Treatment adherence to oral glucose-lowering agents in people with diabetes: Using the Brief Medication Questionnaire

Adherence to diabetes treatment is a complex process that involves lifestyle changes such as meal planning, physical exercise, glycaemic monitoring and taking prescribed medication. The most commonly used instruments to measure medication adherence in the literature are the Morisky–Green Test (Morisky et al, 1986), the Brief Medication Questionnaire (BMQ; Svarstad et al, 1999) and the Measure of Adherence to Treatment test (Delgado and Lima, 2001).

The current study aimed to assess medication adherence to oral glucose-lowering agents among individuals with diabetes through the use of the BMQ, which assesses drug regimens, beliefs and patient recall.

Methods
Participants
Participants were recruited from the ATEMDIMEL (Apoio Telefônico para o Monitoramento em Diabetes Mellitus [Telephone Support for Monitoring Diabetes]) programme in a primary care service facility at a public hospital in the state of São Paulo, Brazil. Data were collected from May to October 2013. ATEMDIMEL participants of both genders, aged ≥18 years and taking oral glucose-lowering agents were eligible for recruitment. Participants were selected by convenience, through invitation by phone call.

Two instruments were used for data collection during interviews: one for sociodemographic and clinical data and the second, the BMQ. Participants were also assessed in terms of their weight, height, blood pressure and clinical laboratory data. The study was approved by the Human Subject Research Ethics Committee of the Ribeirão Preto College of Nursing (University of São Paulo).

Brief Medication Questionnaire
The BMQ is composed of three sections that identify self-reported barriers to adherence with respect to treatment regimens, patient beliefs and patient recall (ability to remember to take medication). In the original study, the regimen section showed 80% sensitivity and 100% specificity to predict repeated non-adherence (Svarstad et al, 1999). The instrument’s total score consists of the sum of the scores of its three sections. Participants receive a score of 1 if their response indicates potential non-adherence and 0 if it indicates adherence.

The score for the regimen section ranges from 0 to 7, and the belief and recall sections both range from 0 to 2. Thus, the maximum total BMQ score is 11 and any score greater than zero for any one of the sections indicates potential non-adherence to prescribed treatment. The BMQ classifies participants into the following categories: adherence (no positive response in any domain), probable adherence (positive response in one domain), probable low adherence (positive responses in two domains) and low adherence (positive responses in all three domains).

Data analysis
The data were analysed through descriptive statistics, which allowed us to characterise the sample regarding the investigated variables. In order to verify associations between qualitative variables, the data were submitted to Fisher’s exact test. Furthermore, any associations were measured using logistic regression models (Hosmer and Lemeshow, 1989) and the raw odds ratios were calculated and adjusted with respective confidence intervals of 95%. All statistical analyses were conducted with the help of SAS^® 9.0 software (SAS Institute, Cary, NC, USA). P-values lower than 0.05 were considered significant.

Results
Sociodemographic and clinical data
The majority of participants were women, aged <65 years, married, and with an average of 6.7 years of schooling. Participants aged ≥65 years had a 2.8 times greater chance of adhering to the proposed medication therapy (defined as achieving a BMQ rating of adherence or probable adherence; Table 1). The most common duration of diabetes was 11–20 years, and hypertension was the most common disease associated with diabetes; overall, 40% of the sample was classified as having stage 1 hypertension (Table 2). Participants were eight times more likely to adhere to oral glucose-lowering agents in the first 5 years of treatment and those with a normal waist circumference (<102 cm in men and <88 cm in women; World Health Organization [WHO], 2008) were twice as likely to be adherent as those with a greater waist circumference (Table 2).

BMQ scores
Total BMQ scores showed that 25.0% of participants had complete adherence and 21.7% had probable adherence to oral glucose-lowering therapy. Regarding the three individual domains, the regimen domain indicated an 81.7% rate of adherence and a 5.0% rate of probable adherence, the belief domain indicated 55% adherence and 25% probable adherence, and the recall domain indicated 35.0% adherence and 33.3% probable adherence (Table 3).

