An evaluation of community pharmacy-based services for type 2 diabetes in an Indonesian setting: patient survey

Setting and sample recruitment

The aims of the study included the estimation of the prevalence of service usage and patient views. A sample size of 200 was defined which if broadly representative of the population of pharmacy patients with type 2 diabetes in Surabaya, the 95% confidence intervals for the true prevalence estimates would be within 7% of the figures obtained from the sample (based on a prevalence estimate of 50%). For the analysis of characteristics associated with respondents’ views, a sample of 200 would be expected to be adequate to identify any independent variables exhibiting a moderate to small effect size (with power = 80%, α = 0.05) (Tabachnick & LS Fidell, 2007).

A previous pharmacist survey had been conducted on a random sample of 400 community pharmacies in Surabaya (60% response rate), providing data on the characteristics of the community pharmacies and diabetes services provided (reported elsewhere) (Wibowo et al., 2015). Based on the pharmacy characteristics data, 10 community pharmacies were purposefully selected as sampling sites in this study, aiming to include different geographical areas and socio-economic levels in Surabaya. Surabaya consists of 31 sub-districts which can be categorised into five geographical areas, namely: centre, west, east, north and south; and four socio-economic levels, from high (labelled ‘1’) to low (labelled ‘4’) (Nugroho, 2010). Each pharmacy was responsible for recruiting approximately 20 patients. Patients eligible for the survey were those aged over 18 years, with a diagnosis of type 2 diabetes for which they were receiving oral antidiabetic medications. Patients were recruited as they were seeking oral antidiabetic medications at these pharmacies, and their written consent was obtained.

Data collection

Data analysis

Descriptive statistics were used to summarise the patient characteristics: demographics (Section A), diabetes profile (Section C), and monitoring profile (Section D). SPSS version 19.0 was used to perform the analysis.

In relation to diabetes services (Section B), frequencies were calculated for binary responses (‘yes’/’no’) related to the patient usage of services and for responses from Likert scales related to the extent of patient agreement regarding pharmacists’ roles. Moreover, content analysis was used for responses from the open-ended question to explore patient views on the five priority services that should be provided at the pharmacies. An initial coding frame structure was established from the generic model generated from the literature. The responses were coded (if new codes emerged, they were added to the thematic codes in the coding frame), and frequencies were calculated for each code (Rose, Spinks & Canhoto, 2015).

It was of particular interest to identify characteristics associated with the views of patients related to pharmacy-based diabetes services. For this reason, the responses regarding patients’ views for each type of service were classified into binary variables, to indicate ‘strong agreement’ (Likert scale ratings of 5–6) versus ‘ambivalence or disagreement’ (Likert scale ratings of 1–4) that a service should be provided by the pharmacy. These binary variables were used as dependent variables in logistic regression models to identify patient characteristics associated with the strong support for each role. Some roles were considered to be different aspects of an overarching role. For example, the roles to provide information on: the taking of medications, use of insulin devices, storage of medications, precautions, and adverse effects were all components of ‘medication education.’ In order to analyse characteristics associated with strong support for an overarching or ‘composite’ role, the arithmetic average of the Likert responses for the component roles was calculated and then converted to a binary variable in a manner similar to that used for the individual roles (scores of 5 or more were taken to indicate strong support, otherwise ambivalence or low support). Taking the simple average of the component roles implicitly gives equal weight to each of the roles within a composite. It would have been preferable to use Factor Analysis to identify if some of the component roles were more important than others, but the sample size was considered too small for this refinement. It is generally recommended that sample sizes of 300 or more should be used to obtain stable factor loadings (Tabachnick & LS Fidell, 2007). Patient characteristics included as independent variables were gender, age, education, employment, income, health insurance cover, diabetes organisation membership, duration of diabetes (time since diagnosis), risk factors for complications, complications and diabetes (glycaemic) control. The models also included an independent variable indicating patients’ previous use of the service (binary responses: ‘yes’/’no’). For dependent variables which were ‘composite’ (for example ‘medication education’), the binary responses indicating previous use of each component service were treated as numeric (zero for ‘No’, one for ‘Yes’), and their mean was calculated and used in the model as an independent variable to show the degree of previous use of the composite service. A mean value close to 1 indicated that most components of the composite role had been used, while a lower value indicated less use. A backward elimination strategy was used to identify all the variables which significantly contributed to each model. Through this approach all independent variables were included initially, and then the least significant variable was dropped (one at a time) until the p-value associated with each of the variables remaining in the model was less than 0.05.

