Abstract
Background
Visual analogue scale (VAS) is one of the simplest to measure medication adherence. It has neither been widely used for Non communicable diseases (NCD) nor validated for in the Indian setting. We examined the validity of self-reported medication adherence measures in relation to HbA1C in a rural population with diabetes mellitus (DM).
Methods
Participants with DM was administered VAS, Diabetes Self-Management Questionnaire (DMSQ) and assessed for missed pills. Descriptive statistics and logistic regression analysis were done.
Results
We recruited 1347 participants and 84% of them reported being 100% adherent as per VAS and 83.8% stated that they did not miss any pills. However, 58.2% of participants who reported having 100 % adherence had poor glycaemic control, as did 58.1% of those who did not miss any pills. None of the diabetic self-care measures was significantly associated with glycaemic control.
Conclusion
We found a lack of association between self-reported adherence measures and glycaemic control in participants with DM suggesting that self-reported adherence scales may not be valid in this population.
Keywords: medication adherence, self-reported adherence, diabetes, NCDs, rural India
Introduction
The prevalence of poor glycaemic control is high (76.6%) among diabetes mellitus(DM) patients in India [1]. Non-adherence to oral hypoglycaemic agents (OHAs) and insulin was 74 % in a rural South India study, with the most common reasons reported for missing medications being distance and lack of transport to a health facility, and forgetting to take medications during travel [2,3]. An adequate assessment of medication adherence in diabetes is thus crucial for researchers to accurately estimate the therapeutic efficacy of the medications under trial and for clinicians to optimize the treatment regimen. [4,5]
There are a few measures that have been found to be valid and useful measures of adherence in Low-and Middle-Income countries (LMIC). One of the simplest measures is the visual analogue scale (VAS), which asks participants to self-report the proportion of medication taken in the past month on a scale from 0–100. It has been found to have high levels of concordance with many other measures of anti-retroviral therapy (ART) adherence, including viral suppression in people living with HIV [6]. Since it’s a continuous measure, it allows more advanced analytic possibilities than data from categorical response sets or Likert scales [7]. However, the VAS has not been widely used for non-communicable diseases including diabetes. Our objective was to examine the validity of VAS in relation to levels of glycated haemoglobin (HbA1c).
Materials and methods
The cross-sectional analyses in this paper are based on the baseline data from a cluster randomized controlled trial in 49 primary health centres in rural southern India evaluating a collaborative care intervention to improve mental health outcomes in participants with co-morbid physical conditions [8]. Details about the trial including the original sample size calculation, is reported elsewhere [8]. Participants included in the present medication adherence analyses were on oral hypoglycaemic agents DM and at least 30 years old. [8]
The baseline data collection was done between May 2015 and November 2018. Blood glucose levels were measured via HbA1C and classified as good control (<= 8.0) and poor control (> 8.0) which is based on a Guidance Statement Update From the American College of Physicians [9].
Medication adherence was measured using both the VAS and self-reported missed pills in the last month. Based on the distribution of the responses, we dichotomized VAS and missed pills. In addition, we used nine of the 16 items of the Diabetes Self-Management Questionnaire (DSMQ), covering three subscales - Glucose Management, Dietary Control, and Physical Activity [10]. We tabulated glycaemic control by self-reported adherence measures (VAS & missed pills). In the logistic regression model, glycaemic control was the dependent variable and independent variables were VAS, self-reported missed pills, DSMQ scores and socio-demographic characteristics.
Results
Of the 2486 cohort participants, 1492 (60%) were under treatment for diabetes at baseline, of which 1347 (90.3%) were on oral hypoglycaemic agents and were included in the present analyses. A majority (56%) of participants did not have any formal education and more than two-thirds reported earning less than INR 5,000 (68.7%). (Table 1)
Table 1.
Sociodemographic Characteristics of the study population (n= 1347)
| Variable | Category | Frequency | Percentage |
|---|---|---|---|
| Gender | Male | 371 | 27.5 |
| Female | 976 | 72.5 | |
| Marital status | Married | 886 | 65.8 |
| Widowed | 431 | 32.0 | |
| Separated or Divorced | 17 | 1.3 | |
| Never married | 13 | 0.9 | |
| Age in years | 30–44 | 96 | 7.1 |
| 45–54 | 274 | 20.3 | |
| 55–64 | 491 | 36.5 | |
| 65–74 | 409 | 30.4 | |
| >75 | 77 | 5.7 | |
| Religion | Hindu | 1326 | 98.4 |
| Others | 20 | 1.5 | |
| Missing | 1 | 0.1 | |
| Monthly income (INR)* | <5000 (~ USD 60) | 926 | 68.7 |
| 5001–10000 (~ USD 61–121) | 321 | 23.8 | |
| >10000 (> USD 121) | 100 | 7.4 | |
| Education | No formal education | 754 | 56.0 |
| Primary (1–7 years) | 403 | 29.9 | |
| Secondary or higher | 189 | 14.0 | |
| Missing | 1 | 0.1 | |
| Body mass index (BMI) | Underweight (<18.5) | 43 | 3.2 |
| Normal (18.5–22.9) | 433 | 32.1 | |
| Overweight (23–24.9) | 287 | 21.3 | |
| Obese (>=25) | 579 | 43.0 | |
| Missing | 5 | 0.4 |
Approximately <13 USD/month qualifies the family to be poor in rural areas.
