Inclusion And Exclusion Criteria

Published Date: 02 Nov 2017

Abstract

Background: Singapore is facing an aging population and as a result there are increasing elderly hospital admission, readmission and rising healthcare cost. Some healthcare organisations are either practicing or considering the implementation of an inpatient self-administered medication programme (SAMP) to counter this issues. Some studies had suggested that SAMP can improve elderly medication adherence and knowledge with positive feedback.

Literature Review: The review aimed to gather information to evaluate inpatient SAMP research studies on elderly patients’ medication adherence, knowledge and satisfaction. The research studies were systematically acquired and appraised. The finding suggested limited up-to-date SAMP researches. Most of the SAMP research studies were poorly designed and methodological flawed. Therefore, SAMP effectiveness has not been proven but cannot be refuted. More well-designed and high quality SAMP researches, especially in the local setting are needed.

Proposal: The proposal aimed to establish an evidence-based practice project of the non-adoption of SAMP in Singapore inpatient settings. The project also identifies the constraints related to non-evidence-based practice in healthcare, in order to correct them. The proposal is guided by the Joanna Briggs Institute (JBI) model of evidence-based healthcare and Institute of Medicine (IOM) recommendations for assessing evidence on clinical effectiveness.

Conclusion: Evidence-based practice is the foundation of maintaining a high quality healthcare standard. Therefore, it is important to promote healthcare practice that is based on rigours research evidences. However, through this paper, many will understand that implementing evidence-based practice is not always that straight forward. It could be due to lack of available evidence, constraints, influences, pressures, etc. These factors have to be taken into consideration when proposing evidence-based practice. It is also important to assess the organisation and individuals’ drivers and barriers to change as it would aid the planning and facilitate the change.

Introduction

An inpatient self-administered medication programme (SAMP) was first developed more than 50 years ago (Parnell and Marie 1959) and is currently practiced in developed countries like Australia (Davis et al 2002), United Kingdom (UK) (Brophy 2010) and United States of America (USA) (Florida Department of Elder Affairs 2012). It is a programme designed to engage patients in medication self-management and allows healthcare professionals, like nurses and pharmacists, to provide patient-centred medication education (Wright et al 2006). SAMP had caught the attention of Singapore healthcare organisation, whereby some hospitals are either practicing or considering the programme. Some speculation for Singapore’s sudden interest in SAMP may be due to its growing aging population, increasing healthcare cost and the tendency to adopt foreign practice. Due to physiological aging process, an older individual have a prevalence of chronic diseases, which require long-term medications and admissions to hospitals (Cheah 2001). In 2012, 11.1% of Singapore’s total population were made up of residents aged 65 years and older (Department of Singapore Statistic (DOSS) 2012). According to the statistics, it is believed that the elderly population would double by the year 2030 (DOSS 2012). In 2011, majority of hospital admissions are residents 65 years and above (DOSS 2012). With this, hospital admission and healthcare cost will rise with the aging population. According to Lim et al (2011), the hospital readmission rate in Singapore in those aged 65 years and older was 19% in 2010. Elderly hospital readmission has been associated with poor medication management at home (Col et al 1990, Tran et al 2011). Therefore, having a programme that may improve medication adherence and knowledge after discharge seems very attractive.

In a modern and developed country like Singapore, evidence-based practice holds up most importance and is crucial in maintaining and promoting Singapore’s high healthcare standards. Evidence-based nursing practice is defined as "the central to the notion of best practice, nurse accountability, and the need to ensure that nursing activities are transparent and safe" (Barker 2010 pg 3). Therefore, this paper aims to systematically gather and critically appraised SAMP research studies, in order to provide scientific evidences of the topic area through a review. The second section of this paper aims to propose an evidence-based practice project based on the review’s conclusion. The proposed project aims to aid evidence-based practice in Singapore’s healthcare organisations that are either practicing or considering SAMP.

Why SAMP?

One suggested definition of SAMP is the ability of patients to assume responsibility for taking their own medications under staff supervision whilst in hospital (Dooley et al 2005). It is also a strategy to identify and address problems associated with the discharge process (Davis et al 2002). This is true, as there is little designated time for nurses and pharmacists to deliver medication education to their patients only until discharge. However during discharge, patients are overwhelmed with information in a short period of time (Puvanendran 2011). Some possible benefits of having an inpatient SAMP are the improvement of patients’ competency and adherence to managing their own medications after discharge (Lever et al 2008). Reason being they have been practicing it throughout their stay in the hospital. In addition, nurses can detect trouble areas before discharge and intervene accordingly (Tran et al 2011). SAMP could reduce hospital readmission due to ineffective medication management (Grantham et al 2006), promotes patients’ self-care (Manias et al 2004) and may also increase patients’ medication knowledge (Furlong 1996, Ryan 1999).

Aim, Objectives and Research Questions

This review aims to gather information to evaluate research studies done on inpatient SAMP for elderly. The objective is to critically appraise the SAMP research studies, to determine any methodological issues of the research studies and how to overcome them. The review’s research question is does an inpatient SAMP improve elderly patients' adherence and knowledge to their own medications after discharge? The second research question is does the inpatient SAMP increase elderly patients’ satisfaction in medication serving in the hospital?

Inclusion and Exclusion Criteria

This review involves empirical quantitative studies, literature reviews and pilot studies that had researched on an inpatient SAMP that either improves and/or detects patients’ adherence and/or knowledge to their medication regime. The inclusion of pilot studies was due to the limited up-to-date research articles available on SAMP. This review also involves empirical qualitative studies that explore patients’ satisfaction on SAMP. Initially, articles published from 2003 to 2013 were set in inclusion criteria to guarantee up-to-date articles. However, after the databases search, there was very limited SAMP research articles as there were very few done in the last ten years. Most SAMP articles were found to be published in the 1970s to 1980s. This is evident from the Medline database search, where forty SAMP articles were acquired. This lead the author to speculate whether it was an idea that had past its time. If so, why SAMP is still implemented in developed countries? A rationale of the peak in SAMP studies in the 1970s to 1980s could be because it was a new intervention then. But to include papers that old would substantially reduce the rigor of this review. In addition, clinical practice and the sample population were very different back then compared to current times. In order to widen the scope, the year of publication was pushed back to 1990. Clinical practice and the sample population would not have differed much in the nineties compared to current times. All included publication must be in the English language. Also due to limited research articles, studied participants’ age where possible specifically related to people aged 60 and older. Participants’ regardless of nationality, ethnicity and illnesses (that does not limit the ability to self-administer medication) pertaining to older adults were included. The programme has to commence in an inpatient setting. The medication forms could be oral or topical medication, inhalers, or subcutaneous injections. The exclusion criteria were participants recruited in the intensive care unit or emergency department. The rationale for this exclusion was that participants should be in a state where they were clinically stable and their medication regime should have little changes. Finally, descriptive and anecdotal articles were excluded due to lack of rigour. A list of the inclusion and exclusion criteria can be found on Table 1 in the appendix.

