,
3]. A consequence of multimorbidity is polypharmacy, affecting most nursing home patients, who on average use eight different drugs every day, and two on demand [
4]. The altered pharmacokinetics and pharmacodynamics of the aged body increase susceptibility to adverse events from drugs [
5]. A study by Soraas et al. (2014) found that the use of eight or more drugs significantly increased the risk of drug-drug interactions [
6], which in turn leads to falls, cognitive decline, medication-related problems, and even increased mortality [
7‐
10].
In nursing homes, physicians frequently prescribe drugs without a proper clinical evaluation of the patient [
11]. Dementia reduces the patient’s ability to report effects and side effects of the treatment and depend on a proxy-rater who has known the patient over time to conduct a medical evaluation. It has been suggested that a systematic review of medications might be a procedure that will help ensure safe and appropriate medical treatment. Randomized controlled trials in nursing homes have been used to test medication reviews including: expert advice, the use of explicit prescribing lists, and multidisciplinary teams with the general practitioner (GP) involved [
12‐
14]. They all demonstrated a reduction in the number of drugs without a detrimental effect on the patient’s health. Other approaches have been tested to improve drug prescribing by involving: a pharmacist, electronic prescribing aides, and the use of explicit prescribing criteria [
15‐
17]. The interventions often comprise of multiple components, require new knowledge and involve different professions; it has been demonstrated that these elements impede implementation [
18].
To ensure that complex interventions are successful and sustainable over time, different approaches have to be combined, and the implementation process must be planned and described in detail. Yet, multicomponent intervention studies have fallen short in reporting their implementation strategies and evaluation there-of [
16]. Consequently, lack of efficacy may be caused by poor implementation, the ineffectiveness of the method or a combination of the two. We believe that we can improve prescribing and the medication review procedure in nursing homes by incorporating clinical assessment using tools validated for people with dementia, and by testing to ascertain whether the intervention was carried out successfully.
In this paper, we describe a novel implementation strategy for an interprofessional medication review. It is based on systematic clinical evaluation of the patient and collegial mentoring of the nursing home team. We describe the process of implementation and report the findings for the following research questions:
1.
How did nursing home staff receive the intervention?
2.
To what degree was the medication review implemented successfully?
3.
What are the barriers and facilitators for implementing interprofessional medication reviews in nursing homes?
Results
Twenty of the 36 intervention units (56%) reported having had a focus on medication reviews in the three years preceding the intervention. Six units had participated in the Norwegian Patient Safety Campaign on improving prescribing in nursing homes [
23], five had been a part of a research project focusing on medication review, and eight had a local focus on medication reviews. No pharmacists attended the medication reviews or the education, mostly because the units did not have any pharmacists on staff (Table
3).
Table 3
Experience, education and workload in health professionals in the intervention group
Staffing, number of patients per nursing staffa (range) | Daytime | 3.2 (1.6–4.0) |
Evening | 4.7 (2.3–6.0) |
Nighttime | 13.0 (4.0–30.3) |
COSMOS ambassadors with direct patient contact, N (N per cluster)
| 73 (2.0) |
Registered nurses, N (%) | | 44 (61%) |
Licensed practical nurses, N (%) | | 9 (12%) |
Unknown education | | 19 (27%) |
Physicians
| | 21 |
Age in years, mean (SD) | | 48 (12.9) |
Female, N (%) | | 8 (38%) |
Mean number of patients in the study, per physician, mean (range) | 22 (8–28) |
Minutes per patient per week, mean (range) | | 20 (8–42) |
Twenty-one physicians had the medical responsibility for the nursing home units (Table
3). Seven physicians (33%) were full-time nursing home physicians and 14 (67%) were GPs with visiting hours at the nursing home. Thirteen had a specialization, 12 of which were in family medicine, and one in internal medicine. The nursing staff in the units consisted of registered nurses, licensed practical nurses, and staff without training. In 30 (45%) units, most staff were hired in 75–100% of full-time equivalent positions; in 24 units (36%), most staff held 50–75% of full-time equivalent positions.