Association of adherence scores with HbA_1c
Regarding HbA_1c levels and BMQ adherence scores, our investigation revealed one participant whose scores for the regimen and belief domains classified him as adherent but who had an HbA_1c of 139 mmol/mol (14.9%). The recall domain presented the most conflicting data, indicating adherence in one participant who had the study’s highest HbA_1c level: 145 mmol/mol (15.4%).

Analysis of total BMQ ratings showed that the median HbA_1c for the combined domains was ≥65 mmol/mol (8.1%), regardless of adherence score. Median values ranged from 65 mmol/mol (8.1%) in adherent participants to 75 mmol/mol (9.0%) in low-adherence participants, and people with higher adherence scores had lower mean HbA_1c values than those with lower scores (Table 4).

Figure 1 illustrates HbA_1c values in participants who demonstrated medication adherence and non-adherence.

Discussion
The purpose of oral glucose-lowering agents is to reduce the risk of acute and chronic complications of diabetes, thus improving quality of life for people with this condition and their families. Barriers to adherence are often observed in chronic conditions such as diabetes, which require lengthy and complex treatments and lifestyle changes (Seley and Weinger, 2007).

The present study adopted the WHO definition of adherence, characterised as the extent to which a person’s behaviour corresponds to recommendations from healthcare professionals regarding a proposed treatment. In other words, adherence is a person’s ability to put recommendations made by healthcare professionals into practice (WHO, 2003).

The innovations presented by the BMQ for measuring the use of oral glucose-lowering agents makes it difficult to compare our results with those of other studies. Other authors measured medication adherence to glucose-lowering agents by using different instruments, such as the Measure of Adherence to Treatment and the Morisky–Green tests, and found values higher than those in the current study (Groff et al, 2011; Carvalho et al, 2012). Comparisons between the BMQ and these instruments showed that the BMQ is a more complete instrument for analysing the complex subject of adherence, as it is divided into three domains that indicate difficulties regarding regimens, beliefs and memory. On the other hand, we must mention the limitations of using a self-reporting instrument in which participants can manipulate their responses.

Results regarding glycaemic control and its association with adherence to oral glucose-lowering agents revealed a median HbA_1c of ≥65 mmol/mol (8.1%) across all three domains. Other studies investigating medication adherence to diabetes drugs have shown characteristics similar to these (Gimenes et al, 2009; Faria et al, 2013).

The belief domain revealed an association between adherence score and the number of medications taken daily, and between adherence score and the educational process conducted by health professionals before and during treatment. This finding can be explained by the difficulty for people aged ≥65 years who are taking a great number of medications every day, and also by the effects of the professional–patient relationship on treatment adherence (WHO, 2011). Another significant factor in treatment adherence is the level of self-efficacy presented by people regarding their ability to follow through with the proposed treatment (Gherman et al, 2011).

Another representative piece of data was that one participant who was considered adherent according to the score in the recall domain presented with an HbA_1c of 145 mmol/mol (15.4%). This finding suggests that medication recall and awareness of the importance of taking oral glucose-lowering drugs do not guarantee that people will adhere to medication or the other pillars of metabolic control: physical activity and diet. It is also consistent with the idea that non-adherence does not cause short-term harmful effects, even if it increases the probability of late and untreatable future complications (Loke et al, 2012).

Research limitation
As with all research involving self-report questionnaires, these findings are limited by the possibility that participants can easily manipulate the answers given and, consequently, transmit a false impression of medication adherence.

Conclusion
Owing to the high prevalence of diabetes and the complexity of treating such chronic health conditions, it is essential that adherence be studied in order to improve the quality of healthcare and reduce the complications of the disease. Although we did not find any statistically significant associations between sociodemographic and clinical variables and medication adherence to oral glucose-lowering agents among these people, some factors came to our attention. For example, the probability of individuals adhering to the use of medication was eight times higher in those who were in their first 5 years of treatment. As most of the population in our study had been diagnosed with diabetes for 11–20 years, this suggests that education strategies must be used to avoid deterioration in adherence during the course of the disease.

We conclude that the BMQ can serve as a support tool in intervention studies as, in addition to measuring adherence, it also investigates barriers to this process that can be overcome with the help of health education.