Methods

Databases used in the literature search are the champion database for nurses, the Cumulative Index to Nursing and Allied Health Literature (CINAHL) via EBSCO, a drug research database, EMBASE via Ovid, Medline via PubMed, Wiley library and the systematic reviews database, Cochrane library (Cullum et al 2008). The use of different databases was to explore a greater pool of journal articles, in ordered to acquire the most appropriate and quality articles for the review. Formulated keywords used in the databases search were self-administration, medication, programme, scheme, hospital, inpatient, elderly, aged, older people, compliance, adherence, and medication knowledge. The databases searches were conducted using the Boolean operator. Each keyword was entered into the Boolean operator to evaluate the amount of hits achieved. Then, the keywords were mix and match, and combine together using ‘AND’ and ‘OR’ to search for the articles. Relevant articles were sieved out and read by the author. Articles were selected based on the inclusion and exclusion criteria.

Results

The final search results from each database can be found on Table 2 in the appendix. Do note that the databases had yielded a repeat of least one article. The reasons for not including many of the relevant articles were firstly, some were not available and/or free in full text electronically. Next, many articles did not involve a SAMP and many were from before the 1990s. A total of thirteen empirical studies were selected. Of that, eight were quantitative and three were qualitative studies. Under the quantitative studies, two were Randomised Control Trial (RCT) (Proos et al 1992, Lowe et al 1995) and two were quasi-experimental studies (Furlong 1996, Thornett et al 1994). There was also a prospective cohort study (Tran et al 2011), a post-test only with no control group study (Barry 1993), a post-test only with control group but no randomisation study (Wood et al 1992) and finally, one questionnaire study (Noy 1997). Under the qualitative studies, there was an action research study (Ryan 1998) and one questionnaire study (Deeks and Byatt 2000). Two pilot studies (Grantham et al 2006, Lam et al 2011) were also selected for the review. The summaries of each empirical and pilot study can be found on Table 3 in the appendix. Finally, three literature reviews (Collingsworth et al 1997, Ryan 1999, Wright et al 2006) were also selected for the review. The summaries of each literature review can be found on Table 5 in the appendix. This paper had selected a total of fifteen articles for its review.

Critical Literature Review

The review is written in three parts, where it explored whether SAMPs improved patients’ medication adherence, knowledge and satisfaction with their care, accordingly. The Critical Appraisal Skills Programme (CASP) for RCT was used as a guide to critically appraise the seven quantitative (except cohort study) and two pilot studies. The rationale is there was no validated tool for critically appraising study designs like quasi-experimental, questionnaire, pilot and post-test studies. The CASP for making sense of cohort research was used as a guide to critically appraise Tran et al (2011) study and the CASP for qualitative research was used to appraise the two qualitative studies. The CASP for review was used as a guide to critically appraise the three literature reviews.

Do SAMPs improve patients’ medication adherence?

A total of five primary studies (Wood et al 1992, Proos et al 1992, Lowe et al 1995, Lam et al 2011, Tran et al 2011) measured adherence as an outcome effect to SAMP. Two literature reviews (Ryan 1999, Wright et al 2006) had address medication adherence with regards to SAMP. Both reviews (Ryan 1999, Wright et al 2006) had reviewed similar primary studies with this paper, and had utilised a systematic search method. Wright et al (2006) had also used CASP as a guide to appraise the research articles and Ryan (1999) had used a criteria identified by Roe (1993) to critically review the articles. Three studies (Wood et al 1992, Proos et al 1992, Lowe et al 1995) had control but one study (Proos et al 1992) compared SAMP with a structured teaching group. Of those, only two studies (Proos et al 1992, Lowe et al 1995) were randomised. However, the authors had not used a proper randomisation technique and were not transparent about it. Proos et al (1992) did not mention how the participants were allocated into the groups. Conversely, Lowe et al (1995) selected four medical wards with interest in elderly patients in their trial. Two of the wards were randomly allocated to the SAMP group and the other two acted as control (Lowe et al 1995). Randomisation in intervention studies like these are important as it is the preferred method for equalising conditions and differences of any confounding factors across the groups (Holly et al 2011). It eliminates both conscious and unconscious bias regarding the selection of participants to the different groups (Salmond 2008). There was also no mention of concealment of allocation during randomisation in both studies (Salmond 2008). Researchers need to be transparent in this area for readers to decide whether the process used was truly random and adequate. Due to the lack of proper randomisation, there is a possibility of selection, allocation and confounding bias (Salmond 2008).

None of the studies here had performed blinding except Proos et al (1992) study, which will be discuss in detail under patients’ knowledge. Blinding is important in RCT studies as it reduces the confounding factors (Salmond 2008). It is understandable that blinding of the participants and nurses was not possible as the participants were receiving and the nurses were administering the care in these studies. Therefore, there is the possibility of observer and performance bias (Salmond 2008). However, blinding of the investigators seems visible in these studies and thus, efforts should be made.