The patients had a mean age of 87 (SD = 7.7) years, 73% were female, and they had a mean of 4 (SD = 3.3) registered diagnoses each. During the four months of the study, 33 patients died and 14 moved to another institution. All participants used a mean of 7.6 (SD = 3.8) drugs each day, ranging from 0 to 19, and had on average 3.4 (SD = 2.3) on demand prescriptions, ranging from 0 to 17. The most frequent regular drug groups were: laxatives (N = 172, 58%), antithrombotic agents, (N = 155, 52%), acetaminophen (N = 136, 46%), antidepressants (N = 118, 40%), and high ceiling diuretics (N = 95, 32%). The most frequent drugs on demand were acetaminophen (N = 147, 50%), anxiolytics (N = 134, 45%), opioids (N = 106, 36%), hypnotics (N = 82, 28%), and laxatives (N = 79, 27%).
Education of nursing home staff
The education program was attended by 105 nursing staff. All units in the intervention group participated and all units sent the required two ambassadors, the average number was three participants per unit. The attendees were mainly two registered nurses and a unit manager or a licensed practical nurse for each unit. Seven of the 21 physicians attended the education program. The non-attending physicians cited limited time and lack of relevance as reasons for not attending. The same number of people attended the midway evaluations as the education program. These meetings had no management level staff or physicians present, but more regular nursing staff participated.
The training materials were clearly visible in all units during visits from the researchers. Ten units asked for an additional supply of flash cards because of their popularity; seven units wanted extra training loose-leaf binders.
Interprofessional medication review based on collegial mentoring
All units conducted the medication review. Some physicians had responsibility for several units and thus the visits were coordinated so that they could be performed in one appointment. Six units took the opportunity to have additional medication reviews. All patients were assessed prior to the scheduled medication review. The nursing home physician and a nurse from the unit attended each medication review; in addition, an extra nurse or the unit manager attended about half of these medication reviews. Using the COSMOS method, the group spent on average 1.5 h (range 1 to 2 h) performing medication reviews for eight patients. The first few patient cases were discussed extensively, while decisions were easier to make with the subsequent cases, because similar issues had already been discussed in previous cases.
Implementation process
All units used the patient logs, though 57 patients (19%) were missing log entries over four months, attributable to either death or being moved from the unit during the intervention period (77%). As the total number of patients remaining in the study declined, the total number of patients changed throughout registration. For the first four weeks, the log was filled in for 211/294 (73%). The completion rates for week 8, 12, and 16 were: 206/288 (71%), 140/276 (51%), and 198/271 (73%), respectively. Of the patients with at least one entry in the logs, 220 (92%) had received a medication review. Medication indications were recorded for 200 (83%) of the patients. For 184 (77%) of the patients, either the patient himself/herself or relatives were informed about changes in drugs, and for 34 (14%) patients the nurse filling in the log did not know whether this information was given. In 72 (30%) of the cases, a drug was reinstated after a pause, and changes in health were documented in 204 (77%) of the patients (Table
4).
Table 4
Feedback by patient logs
Question in patient logs | N | % | N | % | N | % |
Had at least one medication review | 220 | 92% | 16 | 7% | 3 | 1% |
Indication on each drug | 200 | 83% | 36 | 15% | 4 | 2% |
Informed patient and/or relative about change | 204 | 85% | 9 | 4% | 26 | 11% |
Reinstated drug after pause | 72 | 30% | 141 | 59% | 27 | 11% |
Documented change in patient health | 184 | 77% | 20 | 8% | 34 | 14% |
Healthcare staff reported potential barriers and promotors for the implementation process during the medication review and the midway evaluation (Table
5). The staff welcomed this approach, which they felt, created an arena for learning, engagement and further development. One participant also expressed this enthusiastically:
I want to run back to my nursing home unit and look over all the medication charts right away!Table 5
Barriers and promotors for good implementation
New and difficult clinical instruments | Engagement |
Lack of competence | Arena for learning |
Practical challenges with changing drug regimes | Introducing a colleague to discuss difficult decisions with |
Poor knowledge about electronic patient records | The intervention was perceived as important and relevant |
Lack of time | Improved communication |
Ethical dilemmas | Pleased relatives |
The implementation of the systematic clinical evaluation by means of validated assessment instruments was straightforward for most of the nursing staff. Meanwhile, some units struggled with new or unfamiliar instruments and some of the nurses were less interested or did not regard the evaluation of the patient as their responsibility. Licensed practical nurses also felt less competent to evaluate changes in patient health or to communicate with the relatives in connection with the medical changes. The introduction of the word “pause” was regarded as a relief because it was easier for the relatives and the other staff in the unit to accept a pause rather than discontinuation of a drug. Practical difficulties where mainly related to the use of multi-dose dispensed drugs when several medications were changed at the same time. The doctors also had varying degrees of knowledge about the electronic patient record and some found it difficult to alter prescriptions. Thus, they depended on nurses to be present at the medication review and to follow up and document the changes in the system. Despite the fact that most GPs did not attend the seminar at baseline, the collegial monitoring was seen as positive and they were receptive to the medication review with the researchers. They appreciated the discussions with the nurses and researchers in the interprofessional setting. Not surprisingly, there was not always agreement in their professional opinions but observations made by the nurses in advance or related to scorings by the clinical assessments, influenced the prescribing routines positively and improved the communication between the staff members.