More than half of the participants had poor glycaemic control (56.5%, 785/1388) although the majority (84 %, 1125/1338) reported 100% adherence per the VAS. Similarly 83.8%,1119/1334 stated that they did not miss any pills in the previous month. Despite the high levels of self-reported adherence 58.2% of participants who reported having 100% adherence according to the VAS had poor glycaemic control, as did 58.1% of those who self-reported noting missing any pills in the past month. Self-reported adherence was not significantly associated with glycaemic control in univariate analysis (p > 0.05) (Table 2).
Table 2.
Number and percent with glycaemic control by adherence (Univariate analysis)
| Measurement | Adherence | Glycaemic control | Total | P value* | |
|---|---|---|---|---|---|
| Poor (HbAlC >8) | Good (HbA1C <=8) | ||||
| Self-report per VAS | < 100% adherence | 121 (56.8%) | 92 (43.2%) | 213 | 0.701 |
| 100% adherence | 655 (58.2%) | 470 (41.8%) | 1125 | ||
| Self-reported missed pill in the previous month | No missed pills | 123 (57.2%) | 92 (42.8%) | 215 | 0.811 |
| Missed pills | 650 (58.1%) | 469 (41.9%) | 1119 | ||
for Pearson Chi-square test (1 degree of freedom)
The mean score (out of 10) of diabetic self-care subscales for medication, diet and physical activity were 9.0 (SD-1.7), 7.9 (SD-1.8), and 6.9 (SD-2.4) respectively. Logistic regression analyses (Table 3) showed that none of these three measures was significantly associated with glycaemic control. Similarly, VAS-based medication adherence lacked a significant association with glycaemic control.
Table 3.
Multivariate logistic regression analysis of glycaemic control on medication adherence and diabetes self-care
| Predictor | Unadjusted Odds Ratio | Cluster robust SE | 95% Confidence Interval | p-value | Adjusted Odds Ratio (n=1332) | Cluster robust SE | 95% Confidence Interval | p-value |
|---|---|---|---|---|---|---|---|---|
| Adherence related variables: | ||||||||
| VAS 100% adherence* | 0.94 | 0.14 | 0.70 – 1.27 | 0.700 | 0.80 | 0.17 | 0.53 – 1.20 | 0.281 |
| Diabetes Self Care, Medication subscale |
1.04 | 0.03 | 0.98 – 1.10 | 0.161 | 1.06 | 0.04 | 0.98 – 1.15 | 0.148 |
| Diabetes Self Care, Diet subscale |
1.00 | 0.03 | 0.94 – 1.07 | 0.954 | 0.99 | 0.03 | 0.92 – 1.06 | 0.744 |
| Diabetes Self Care, Activity subscale |
1.02 | 0.03 | 0.97 – 1.07 | 0.519 | 1.01 | 0.03 | 0.96 – 1.07 | 0.702 |
| Sociodemographics: | ||||||||
| Age * | 0.140 | 0.158 | ||||||
| 45–54 | 1.16 | 0.29 | 0.72 – 1.88 | 1.19 | 0.30 | 0.72 – 1.94 | ||
| 55–64 | 1.15 | 0.27 | 0.73 – 1.82 | 1.21 | 0.30 | 0.75 – 1.96 | ||
| 65–74 | 1.33 | 0.33 | 0.81 – 2.16 | 1.40 | 0.38 | 0.83 – 2.38 | ||
| >=75 | 2.17 | 0.67 | 1.18 – 3.99 | 2.31 | 0.78 | 1.20 – 4.47 | ||
| Female* | 1.02 | 0.12 | 0.80 – 1.29 | 0.882 | 1.10 | 0.17 | 0.81 – 1.48 | 0.544 |
| Married* | 0.86 | 0.09 | 0.70 – 1.06 | 0.158 | 0.91 | 0.12 | 0.71 – 1.17 | 0.475 |
| Working* | 0.91 | 0.09 | 0.75 – 1.12 | 0.387 | 0.99 | 0.12 | 0.78 – 1.25 | 0.916 |
| Education* | 0.175 | 0.072 | ||||||
| Primary | 1.16 | 0.10 | 0.99 – 1.36 | 1.22 | 0.11 | 1.01 – 1.46 | ||
| Secondary or higher | 1.11 | 0.17 | 0.82 – 1.50 | 1.25 | 0.21 | 0.90 – 1.72 |
Reference: VAS- <100% adherence ; Age −30 to 44 years ; Gender-male; Marital status – others ; Occupation- not working; Education- no formal education
Discussion
The results show a significant discrepancy between self-reported medication adherence and glycaemic control in this study population, with more than half of the participants reporting 100% adherence, even though only 41.8% of them had good glycaemic control. While it is possible that other factors, such as diet and exercise, contributed to their lack of glycaemic control, it appears more likely that they overestimated their adherence [11]. Consistent with our findings, this phenomenon has also been observed when self-reported adherence measures have been used in other medical regimens, including HIV [12–15]. Similarly, several studies have observed inaccuracies in self-reported HIV risk and treatment adherence behaviours as well as medication adherence in non-communicable diseases such as HTN and patients on anticoagulant therapy. The authors hypothesize that self-reported adherence scales likely lead to overestimates, due to social desirability bias [16–19]. Though subjective measures are inexpensive; easy to administer; identify beliefs and barriers; and are flexible to accommodate in different settings, it has also been established that they have relatively lower sensitivity and specificity, and are affected by the communication skills of the interviewer as well as patient’s desirability bias. [20]