All studies samples size was small and thus, representing the target population was in question. This was found to be similar with Wright et al (2006) review. The largest sample was 92 participants (Proos et al 1992) and the smallest was 24 participants (Lam et al 2011), which was not surprising as it was a pilot study. Tran et al (2011) a cohort study target population was the elderly population, however, the participants’ age ranged only between 79 to 90-years-old. It failed to include elderly aged 60 to 78-years-old. Four studies (Wood et al 1992, Proos et al 1992, Lowe et al 1995, Lam et al 2011) had an intervals ranged from 10 days to 3 months post-discharge follow-up. One study (Tran et al 2011) did the evaluation inpatient. Without a post-discharge follow-up the researchers cannot determine participants’ who passed the programme had remained independent and adherence to their medications after discharge. All studies with post-discharge follow-up had attritions. Only two studies (Proos et al 1992, Lowe et al 1995) explained their attritions in some detail. In Proos et al (1992) study, there was a 51% drop in feedback from the participants from both groups during the one month follow-up. The researchers should have taken precautions to prevent such a huge drop-out and, explore and explain the reasons for the poor reply. In Wood et al (1992) study, four participants from both groups had loss-to-follow-up during the three months post-discharge pill count and did not explain the reason for the lost. Three studies (Proos et al 1992, Lowe et al 1995, Lam et al 2011) had included the participants that had dropped out results when appropriate. All participants were analysed in the groups to which they were randomised or allocated to. Only Proos et al (1992) did a sample size calculation in their study. To achieve 80% confidence in detecting a difference between the two groups, the total sample size needs to be 112 participants (Proos et al 1992). Unfortunately, they did not meet that amount and lost almost half of their participants during the follow-up. This can be avoided by setting a larger sample size at the start of the study, to give room for drop-outs. In addition, power calculation should be set at 90% statistical power not 80% (Chow et al 2003). A sample size calculation of 90% statistical power was to ensure that the study had enough participants to minimise the play of chance (Chow et al 2003).

To evaluate adherence, four studies (Wood et al 1992, Proos et al 1992, Lowe et al 1995, Lam et al 2011) used pill counts. Of those, one study (Lam et al 2011) did not use pill counts during their post-discharge follow-ups. Studies had shown that the limitations of pill counts either masked excessive medication taking, medication dumping especially if the patients are aware that the assessor suspect non-adherence, or patient may adjust the number of pills either deliberately or inadvertently (Lowe et al 1995, MacLaughlin et al 2005, Wright et al 2006). All post-discharge pill counts were done physically by investigators, except in some participants in Proos et al (1992) study, where the follow-ups were done by telephone thus, the participants had to perform the pill count. Proos et al (1992) study findings may be inaccurate as a result of this. Lowe et al (1995) had taken steps to minimise the drawbacks of pill counts by not telling the participants there will be a pill count and by controlling the amount of medication supplied on discharge. Participants were also told not to use any of their home stock medications, thus making it difficult to adjust the number of pills during the home visit pill count (Lowe et al 1995). The methods used for rating patients as ‘complaint’ varied in the studies. Lam et al (2011) study used the Tool for Adherence Behaviour Screening (TABS) to evaluate adherence in post-discharge follow-up. TABS is a self-reported tool to measure adherence that screens both intentional and unintentional non-adherence, however, it has not been validated (Lam et al 2011). The limitation of self-reporting tools is the chances of participants reporting high adherence, especially post-intervention cannot be ruled out. One study detects and addressed the barriers to medication adherence prior to discharge (Tran et al 2011).

All studies here used statistical analysis and presented their findings with a P value of less than 0.05. In addition, two studies (Lowe et al 1995, Lam et al 2011) had included the Confidence Interval (CI) set at 95% with their results. None of the study had measured an effect size, which is important in indicating clinical significance. Lowe et al (1995) study findings had a fairly wide CI ranges which indicate poor precision. In Lowe et al (1995) findings, ten days post-discharge, SAMP participants yielded a statistical significant 95% mean compliance score compared to the control group of 83% (difference 12%, CI 4%-21%, P<0.02). In Wood et al (1992) study, at two weeks post-discharge, the SAMP group compliance was significantly better compared to the control group (p<0.02). Only 6% in the SAMP group made serious errors compared to 47% in the control group (Wood et al 1992). At three months post-discharge, compliance was still significantly better in the SAMP group compared to the control group (p<0.05) (Wood et al 1992). But, no significant difference was found between the findings for participants with less than 15% non-compliance (Wood et al 1992). In Lam et al (2011) study, participants who passed SAMP had achieved statistical significantly reduced non-adherence score with a narrow CI when compared the two weeks post-discharge score with immediately prior to level two (mean score: 5.7, CI 4.5-6.8, P=0.02). After participation in SAMP, a trend in improved adherent behaviour was found to be not significant (P=0.08) (Lam et al 2011). Lam et al (2011) findings should be taken with precaution as the recruitment of a relatively high functioning participant may explain the high baseline score (mean score: 89.2%). With the high baseline, it may limit the amount of improvement that could be demonstrated. Also, with no control group for comparison, the instrument may have acted as an educational intervention which could have helped improved the participants’ medication management.

Out of 50 participants that passed SAMP in Tran et al (2011) study, seven passed with nursing interventions. In summary, 30.6% participants had barriers to accurate and safe self-medication were identified that may not have been detected by the usual discharge process. Statistical significances was seen when variables such as regular medication at the start of SAMP, change made to regular medication during hospitalisation, Mini-Mental State Examination (MMSE) score, age, length of hospital stay and Medication Regimen Complexity Index (MRCI) score were compared between participants that passed and failed SAMP (Tran et al 2011). However, two of the assessment tool were not available for during routine assessment and were therefore not completed by some participants, which may have reduced the strength of the analysis (Tran et al 2011). Proos et al (1992) compliance score for the SAMP group was 94.00 (SD 6.54) and 90.05 (SD 11.55) for the structured teaching group. The difference in score between the two groups was not statistically significant. It would give a better comparison to usual practice if Proos et al (1992) used the nurse-administration of medication for its control arm. According to Wright et al (2006), the studies’ results are inconclusive because it is evident that some participants will comply and others will have significant non-compliance issues regardless of whether they participated in SAMP. Ryan (1999) review did not contribute much on the critical analysis of the research articles under SAMP and patients’ adherence.

Do SAMPs improve patients’ medication knowledge?

