Background
Diabetes Mellitus is among the most prevalent chronic illnesses in the United States, affecting nearly 24 million Americans [
1]. In response to the Institute of Medicine’s calls for patient-centeredness [
2], innovations in diabetes care have increasingly made patients’ perspectives central to the process and outcomes of care. These advances, which include the Chronic Care Model [
3], the Patient-Centered Medical Home [
4], and various patient-engagement interventions [
5,
6], all focus on patient-centeredness in the process of care. However, there is a need to move beyond the process of care and develop patient-centered outcomes to assess the impact of these innovations from the patient perspective.
As with many chronic diseases, diabetes patients are less concerned with clinical biomarkers [
7] such as hemoglobin A
1c, blood pressure, or lipid levels, and are more concerned with physical and social function, emotional and mental health, and the burden of illness and treatments on daily life [
8]. Quality of life measures, which include many of these domains [
9] are thus more meaningful and relevant outcomes from the patient perspective. The development of quality of life measures that are associated with future clinical outcomes would enhance shared decision making by framing treatment options in a context that is pertinent to patients [
10].
In diabetes care, general health status measures such as the SF-36 and the EQ-5D are commonly used to assess patients’ quality of life [
11‐
18]. Although these measures are useful in comparing patient health status across different illnesses, they often cannot capture distinctive aspects of specific diseases [
19]. Quality of life measures that are disease-specific and associated with clinical outcomes have been developed in other chronic illnesses. For instance, a number of disease-specific quality of life measures in cardiovascular disease [
10,
20‐
22] and cancer [
23‐
28] are predictive of subsequent morbidity and mortality. Diabetes places significant self-management responsibility on patients, and thus warrants the development and validation of clinically relevant and patient-centered quality of life measures. Recent structured reviews [
9,
29,
30] have identified several disease-specific quality of life measures for diabetes. Unfortunately, attempts to understand the association between these quality of life measures and common clinical biomarkers, such as HbA
1c, have been inadequate [
9,
29,
30].
We conducted a retrospective cohort study nested within a randomized comparative effectiveness trial of diabetes self-management interventions to investigate the association of HbA1c and diabetes-specific quality of life. We evaluated the relationship between diabetes-specific quality of life and HbA1c both before and after participants completed diabetes self-management programs.
Discussion
We constructed a retrospective cohort of participants drawn from a randomized comparative effectiveness study to evaluate the relationship between change in HbA1c and Diabetes-39 quality of life. HbA1c at one-year follow-up was significantly associated with overall quality of life on the Diabetes-39. Our multiple linear regression models suggest that improvements in HbA1c among patients completing diabetes self-management interventions are significantly associated with increased quality of life on the diabetes control and sexual functioning subscales of the Diabetes-39. No association was established between changes in HbA1c and the anxiety and worry, social burden, and energy and mobility subscales. Baseline burden of illness, a proxy for baseline quality of life, predicted overall quality of life as well as all subscales of the Diabetes-39, as expected.
This study firmly establishes the relationship between improved HbA
1c, a critical clinical biomarker in diabetes, and the Diabetes-39, a patient-centered diabetes-specific quality of life measure among patients completing a self-management education program. Several previous studies have attempted to explore the relationship between clinical indicators, such as HbA
1c, and a variety of diabetes-specific quality of life measures [
31,
34‐
40]. Unfortunately, these associations have been weak [
41] or nonexistent [
42], present for only very few of a scale’s domains [
36], or are specific to type 1 diabetes only [
34,
35]. Further, prior studies report on measures that have poor evidence for validity and reliability [
32,
35,
36,
41], focus on singular aspects of quality of life (e.g., distress [
37,
38,
41]), ignore key components of quality of life such as physical and social functioning [
9], or include several items that are not diabetes-specific [
9]. Additionally, several reviews of diabetes-specific quality of life measures [
9,
29,
30] have recognized the lack of empirical evidence on the responsiveness of these scales to changes in health status.