Sample recruitment

This study included 10 community pharmacies as sampling points after approaching 11 community pharmacies in Surabaya. One pharmacy refused, as the employee pharmacist was planning to resign and they were in the process of recruiting a new one. From the final 10 pharmacies, a total of 336 questionnaires were distributed and 204 were returned; however, eight patients reported the use of insulin at the beginning of their therapies and were deemed to have type 1 diabetes, leaving a sample size of 196 (a response rate of 58.3%) (Table 1).

Characteristics of pharmacy patients with type 2 diabetes

The demographic information of participating patients (Section A) is summarised in Table 2. Approximately 60% of respondents were female and half of respondents were aged 60 years or older. Approximately half of the respondents did not have health insurance plans.

The diabetes data reported by participating patients (Section C) are shown in Table 3. The median duration of diabetes (time since diagnosis) was seven years. More than 80% of respondents reported that their treatment regimen included oral antidiabetic medications and diet modifications. It is important to note that this variable did not measure whether the doctor made the medication/diet/exercise recommendations, but rather whether the patient remembered and/or reported it. Almost 60% of the respondents reported having at least one diabetes-related complication, and most of the patients reported having at least one risk factor for complications.

Table 4 shows the monitoring profile reported by participating patients (Section D). HbA1c refers to glycosylated hemoglobin, which identifies average plasma glucose concentration over the previous three months. To provide insight to patients’ diabetes (glycaemic) control, variables related to the symptoms of hypo/hyperglycaemia and HbA1c values were combined to indicate: fair-good control (i.e., no symptoms and HbA1c ≤8.0%); poor control (i.e., presence of symptoms and/or HbA1c >8.0%); and unknown (i.e., symptoms ‘none/don’t know’ and/or HbA1c values ‘none/don’t know’). Using this derived variable, most respondents were perceived to have either poor diabetes control (45.9%) or unknown diabetes control (42.3%).

Patients’ use of pharmacy services and their views on pharmacist roles

Responses describing patient usage of pharmacy services are summarised in Table 5. In addition to the traditional role of dispensing (‘treatment administration’), the most frequent services received were ‘patient education’ about medications, particularly directions for use (79.6%) and special precautions to follow (71.9%).

Table 5 also shows responses regarding pharmacists’ roles. All patients agreed with pharmacists’ roles in dispensing. Beyond dispensing, more than 70% of respondents expected pharmacists to provide ‘patient education’ about medications. About half of respondents supported other activities related to ‘patient education’ and ‘monitoring’. There were large gaps between patient usage and expectation regards education on: medication storage and adverse effects, SMBG, prevention of complications, need for regular monitoring, and foot care. In addition, patients expected pharmacists to provide more monitoring services than those currently provided.

In terms of the priority roles of pharmacists, patients’ responses can be seen in Table 6. The top five services perceived by patients as priorities (in addition to pharmacists’ traditional roles of dispensing being already provided) were from the ‘patient education’ domain—education related to medications (i.e., directions for use (64.5%), common/important adverse effects (25.5%), storage requirements (26.6%)) and the ‘monitoring’ domain—monitoring compliance with medications (37.3%). No new services were raised, beyond those already listed in the questionnaire, from this open-ended question, suggesting that from the respondents’ perspective the range of services that should be provided by community pharmacies described in Table 5 was complete.

Characteristics associated with patients’ views on pharmacist roles

Logistic regression models were used to identify patient characteristics which were associated with patients’ views on pharmacists’ roles. The odds ratios of significant characteristics are summarised in Table 7.

Patient experience (previous use) with a service was strongly associated with their views that the service should be provided by pharmacists (Odds Ratios (ORs) 4.4–11.3). Patients with poor/unknown glycaemic control or those who had risk factors for complications were more supportive of pharmacists providing some monitoring services (ORs 2.3–10.2). On the other hand, patients with higher incomes or those who were working were less supportive towards pharmacists providing some education and monitoring services (ORs 0.2–0.5).

Patients’ use of pharmacy services and their views on pharmacist roles

This study indicated little involvement of Indonesian community pharmacies in the care of patients with type 2 diabetes. Respondents mostly utilised pharmacists for their supply role (dispensing). A previous Indonesian study has confirmed the limited services provided to general pharmacy patients (Handayani & Raharni, Gitawati, 2009).