Seven primary studies (Proos et al 1992, Barry 1993, Thornett et al 1994, Lowe et al 1995, Furlong 1996, Noy 1997, Ryan 1998) researched the effect of a SAMP on patients’ medication knowledge. Two literature reviews (Collingsworth et al 1997, Wright et al 2006) had discussed patients’ medication knowledge with regards to SAMP. Collingsworth et al (1997) had utilised a systematic method to assess primary studies and used a framework to critique them. The framework was developed by the Collingsworth et al, but its reliability and validity is unknown. Both reviews (Collingsworth et al 1997, Wright et al 2006) had reviewed similar primary studies with this paper. Five studies (Proos et al 1992, Thornett et al 1994, Lowe et al 1995, Furlong 1996, Noy 1997) had control. Of those, only two studies (Proos et al 1992, Lowe et al 1995) were randomised. One study (Barry 1993) was a post-test only study and one study (Ryan 1998) was a qualitative study. SAMPs structure of these studies varied and may included verbal instruction, information leaflet, an individualised medication booklet, and/or reminders. Four studies (Proos et al 1992, Furlong 1996, Noy 1997, Ryan 1998) utilised a before-and-after study design which tested participants’ knowledge pre and post-intervention. The other studies (Barry 1993, Thornett et al 1994, Lowe et al 1995) tested post-intervention knowledge only. Having baseline information to compare with post-intervention measurement may introduce bias, because it sensitises participants during the baseline assessment (Wright et al 2006). This effect is anticipated to be greater in studies where knowledge is tested more than twice (Wright et al 2006). The before-and-after study designs are methodologically weak because other simultaneous change makes it hard to characterise observed changes to the intervention being studied (Wright et al 2006).

Three studies (Barry 1993, Furlong 1996, Noy 1997) used written questionnaires. Noy (1997) study had adopted a survey approach with open and closed questions, and had kept it short with only seven questions to encourage completion. But had not mentioned how the questions were scored. Barry’s (1993) tool was a six item questionnaire which was attached to their article. The questionnaire was used to only test the knowledge of only one medication and each item was answered with Yes, No or NA (Barry 1993). However, the reliability of what this questionnaire is truly measuring is in question. Barry (1993) questions were structured to measure subjective knowledge. Subjective knowledge is an individual’s perception of how much he/she knows (Popper 1966). Conversely, objective knowledge is a measure of what an individual actually knows (Popper 1966). Therefore, to truly test patients’ medication knowledge post-SAMP, the measure of objective knowledge should be utilised. Both Noy (1997) and Barry (1993) questionnaire were neither tested for reliability nor validity. Furlong (1996) had three separate questionnaires which were designed by the author and extensively piloted to test reliability and validity. But according to Furlong (1996), some questionnaires were administered through an interview, completed by a nurse or relative on the participant’s behalf. One of Furlong’s (1996) questionnaires contained 21 items, which was too lengthy for some and thereby had choose not to complete the questionnaire. These would impact the accuracy of the findings. In addition, even with three different questionnaires, it still contained some similar questions. Therefore, participants may have experienced a learning effect.

The reminding three studies had used structured interview method (Proos et al 1992, Lowe et al 1995, Ryan 1998) and one study had used semi-structured interview method (Thornett et al 1994) to collect knowledge data. Both Proos et al (1992) and Ryan (1998) had used a questionnaire to guide the interviews. Lowe et al (1995) tested participants’ knowledge by asking the purpose of each of medication. The form of participants’ response was noted in three studies (Proos et al 1992, Lowe et al 1995, Ryan 1998) and only Thornett et al (1994) study was unclear. Only Proos et al (1992) had provided details on how the questions were scored. The use of structure interviews is subjected to interviewer bias although this can be reduced if the investigator was blinded (Wright et al 2006). Blinding of the investigators was only performed in Proos et al (1992) study. Two investigators, blinded from the participants’ groupings, separately and independently scored the questionnaires (Proos et al 1992). The investigators would then discuss and reach consensus on the few discrepancies in the score (Proos et al 1992). By doing so, it would prevent or reduce observer, detection and outcome bias (Salmond 2008).

The largest sample was 107 participants (Noy 1997) and the smallest was 20 participants (Ryan 1998). Four studies (Barry 1993, Thornett et al 1994, Furlong 1996, Noy 1997, Ryan 1998) had used non-probability sampling which would impact the generalisability of the target population. Four studies (Proos et al 199, Lowe et al 1995, Furlong 1996, Noy 1997) had attritions during post-discharge follow-up and one study (Ryan 1998) had attrition during the study. Barry (1993) study was not clear on the reporting and therefore, was unable to determine the total sample size and the attrition rate. Thornett et al (1994) had no attrition. The highest attrition rate was 82% (Noy 1997) and the lowest was 10% (Lowe et al 1995). According to Noy (1997), the very high attrition was due to having questionnaire questions that were difficult to interpret and hard to understand. In addition, 28% of participants’ data could not be used due to the nurses forgetting to issue out the questionnaires. The questionnaire should be designed using simple language, avoid jargons and ensure it is not too lengthy. A limitation of having the nurses undertake the study evaluation would result in poor control for investigators and observational bias. The researcher should also consider the nurses workload before involving them in data collection. A recommendation would be hiring a designated data collector.

Three studies (Proos et al 1992, Lowe et al 1995, Furlong 1996) used statistical analysis and presented their findings with P value (<0.05). None of the study had measured an effect size. Proos et al (1992) findings demonstrated a statistical significantly better mean pre and post knowledge test score in medication (P<.025), frequency (P<.025), purpose (P<0.01) and side effects (P<0.01) when compared with the structured teaching group. Only dosage was found to be not statistically significant (Proos et al 1992). In Lowe et al (1995) findings, 90% of SAMP participants knew the purpose of their medications compared to 46% of participants in the control group (difference 44%, CI 29%-63%, P<0.001). In Furlong (1996) study, both group had a significant improvement in medication name and dose, and purpose from admission to 8 weeks post-discharge. A significant improvement in medication timing was only seen in the SAMP group from admission to post-discharge (Furlong 1996). Both groups did not show significant improvement in medication side-effects (Furlong 1996). The results have to be viewed with caution due to the small sample size in the nurse-administered group because for statistical reliability, the sample size should be twenty or more (Furlong 1996, Wright et al 2006). Barry (1993) study findings were that knowledge in medication name, dose, two side effects and special administration instructions were higher in SAMP participants post-implementation. However, medication timings and purpose were found to be higher pre-implementation (Barry 1993). The data is highly likely to be skewed due to more participants being assessed after the programme compared to pre-implementation, and the test of subjective knowledge.