Our work thus demonstrates that self-report of medication adherence in this rural Indian setting has limitations despite using standardized measures and conducting extensive formative work prior to data collection, including cognitive interviews to ensure comprehension and semantic equivalence. Participants were also assured that their responses would not be communicated to their healthcare providers. Our findings suggest that biological and objective measures in these settings are more likely to be valid. While objective measures such as pill counts, electronic monitoring, secondary database analysis and biochemical measures overcome the challenges of subjective measures, they are expensive, sometimes invasive and often time consuming [20]. Doggrell SA and Warot S conclude that the most of studies demonstrate a relationship between HbA1c levels and adherence to anti-diabetes medications, and this relationship appears to be irrespective of the method employed to measure adherence [21]. However, a range of HbA1c readings is required to demonstrate a relationship. In low-income populations, the association is not always evident and these findings are in concordance with ours [21]. Similarly, a self-reported tool in the Mexican population has been found to be valid and reliable but has not been proven to have a positive impact on glucose levels [22].
The major strength of our study was that it was population-based with a large cohort. We used validated tools and trained field staff for data collection and used a biological measure (HbA1C) to glycaemic control. There are also a few limitations of our study. Since HbA1C can be influenced by several other factors such as lifestyle, diet, and stress, we tried to measure some of the lifestyle factors through DSMQ to examine whether they accounted for the poor glycaemic control. However, it appears that these measures too may have been influenced by social desirability biases, so we are unable to rule them out as a contributing factor. This study calls for developing simple, inexpensive objective measures that can be used in resource-limited settings.
Conclusion
Self-reported adherence scales may not be valid in the study population. Future research should focus on developing simple, inexpensive objective measures that can be used in resource-limited settings to accurately estimate medication adherence.
Highlights.
Visual analogue scale (VAS) is a simple tool to measure medication adherence
VAS has not been widely used or validated in Indian setting
Self-reported adherence scales may not be reliable in a rural population
Need for innovative, easy-to-administer measures that can be used in rural settings
Acknowledgement
We would like to thank all the study staff of the HOPE trial as well as the participants who so generously shared their thoughts on some very sensitive topics.
Funding
This work was supported by the US National Institute of Mental Health [R01MH100311]. The funding body did not have any role in the design of the study and collection, analysis, and interpretation of data and in writing the manuscript.
Footnotes
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Ethics statement
Ethics approval was obtained from the Institutional Ethical Review Board at St. John’s Medical College and Hospital (reference 38/2013) and the University of California, San Francisco. All the participants were recruited for the study after obtaining informed consent (reference 12–10470).
Conflict of interest statement
None of the authors declare any conflict of interest.
Availability of data and materials
The datasets used and/or analysed during the current study are available from the corresponding author upon reasonable request.