Amongst non-supply roles, many respondents chose roles closely related to dispensing as the priority roles of pharmacists, i.e., education related to medications, and monitoring compliance with medications. Similar findings were evident from some international studies involved diabetes patients (Abduelkarem et al., 2003a; Weitzman et al., 2009; Brown & Green, 2000; Hermansen-Kobulnicky & Worley, 2008; Twigg et al., 2013). Two qualitative studies indicated that patients identified the primary expertise of the community pharmacist as medicines supply, and there were mixed perceptions of community pharmacists’ roles extending to advising on prescription medicines, providing disease-related/health advice or providing monitoring services (using clinical testing devices) (Twigg et al., 2013; Gidman & Cowley, 2013).

It is interesting that the responses to the open-ended question that patients perceived that services related to dispensing and patient education were those that should be provided to type 2 diabetes patients from community pharmacies (Table 6). All of those they considered should be provided had been included in the options that could be provided (Table 5). This also indicated that patients did not have additional requirements for services not identified initially for this study.

Characteristics associated with patients’ views on pharmacist roles

The logistic regression models consistently found that a patient’s support for a service was influenced by their experience (previous use) of the service (Odds Ratios, ORs ≥4.4). Supporting this finding, studies worldwide have shown that type 2 diabetes patients have increased perceptions of pharmacists’ ability to assist them after receiving pharmacy-based services (Fera et al., 2008; Garrett & Martin, 2003; Hughes, 2006; Abduelkarem & Sackville, 2009; Hales et al., 2010). It should be emphasised that most patients in this study (at that time) received limited services from community pharmacies, thus they might not be aware of what pharmacists should and could do.

The implementation of Jaminan Kesehatan Nasional—JKN (National Health Coverage) in 2014 provides the best opportunity to optimise the use of Indonesian community pharmacies. It is important for the Government and IAI to establish an agreement on the basic services that should be available in community pharmacies. While the current payment under the scheme includes a very low prescription fee (Kementrian Kesehatan Republik Indonesia, 2014), the IAI should negotiate adequate remuneration for pharmacists to provide the services, thus enabling community pharmacies to remain viable.

In addition to patients’ past experiences, patients who were working and/or had higher incomes were generally less supportive of some of the proposed education or monitoring services (ORs ≤0.5). This might be because these groups of patients were likely to be younger (mean age 63.7 years for non-workers versus 54.0 years for workers, p < 0.0001; and mean age 60.4 years for income ≤Rp 5 million versus 58.6 years for income >Rp 5 million, p = 0.358). Two previous studies have reported that older patients, or those living with diabetes for a long time, were more supportive of pharmacists’ contributions (Twigg et al., 2013; Ibrahim, Al Tukmagi & Wayyes, 2013). It has been suggested that elderly people are one of the groups whose need for additional advice on medications and other related services has been demonstrated (Cartwright & Smith, 1988).

Notably, patients who had risk factors for complications and/or had poor/unknown glycaemic control were much more supportive of pharmacists monitoring treatment outcomes or adverse drug reactions (ORs ≥2.3). Such patients might reflect those with lower health status, representing a target group who might be more motivated and responsive to pharmacists’ involvement. It has been suggested that patients who benefit most from pharmacist-led education/coaching and disease state management services include those with poor glycaemic control and multiple comorbidities (Sisson & Kuhn, 2009). It is evident that overall treatment outcomes in this patient cohort are concerning and expanding the community pharmacy role into patient management needs to be investigated.

Limitations

The purposeful sampling method used in the study (20 patients from each of 10 pharmacies) was considered the only feasible manner by which it could be conducted. The request for approximately 20 patients was to ensure that each pharmacy included a range of patients in the sample and in many cases this about half of their current type 2 diabetes patients. No full list (sampling frame) of pharmacy patients with diabetes exists in Surabaya, so it was not possible to obtain a truly random sample of patients with this condition. Thus, there is a possibility of non-respondents not sharing the same practice and/or views of respondents, and some caution should be exercised in generalising the findings. However, the pharmacies covered a wide range of settings (geographic and socioeconomic), so that no particular background group of patients would be excluded; and achieved a sound response rate of approximately 60%. The characteristics of the respondents in this study were comparable to those of a population study involving all type 2 diabetes patients visiting 18 medical centres across Indonesia between November 2008 and February 2009 (N = 1,785) with respect to age (60 years versus 59 years, respectively), gender (female 59.2% versus 55.2%, respectively), and duration of diabetes (7 years versus 8 years, respectively) (Soewondo et al., 2010). Hence, although the sample is not randomly selected and the risk of response bias might limit the generalisation, the views of participants give some insight into the diabetes services used or desired at community pharmacies in Surabaya.