Thornett et al (1994) study findings were that sixteen SAMP participants stated that they knew their medication’s name and purpose. Eleven SAMP participants and only three control participants mentioned that they knew more about their own medications since their stay in the hospital (Thornett et al 1994). However, there was very little difference in medication side-effects knowledge between the two groups (Thornett et al 1994). This study was found to be similar to Barry (1993) study where it tested subjective knowledge. The study reporting was also inadequate as the article was only two pages. In Noy (1997) study, before SAMP, participants’ knowledge of their medication span from 0-100%. However, after SAMP, approximately three-quarters of the participants knew 81-100% of the expected information (Noy 1997). The medication side-effects were an area that the participants know least about (Noy 1997). But this changed from 75% of participants knowing nothing about their side-effects to 25% after the programme (Noy 1997). The amount of participants knowing all their side-effects increased from 2% to 38% (Noy 1997). Due to too small of a sample during the six weeks post-discharge, it is hard to draw any conclusion from the data. Ryan (1998) finding were that all participants had an improvement in their medications knowledge in approximately two weeks, but the knowledge improvement varied between participants. There was an increase of 25%, which suggest that participants were more knowledgeable about their medications at the end of the programme (Ryan 1998). Participants’ motivation was an important factor in learning because high numbers of successful participants were found to be living alone, thus had appeared to have the desire to self-manage their medications (Ryan 1998). There were some limitations to Ryan (1998) study, firstly, an ethics committee was not involved as it claimed to have no major ethical issues associated with the study. Next, the participants’ demographics and data analysis was not discussed at all. Most of the studies contained some education component which makes it hard to determine whether SAMP, the educational component or both is responsible for increasing the participants’ knowledge (Collingsworth et al 1997).

Do SAMPs improve patients’ satisfaction with their care?

It is difficult to measure satisfaction in healthcare, particularly when patients’ are still in the hospital or required further care, as this may bias their response (Wright et al 2006). This can be minimised by measuring patients’ satisfaction during discharge. Five primary studies (Lowe et al 1995, Furlong 1996, Ryan 1998, Deeks and Byatt 2000, Grantham et al 2006) researched patients’ satisfaction on SAMP was selected for this review. One literature review (Wright et al 2006) had reviewed studies with patients’ satisfaction in regards to SAMP and had reviewed similar articles as this paper. Two were controlled studies (Lowe et al 1995, Furlong 1996). Of those, one study (Lowe et al 1995) was randomised. Two studies (Ryan 1998, Deeks and Byatt 2000) were qualitative design and one study (Grantham et al 2006) was a pilot study. All studies used either written questionnaire (Furlong 1996, Ryan 1998, Deeks and Byatt 2000, Grantham et al 2006) or structured interview (Lowe et al 1995). One study used Likert scale (Furlong 1996), and Deeks and Byatt (2000) used a variety of questioning methods. Details of the questions were not provided in two studies (Lowe et al 1995, Ryan 1998). One study (Deeks and Byatt 2000) mailed the questionnaire to the participants and another study (Grantham et al 2006) administered the questionnaire during discharge. One study (Ryan 1998) administered the questionnaire at the end of the programme and Lowe et al (1995) study conducted the interview at participants’ homes. The likelihood of misunderstanding or inaccurate reporting is the limitations of a self-completed written questionnaire (Wright et al 2006). It was unclear when Furlong (1996) had measured participants’ satisfaction. There was a significant imbalance of older participants in Deeks and Byatt (2000) study control group. This may be due to the unclear inclusion and exclusion of participants in the SAMP group which may have resulted in a large amount of participants who could not self-administer their medication to be in the control group.

The best response rate was 90% (n=79/88) (Lowe et al 1995) and the worst was 16% (n=34/207) (Grantham et al 2006). 95%-80.9% of participants who had an experience SAMP responded in favour of self-administration compared to nurse-administered (Lowe et al 1995, Furlong 1996, Deeks and Byatt 2000). There was no figure given in two studies (Ryan 1998, Grantham et al 2006) of participants’ satisfied with SAMP. Ryan (1998) only mentioned that there was a high level of satisfaction with the programme and 75% of participant would recommend it to others. Grantham et al (2006) stated that all participant surveyed had reported satisfaction with SAMP in level of support from staff, the amount of information provided and the use of the medication information card. In Furlong (1996) study, five participants from the control group preferred nurse-administration because they believe it was more convenient and save the nurses time. Conversely, nine participants from the intervention group had preferred SAMP because it saved nurses time. Only Deeks and Byatt (2000) study had statistical analysis. Deeks and Byatt (2000) reported that participants in the intervention group were significantly more likely to prefer SAMP compared to those who were not (P<0.0001). Less than 10% of older participants from the intervention group would not prefer SAMP in the future compared to those aged under 60-years-old. In contrast, younger participants from the control group would significantly prefer SAMP in the future compared to participants aged 60 years and over (P<0.005). Majority of participants who had experienced a SAMP will be satisfied and would choose it again, but most of the participants who had never experienced a SAMP would most likely choose nurse-administration (Wright et al 2006).

Discussion

Most of articles participants’ age can correlate to the research population of this paper. The articles participants’ illnesses and clinical settings do not differ much to the local population. However, there may be cultural difference as the articles participants’ are mainly Caucasians and this review’s population are mainly Asians. In the Asian culture, majorly of elderly are known to be more dependants compared to their western counterparts. The finding of this review concludes that SAMP researches lacks of appropriate well design studies and the use of sound research methodologies. Most of the studies reviewed had serious methodological flaws. In addition, there is very limited research done on SAMP especially in the last ten year. Many articles reviewed here are considered out-dated. Some of the quasi-experimental studies based on single sites, limits the generalisability. All studies used too small sample size or had low responses rate, and little or no attempt was made to establish the impact that non-responders or drop-outs might have on the results. Most of the studies were either conducted in one or two wards in a hospital, which also contributes to poor generalisation. All the RCT studies had no proper randomisation and allocation of concealment. Only one had performed blinding in their study. Most of the studies measuring instruments were not validated and reliable. Methods like pill counts to measure adherence and written questionnaire to measure knowledge may be bias and inaccurate. On top of that, some studies did not follow a standardised method of collecting the data. All the studies did not utilise a sample size calculation of 90% statistical power to ensure adequate sample size to reduce the play of chance. Although most of the studies yielded statistical significances in their findings, it should be viewed with caution due to many confounding factors that may influence the results accuracy. Some studies had drawn and concluded the effectiveness of SAMP without a firm foundation. Many articles reporting were not transparent, which make conclusion hard to draw. This review findings correlates with Collingsworth et al (1997), Ryan (1999) and Wright et al (2006) reviews’ conclusion. Evidence of an inpatient SAMP benefits on elderly medication adherence and knowledge had neither been proven nor refuted. This can only be known through high quality researches and subsequent systematic reviews. All articles reviewed here had recommended the need for more well designed SAMP researches.