References
- 1.Borgharkar SS, Das SS. Real-world evidence of glycemic control among patients with type 2 diabetes mellitus in India: the TIGHT study BMJ Open Diabetes Research and Care 2019;7:e000654. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Venkatesan M, Dongre AR, Ganapathy K. A community-based study on diabetes medication non-adherence and its risk factors in rural Tamil Nadu. Indian J Community Med 2018;43:72–6 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Sankar UV, Lipska K, Mini GK, et al. The adherence to medications in diabetic patients in rural Kerala, India. Asia Pac J Public Health. 2015. Mar;27(2):NP513–23. [DOI] [PubMed] [Google Scholar]
- 4.Bosworth H Improving patient treatment adherence: A clinician’s guide. New York: Springer; 2010. [Google Scholar]
- 5.Farmer KC. Methods for measuring and monitoring medication regimen adherence in clinical trials and clinical practice. Clin Ther 1999;21:1074–90. [DOI] [PubMed] [Google Scholar]
- 6.Ekstrand ML, Shet A, Chandy S, et al. Suboptimal adherence associated with virological failure and resistance mutations to first-line highly active antiretroviral therapy (HAART) in Bangalore, India. Int Health 2011;3(1):27–34. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Finitsis DJ, Pellowski JA, Huedo-Medina TB,et al. Visual analogue scale (VAS) measurement of antiretroviral adherence in people living with HIV (PLWH): a meta-analysis. J Behav Med. 2016. Dec;39(6):1043–1055 [DOI] [PubMed] [Google Scholar]
- 8.Srinivasan K, Mazur A, Prem KM,et al. Improving mental health through integration with primary care in rural Karnataka: study protocol of a cluster randomized control trial. BMC Fam Pract. 2018;19:158. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Qaseem A, Wilt TJ, Kansagara D, et al. Hemoglobin A1c Targets for Glycemic Control With Pharmacologic Therapy for Nonpregnant Adults With Type 2 Diabetes Mellitus: A Guidance Statement Update From the American College of Physicians. Ann Intern Med. 2018;168(8):569–576 [DOI] [PubMed] [Google Scholar]
- 10.Schmitt A, Gahr A, Hermanns N,et al. The Diabetes Self-Management Questionnaire (DSMQ): development and evaluation of an instrument to assess diabetes self-care activities associated with glycemic control. Health Qual Life Outcomes. 2013. Aug 13;11:138. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Parajuli J, Saleh F, Thapa N,et al. Factors associated with non adherence to diet and physical activity among Nepalese type 2 diabetes patients; a cross sectional study. BMC Res Notes 2014;7:758. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Simoni JM, Frick PA, Lockhart D,et al. Mediators of social support and antiretroviral adherence among an indigent population in New York City. AIDS Patient Care STDS 2002;16:431–9. [DOI] [PubMed] [Google Scholar]
- 13.Hewett PC, Mensch BS, Ribeiro MC, et al. Using sexually transmitted infection biomarkers to validate reporting of sexual behavior within a randomized, experimental evaluation of interviewing methods. Am J Epidemiol. 2008;168(2):202–11. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Mensch BS, Hewett PC, Abbott S, et al. Assessing the reporting of adherence and sexual activity in a simulated microbicide trial in South Africa: an interview mode experiment using a placebo gel. AIDS Behav. 2011;15(2):407–21. [DOI] [PubMed] [Google Scholar]
- 15.Gnambs T, Kaspar K. Disclosure of sensitive behaviors across self-administered survey modes: a meta-analysis. Behav Res Methods. 2015;47(4):1237–59. [DOI] [PubMed] [Google Scholar]
- 16.Tedla YG, Bautista LE. Factors associated with false-positive self-reported adherence to antihypertensive drugs. J Hum Hypertens. 2017. May;31(5):320–326. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Mondal H, Mondal S. Social desirability bias: A confounding factor to consider in survey by self-administered questionnaire. Indian J Pharmacol 2018;50:143–4. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Mehta K, Ekstrand ML, Heylen E,et al. Adherence to Antiretroviral Therapy Among Children Living with HIV in South India. AIDS Behav. 2016. May;20(5):1076–83 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Pai-En CH, Hsuan-Ming TS, Chuan-Hsiu TS. Discrepancy among Self-Reported Adherence, Prescription Refills, and Actual Anticoagulant Control. Journal of Nursing Research. 2020. Feb 1;28(1):e63. [DOI] [PubMed] [Google Scholar]
- 20.Lam WY, Fresco P. Medication Adherence Measures: An Overview. Biomed Res Int. 2015;2015:217047. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Doggrell SA, Warot S. The association between the measurement of adherence to anti-diabetes medicine and the HbA1c. Int J Clin Pharm. 2014;36(3):488–497. [DOI] [PubMed] [Google Scholar]
- 22.Muñoz-López DB, Reyes Pérez V, Garay-Sevilla ME, Preciado-Puga MDC. Validation of an instrument to measure adherence to type 2 diabetes management. Int J Clin Pharm. 2021;43(3):595–603. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Data Availability Statement
The datasets used and/or analysed during the current study are available from the corresponding author upon reasonable request.