Several physicians suggested that the collegial support from the researchers and the interprofessional discussion provided help in the decision-making processes, as one colleague commented: It is never easy to find the correct timing for deprescribing, for instance, anticoagulants to prevent a stroke. It is an ethical issue, you know. Notably, these judgments were often related to drugs that were initiated during a hospitalization, often a long time ago, even when indications were no longer relevant.
Lack of time was a key barrier mentioned by all participants, including limited time for training, evaluation of the patients, and proper documentation. Meanwhile, motivated managers were able to initiate the medication review, despite the time-barriers. One nurse said: Our boss encourages us and has provided a schedule for all the training and observations we are supposed to do; she also checks whether we have done it. To conclude, participants felt that the intervention was needed and relevant. It granted knowledge and highlighted problems that they all felt were present. In addition, the relatives reported to the nurses that they were pleased with this thorough approach.
Discussion
This study describes a new approach for systematic medication review based on clinical assessment of the patient and collegial mentoring of the physicians in Norwegian nursing homes. The implementation of the intervention was highly appreciated and well received by the physicians and created enthusiasm among nursing staff. The approach improved communication between the health personnel, patients, and the relatives. After four months, 92% of the patients had undergone a medication review, changes in the patients’ health were documented for 77% of the patients. Thirty percent of the patients were put back on a deprescribed drug. Lack of time was the most frequently reported barrier against the intervention as well as difficulties in engaging everyone in the unit. The systematic use of clinical assessment tools before and after the medication review was of key importance to the clinician because this facilitated the optimizing of safe prescribing patterns.
To our knowledge, this is the first study that describes the implementation strategy of collegial monitoring combined with thorough clinical testing of nursing home patients in connection with a medication review. Nursing home patients often suffer from neuropsychiatric symptoms, and consequently, psychotropic drugs are frequently prescribed [
36]. To avoid unnecessary drug use, and thus side effects and interactions, it is a prerequisite to assess relevant clinical symptoms before and after treatment has been initiated. In an Australian randomized controlled trial, Potter and colleagues included 96 participants, who were systematically assessed by two researchers [
12]. During the medication review, they were able to withdraw risk-modifying drugs, and to some extent symptom-modifying drugs. However, they did not report how neuropsychiatric symptoms were assessed and re-assessed after optimizing the medication lists. In the WHELD study, Ballard et al. focused on antipsychotic drug use in connection with neuropsychiatric symptoms on 187 nursing home patients [
37]. After 9 months, the intervention group reduced antipsychotic drug use by 50% compared to the control group. The antipsychotic review group experienced a worse outcome of overall neuropsychiatric symptoms after the procedure. Meanwhile, social interaction and exercise proved to be essential to alleviate these symptoms. Both studies describe complex interventions but the implementation strategies were not carried out as planned in these settings. Furthermore, none of these studies included the nursing home physician in the medication reviews. In a different publication [
38], we describe how the COSMOS study showed significant reduction in use of antihypertensive drugs in the intervention group compared to the control group, without any lasting effect on pulse or blood pressure.