Proposal

Due to the lack of SAMP evidence yielded, this proposal aims to advocate the non-adoption of SAMP to Singapore’s healthcare decision makers. As some local hospital has started practicing SAMP, it is important to find out why the decision makers had implemented SAMP with so few drops of rain in this evidence desert. The word ‘evidence’ used here means the basis of belief; the approval or validation that is needed in order to believe that something is true (Miller and Fredericks 2003). According to Pearson et al (2011), in the "real world" of clinical practice and policymaking, decision makers are influenced by a diverse understandings and source of evidence that flows over the situation. These influences could be pressures of society’s greying population, influence from foreign policies, financial constraints, lack of relevant published evidence, etc. Therefore, this paper proposes a project that not only promotes evidence-based practice, but identifies the constraints related to non-evidence-based practice in healthcare, in order to correct them. The first objective is to set up a project to present the lack of evidence in a systematic way. The project will use of the Joanna Briggs Institute (JBI) model of evidence-based healthcare (Pearson et al 2011) and Institute of Medicine (IOM) recommendations for assessing evidence on clinical effectiveness (Eden et al 2008) as guide to provide safe, quality and patient-centred nursing care. The target population here are Singapore healthcare organisations that are either practicing or considering SAMP. The second objective is to include barriers and drivers to change in the project, in order to aid the organisations that decide to adopt the change.

JBI Model of Evidence-Based Healthcare and IOM Recommendations

The JBI model of evidence-based healthcare (Figure 1 on appendix) was developed by Pearson et al in 2005 (Pearson et al 2011). It is for healthcare professional to use research evidence balanced with the context of care, patient values and choices and the experience, expertise and clinical judgment of the professional (Pearson et al 2011). Hence, the healthcare professionals are then in a position to make evidence informed decisions (Pearson et al 2011). The model consists of four major components: healthcare evidence generation, evidence synthesis, evidence/knowledge transfer, and evidence utilization (Pearson et al 2011). The JBI model address evidence-based health in cyclical process where questions, concerns or interests derive from the identification of global healthcare needs by clinician or patients (Pearson et al 2011). Then model continues to address these questions by generating knowledge and evidence to effectively and appropriately meet these needs in ways that are feasible and meaningful to specific population, cultures and settings (Pearson et al 2011). This evidence is then appraised, synthesized and transferred to service delivery settings and healthcare professionals who then utilise it and evaluate its impact on health outcomes, health systems and professional practice (Pearson et al 2011).

The advantage of the JBI model is that it is very comprehensive and still structured for easy understanding. It also recognises all different types of evidence, from systematic reviews to expert opinions. It provides healthcare researchers the guidance when there is lack of evidence in a topic area. A disadvantage is that some may misinterpret the JBI model’s recommendation to use of low quality evidence as "best available evidence" and apply it to practice, when further research is needed. The use of intelligence interpretation of the model’s recommendations is needed to rip the fruits of its practicality.

The IOM is a committee from USA on reviewing evidence to identify highly effective clinical services board on healthcare services (Eden et al 2008). Its framework provides a general blueprint for national clinical effectiveness assessment programme that is quite similar to the JBI model (Eden et al 2008). However, it is not as in-depth and structure. But the IOM committee have useful recommendations that would aid the development of standards and processes that achieve trustworthy, unbiased and clear syntheses of the available evidence on clinical effectiveness for patients, board of the organisation, clinicians, policy makers, researchers and other decision makers (Eden et al 2008).

Global Health

As mentioned above, there are valid drivers and reasons for an inpatient SAMP. According to an Australian study about patients’ perspective on SAMP, a number of patients were interested and fond of the programme (Manias et al 2004). However, the reason for the interests ranged from taking control of their own medication to helping ward nurses save time (Manias et al 2004). Singapore practitioners may be pressured by society’s greying population and the dependent characteristic of our elderly. In addition to the increasing healthcare usage and cost related to admission and readmission due to poor medication management Singaporean elderly. To generate this evidence effectively and appropriately, local qualitative research on patients’ perspective, as well as nurses perceptive, on SAMP should be conducted first (Eden et al 2008, Pearson et al 2011).

Healthcare Evidence Generation

Many authors view quantitative RCT as the gold standard in research design for testing intervention effectiveness (Evans 2003, Mantzoukas 2007, Burls 2009). But it is not possible to conduct a RCT study design in all healthcare researches. According to Pearson et al (2011), when there is no research evidence that exists on that level, other evidence may be represented for the ‘best available evidence’ for a specific question. For healthcare policy and practice, "best" evidence derives from rigorous research, experience, expertise, inference and deduction (Pearson et al 2011). In case of SAMP research studies, the feasibility, appropriateness, and meaningfulness can be positive. Feasibility can be seen from the foreign SAMP policies and guidelines. The policies and guidelines had shown that it is possible to integrate and conduct SAMP into an inpatient setting (Deputy Chief Pharmacist for Clinical Service 2010, Pharmacy Department 2010). In terms of meaningfulness, some qualitative SAMP studies had demonstrated positive experiences by the patients (Manias et al 2004, Deeks and Byatt 2000). For SAMP appropriateness, some studies mentioned that there are predictors to elderly capacity for medication management in a SAMP (Bailey et al 1995, Maddigan et al 2003). Therefore, only for elderly deemed physically, mentally and cognitively fit to self-administer their medication might be able to benefit from the programme.

However, according to the review findings, there are doubts of the effectiveness of SAMP. Some studies mention that they had demonstrated effectiveness in improving patients’ medication adherence and knowledge. However, due to the poor design and flawed methodology of the studies, the true result of the effectiveness is unknown. In addition, other studies had also suggested that educational component and/or reminder devices paired with SAMP will only then make it successful (Ryan 1998, Logue 2002, Shen et al 2006). Also, none of the studies presented in the literature review has included relative risk, absolute risk difference and number needed to treat (Cullum et al 2008). These are the three common methods used to calculate the effect size, which is important in measuring meaningful interpretation of clinical effectiveness (Cullum et al 2008, Burls 2009).

There are a variety of evidence source in healthcare. There are internal sources such as experience, expertise and reasoning, and external source like research (Pearson et al 2011). In order not discount internal sources, as they can be useful, it has to be paired with research to achieve scientific-based evidence (Pearson et al 2011). Nonetheless, the research has to be rigorous (Eden et al 2008). The implementation of SAMP is currently based on internal source and low quality researches which is not robust enough for clinical implementation.

Evidence Synthesis

Both Eden et al (2008) and Pearson et al (2011) agreed that the most reliable evidence synthesis for practice is the systematic review of researches. It is a common, internationally accepted approach to evidence synthesis (Eden et al 2008, Pearson et al 2011). The systematic review is a secondary research that selects, analyse, and critical appraise primary studies of a similar topic (Eden et al 2008). There is a protocol for performing a systematic review to ensure scientific rigor and minimise potential bias (Eden et al 2008). The protocol consists of the review question, background to the review, inclusion/exclusion criteria, search strategy and review methods (Eden et al 2008). The review methods include assessment of methodological quality, data extraction and data synthesis (Eden et al 2008). There is a misconception that systematic reviews are reviews of only RCTs. However, according to Pearson et al (2011), a systematic review may utilise multiple forms of research, as long as it follows a standard criteria that are addressed in the protocol. In the first part of this paper, there are no SAMP systematic reviews available, only literature reviews. All three selected literature reviews (Collingsworth et al 1997, Ryan 1999, Wright et al 2006) and the review conducted in this paper had used a similar protocol. However, the reviews still need more information before publishing and presenting it. It is up-most importance to provide rigorous evidence reviews to determine what is known and not known, uncover the gaps and provide pointers in questions that decision makers need to find answers in SAMP (Eden et al 2008). As both answers and questions are developed, the review takes on a critical role in providing guidance based on evidence which this proposal aims to follow (Pearson et al 2011).

Evidence Transfer

Evidence transfer applies to the practical puzzle of transferring pre-reviewed, appraised, robust evidence to practice area (Pearson et al 2011). It associate capturing, organising and distributing evidence in forms that ensure it is available, accessed and ultimately used in practice (Pearson et al 2011). The transferring of evidence in this paper demonstrates the non-adoption of SAMP into practice as it differs with the best available evidence. There are several critical factors to evidence transfer. First, is the perceived strength or reliability of the evidence that is to be transferred. A study of healthcare professional in Australia, Indonesia, Malaysia, Philippine and Thailand suggest that they are more likely to access and apply evidence-based knowledge if it is clear and provides answers to clinically or policy relevant questions (Turner 2009). Therefore, to make things easier for the decision makers, a grading system of the evidence will be utilised in the literature review (Pearson et al 2011). According to Hutchinson and Baker (1999), this process is important when making recommendation for practice as it facilitate healthcare professionals to distinguish recommendation based on consensus from those based on good quality studies with consistent findings. There are many grading systems available. However, some only accommodate study designs in the hierarchy of evidence (Pearson et al 2011). In SAMP, there are all sort of quantitative and qualitative study designs. To accommodate this variety of evidences, a complex scale developed by the Centre of Evidence Based Medicine in Oxford will be use to grade of evidence in the literature review (Pearson et al 2011).

The second factor of evidence transfer is the way in which the evidence is delivered (Pearson et al 2011). Evidence can be delivered or distributed through written or orals means that are formal or informal and includes education, training, online delivery systems and printed resource (Pearson et al 2011). The author has to tailor information depending on the target group (Jacobson et al 2003). The target group here are policy makers, nurse administrators and nurse clinicians. If it was for the ward nurses or patients, the amount of information presented would be different. For the target group, the review would provide the most reliable foundation for evidence-based information (Pearson et al 2011). The review components should include a comprehensive background that justifies conducting the review, a description of the objectives, the inclusion/exclusion criteria that were used for selecting the primary studies, the search strategy and methods used for critical appraisal, extraction and synthesis of data (Eden et al 2008). The review of literature results should encompass the description of the studies, including the types and numbers of paper indentified (Eden et al 2008). The amount of articles that were included and excluded should be clearly explained, and a summary of the overall quality of the literature identified is presented (Pearson et al 2011). The results section should be organised in a structured and meaningful way based on the objectives of the review and the criteria for considering studies (Eden et al 2008). Therefore, the review in the first part of this paper still needs to be written with the standards above before publishing and presenting it as evidences in the non-adoption of SAMP to the decision makers.

The third factor is presentation of evidence, where readability is probably the foremost consideration when marketing a translation project (Pearson et al 2011). In Majid et al (2011) study of Singapore’s nurses perspective on evidence-based practice, the inability to understand statistical terms and inadequate understanding of the jargon used in research papers were among the top barriers. This suggests that the translator’s task is to break through linguistic barriers in order to make things easy for the readers (Pearson et al 2011). The overall appearance holds equal importance in the end users standpoint as it determines whether or not someone will read it (Pearson et al 2011). The author will use the aid of the JBI COnNECT+, a web-based platform, to format the review (Pearson et al 2011). Therefore, the SAMP literature review design and layout will have visual appeal element to it and is written with consistent terminology, clear definitions and is well structured.

The final factor is organizational context within which evidence transfer occurs. According to Ferlie and Shortell (2001), interventions to change clinical practice must be address in multiple levels in the organisation including the nurses, nurse clinicians, policy makers, the organisation itself and the larger healthcare system. Therefore, the influence cannot be overlooked. There are also two factors to consider. First, it would depend on the organisation culture as Rycroft-Malone et al (2002) mentioned that an organisation that is described as a learning organisation would be more receptive to the promotion of evidence-based practice. Second, the person leading the change needs to have leadership qualities in order to influence and lead others to achieve a common goal (Marchionni and Ritchie 2008). With these four factors and the proper channels within organisation and team system, the evidence transfer can occur (Marchionni and Ritchie 2008). The evidence transfer in this proposal would be achieved through publishing the literature review into journals for wider target audiences and presentation of the reviews findings in the target organisations.