It is of key importance to follow up and document changes after the medication review. We found that nearly 80% of the patients had changes in their health documented after the medication review. We also found that 30% had a drug reinstated after a pause. The study by Potter et al. had 41% unsuccessful withdrawal rate [
12]. A Swedish study focusing on improving health monitoring of nursing home patients reduced drug use and increased documentation and follow-up [
39].
The development, implementation and evaluation of complex interventions as described by the Medical Research Council guidance, UK, is challenging [
40]. The process depends on a range of possible outcomes, the variability in the target population, and the number and content of the elements in the intervention package. Thus far, few trials have focused on the critical issue of whether the implementation of a systematic medication review is feasible in nursing homes or whether the staff would be receptive to the intervention [
16]. Implementation studies allow for testing effectiveness of an intervention and at the same time investigating the implementation [
41]. We found that the medication reviews were implemented in all wards and almost all patients received a medication review. In contrast, in the ARCHUS study from New Zealand, using education and multidisciplinary teams, only 23% of the participants in the intervention group were discussed during the team meetings [
42].
Promotors and barriers affect the implementation process and depend on environmental factors, resources, beliefs about consequences, and social and professional roles [
18]. In our study, time and available resources are considered the greatest barriers to implementation in clinical practice. However, some participants highlighted the value of prioritizing an intervention they perceived as being important, leading to increased knowledge among the healthcare professionals. A systematic review demonstrates that key factor for successful interventions in nursing homes are the involvement of the stakeholders and management-level healthcare professionals and to enlist commitment to support nursing staff to prioritize the intervention [
43].
During the collegial mentoring, we encountered a number of ethical dilemmas when drugs or doses were changed or withdrawn. For instance, the deprescribing of anticoagulants or antibiotics often led to discussions about possible consequences. Interestingly, physicians tend to place more emphasis on actions than on omissions, and guilt deriving from negative consequences of an action is greater than guilt from inaction [
44]. We observed that the process may create an arena for discussion and problem solving, and that it may bolster the communication between the participants. The roles of each participant in the medication review process were clearly defined. This might explain the feeling of improved communication between all the involved parties, leaning a voice in the interprofessional discussion to everybody.
In our study, we met nursing home staff members who were eager to expand their knowledge. Although physicians were often more difficult to include, commitment among staff members has been highlighted as a key factor for successful implementation, whereas a high turnover rate in personnel appears to weaken participation and implementation [
45]. Despite the fact that the physician has the main responsibility for drug prescription, physicians have routinely been ignored in studies on medication reviews and only few studies include the attending physician [
15‐
17,
24‐
26]. Meanwhile, in other countries, engaged pharmacists may assist in medication review. This was not possible in our study, because in-house pharmacists are seldom available in Norwegian nursing homes.
The main strengths of this study are the comprehensive sample size, the variety of units, and the active involvement of the patients’ physicians and nurses. However, this paper also has some limitations. It was not in the scope of the present paper to report patient-related health outcomes, or to evaluate changes in quality of life, as called for by Alldred et al. [
15]. Our focus was to investigate the process in which the intervention was implemented as requested, which is often left out in other studies [
16]. However, effect of the study has been reported elsewhere [
20,
38]. The intervention might seem complex and time consuming, especially the assessment of patients’ pain and neuropsychiatric symptoms. Since these symptoms are common and drugs to treat these conditions are frequently prescribed, these assessments are essential. The use of two dedicated researchers in medication reviews is not feasible in clinical practice. On the other hand, we experienced during the medication reviews that it was the collegial discussion the physicians and nurses valued highly. We therefore suggest that local collegial networks might facilitate medication reviews. Unfortunately, we did not plan to use a structured assessment of the implementation Using a framework like RE-AIM would have strengthened the study [
46]. Future studies may benefit from using such frameworks. The setting and variety of units makes the results generalizable and possible to compare across nursing homes in Norway and comparable countries.
Acknowledgements
We wish to thank the patients, their relatives, and the nursing home staff for their willingness and motivation that made this work possible. Special thanks go to all the participating nursing homes in Øygarden, Sund, Askøy, Bergen, Kvam, Bærum, and Sarpsborg, Norway.