Evidence Utilisation

This section of the model refers to the implementation of evidence into practice (Pearson et al 2011). The task is to change any current practice of SAMP and promote the non-adoption of it. There are three factors to consider and the first is getting the evidence into action (Pearson et al 2011). There are often significant barriers and opposition in the dissemination and implementation of evidence-based practice, thus, makes it such a challenge to healthcare professionals (Hutchinson and Baker 1999). The National Health and Medical Research Council in Australia (NHMRC 2000) suggest four applications in order for the successful translation of research evidence into action. First, it includes good information based on research that is capable of standing up to rigorous critical appraisal which is the literature review (NHMRC 2000). Next are the effective mechanisms and strategies for dissemination that enable nurses and decision makers to assess the information (NHMRC 2000). This would be publishing the review in online journals and setting up educational presentations in the institution itself. Third, is the physical and intellectual environment in which research is valued and uptake of research based knowledge is support and encouraged (NHMRC 2000). This would depend in the organization’s culture towards teaching and learning. Even if the organisation is not as receptive, the form of this proposal does impart, encourage and promote evidence-based practice. Finally, the intervention that promotes the uptake of knowledge and lead to behaviour change (NHMRC 2000). This would be the application of a change model.

The second factor to evidence utilisation is barriers to change (Pearson et al 2011). The ability to overcome barriers would increase the success of change. The need to address not only the information requirements of the healthcare professionals and organisations, but also the social and organisational aspect that conflict with the value of the reviews findings (Pearson et al 2011). The barriers may occur at diverse levels, regardless of whether it is organizational, staff or consumer driven, and the barriers will vary substantially for the stakeholders involved (Eden et al 2008). Reasons for an organisation to continue the practice of SAMP could be that there were already too much invested in the programme. Therefore, it would be a financial barrier. A strategy to identify such barriers prior to initiating any approach to facilitate change is to conducted interviews with all key stakeholders (Pearson et al 2011). The interviews would also provide them with an opportunity to feel involved in the change (Pearson et al 2011). There would not be any major barriers to change if an organisation is only in the mist of implementing or considering SAMP implementation. A solution for organisation that had invested in SAMP implementation could integrate it into research, where their investment into the programme would not go into total waste. Common potential barrier to change are staff knowledge and skill deficit, psychosocial barriers, organisation barriers and resource barriers (Grol and Wensing 2004). Strategies can be developed with the clinical team to overcome or, occasionally, ways around the barriers (Grol and Wensing 2004). The barriers and drivers that influence change at different levels of healthcare are available in appendix Box 1.

The final factor is leading and managing change (Pearson et al 2011). A change that is well planned would be easier to manage compared to change which is imposed, rushed or misunderstood. Therefore, it is important to understand why people resist and the ways in which resistance happens, as it can help healthcare professional lead and smooth the path of change (Pearson et al 2011). Resistance can happen in two ways: resistance effectively hampers and block change, and resistance maintains order and stability. According to Pearson et al (2011), one of the best ways to minimise resistance is to actively involve the decision makers. By giving them the opportunity to contribute and become part of the change would enable them to assist the change instead of hindering it. By doing so, it can create a sense of worth and recognition in the decision makers.

Evidence-Based Practice Change

The transfer of evidence into practice is slow and there is no solution that would guarantee an effective change process (Pearson et al 2011). However, there are models and strategies that could aid the change process. Change model like the Plan-Do-Study-Act (PDSA) cycle assist the overcoming of the natural resistance of an organisational change (Pearson et al 2011). This model can be suggested to the organisation to help them with their change process. Also, when planning change, the individuals concerned and the organisation are two major areas for consideration (Eden et al 2008). Studies evidences suggest that successful change in healthcare requires an atmosphere that is genuinely collaborative, cooperative, democratic, non-hierarchic and involves all stakeholder parties (Eden et al 2008). It is recommended that involving key stakeholders will the aid an effective and coordinate change because the expectations of all parties were involved (Eden et al 2008). Before proposing the non-adoption of SAMP in an organisation, it is important to assess the organisation culture on its receptiveness to evidence-based practice as it will establish the degree of support and the extent to which the proposed change success (Pearson et al 2011). It is also important to assess the organisation structure to which decision-making is decentralised and whether there are sufficient staff members available to support the change process (Pearson et al 2011). The rational for assessing the organisation structure is to respect the organisation’s hierarchy by administering the proposal through ensure proper levels. Another strategy to ensure a successful change is communication. Therefore, the research team should establish an adequate communication system to support exchange related to the change and the implementation process (Pearson et al 2011).

Project Overview

This project will run for seventeen months. After acquiring the funding, office space and hiring staffs, the lead research will train the staff. The team will then work together to improve the literature review with the use of the systematic review protocol, standards set by Eden et al (2008) and Pearson et al (2011) and the Centre of Evidence Based Medicine in Oxford to grade the evidences. Next is to publication of the literature review to journals and online journals that are freely accessible to all healthcare professional especial nurses, students and researchers. One of the researchers will source and contact the organisations that are either practicing or considering SAMP. With permission from the organisation, one researcher will conduct interviews to the key stakeholders of the organisation to assess the drivers and barriers to change, while another researcher will brief designated nurse as project champions. The research team will present the lack of evidences in SAMP in an educational, informative and interactive method in two separate presentations. One presentation will only be for crucial members of the organisation, like nursing directors, policy makers, clinicians, practitioners, managers and researchers. This presentation involves information on how to make the change, solutions to overcome the organisation’s assessed barriers; encourage more rigours SAMP researches to counter the lack of evidence and provide information to other evidence-based interventions to improve elderly patients medication adherence and knowledge. The actual change will not be forced and is up to the organisation’s decision. Once the approval is give to conduct the large scale presentation, the champions are responsible to distribute the flyers to the wards and different departments’ nurses and encourage their attendance. The second presentation will be large scaled as it is open to all of the organisation nurses. There may be more than one presentation to target the shift working nurses, and this presentation will be presented in a less technical method to facilitate easy understanding. The presentation will also take the opportunity to promote and educate on evidence-based nursing practice. Refer to Table 6 on appendix for project timeline.

Evaluation Methods

An external company will be hired to conduct the project evaluation in order to obtain a non-biased evaluation and experience project evaluation advice may be sought (Worthen et al 1997). Some issues that may be faced when collecting the data would be how feasible it is to collect data in the allocated time frame; identification of the structure, process and outcome measure that the author wants to record; and what methods are available to collect data (Pearson et al 2011). Other issues like who will enter the data in a computer programme and who will do the analysis; indication of what may bias the data collection; and who will write the report and/or present the findings would not be an issue if an external company is in-charge of conducting the evaluation (Pearson et al 2011). A before-and-after qualitative design will be used for the presentation evaluations (Robson et al 2001). This design offers evidence about the project’s intervention effectiveness and is