This analysis of HbA
1c and diabetes-specific quality of life addresses many of the limitations of prior studies. The Diabetes-39 diabetes-specific quality of life measure has been recommended for use in research and clinical settings by all of the aforementioned reviews of diabetes-specific quality of life measures [
9,
29,
30]. The instrument has good evidence for validity and reliability, includes several domains that cover many aspects of quality of life, and is applicable to a wide population of patients [
9,
29,
30,
33]. The Diabetes-39 is one of few diabetes-specific quality of life measures that have been shown to be responsive to changes in health status [
39]. Further, this instrument does not impose a definition of quality of life upon respondents, but instead allows patients to frame responses in the context of their own personal conceptualization of quality of life. Also, patients were directly involved in the selection of items for the questionnaire [
33]. These attributes make the instrument highly patient-centered, one of the most critical components to any patient-assessed quality of life measure. Thus, our study focuses on a diabetes-specific quality of life measure that is a prime candidate for analysis.
Our statistical methods also address several prior studies’ shortcomings. While most previous attempts to examine the relationship between HbA
1c and quality of life used simple linear correlations [
34,
41,
42], our analyses included predictive linear regression models. This allows for a more robust analysis and provides a quantification of the impact of HbA
1c on quality of life. To our knowledge, two prior studies have employed linear regression models to assess this relationship [
37,
38]. However, one study [
38] grouped continuous HbA
1c data into two groups. This reduces a model’s ability to quantify the effect of changes in HbA
1c on quality of life, and diminishes the overall robustness of the model. A second study [
37] modeled HbA
1c as the primary dependent variable. This is not in line with the Institute of Medicine’s vision [
2] in which patient-centered measures, such as quality of life, are the ultimate outcomes of care. Our analysis included a regression of continuous HbA
1c data with quality of life as the primary outcome.
Few prior studies have examined the relationship between clinical indicators and diabetes-specific quality of life measures among participants who all completed diabetes self-management programs. These programs were deeply embedded in primary care. One program was led by a primary care physician, while the other was led by nurse educators and registered dieticians. The latter model represents the type of delivery system redesign that is characteristic to many primary care innovations [
3,
4]. Our examination of the relationship between clinical indicators and quality of life outcomes in the context of patient-centered diabetes self-management programs demonstrates that HbA
1c improvements among participants in these programs are associated with better quality of life. Previous studies have included diabetes-specific quality of life among outcome measures [
40,
43]. These studies approach both quality of life and HbA
1c as distinct outcomes, and do not explore the association between the two variables. Unlike prior studies, our study examines the relationship between changes in HbA
1c and diabetes-specific quality of life. In the post-ACCORD era, there has been reduced emphasis on intensive HbA
1c control [
44]. However, the current study suggests that improved HbA
1c resulting from diabetes self-management interventions is associated with better diabetes-specific quality of life. Thus, HbA
1c control is relevant to patient-centered outcomes and should remain a valuable goal in diabetes care.
There were limitations to our study. A sample size of 75 limited the range of analytic strategies that could be employed. The sample size may also have affected the power of our analyses, which may account for the weak association between changes in HbA
1c and some of the Diabetes-39 subscales. The generalizability of our study may also be limited. Our sample is reflective of the United States Veterans Administration patient population, consisting largely of older patients who are predominantly male, of older age, and have significant co-morbidities. Further, all of the participants in our cohort participated in at least one diabetes self-management program. Thus, we were unable to assess the impact of participation in these programs on quality of life as compared with patients who did not participate in any self-management programs. Additionally, the lack of Diabetes-39 data at baseline precluded an examination of the responsiveness of this diabetes-specific quality of life measure over time. However, our analysis does include HbA
1c data from multiple time points and includes a measure of burden of illness at baseline. Many previous studies used cross-sectional data from one time point [
34,
37,
41,
42]. Our analyses included HbA
1c data from both before and after participation in diabetes self-management programs.
Future studies should be certain to collect quality of life data both before and after diabetes self-management programs so that the responsiveness of quality of life measures can be assessed. Subsequent studies should also include larger, more diverse samples to ensure adequate power and generalizability. The inclusion of a control group that does not receive any programs beyond routine care may also allow for future examinations of the impact of diabetes-self management programs on quality of life.
AK is an MD/MPH candidate at Baylor College of Medicine and the University of Texas School of Public Health. AN is the an investigator in the Health Decision-Making and Communication Program, Houston VA Health Services Research and Development CoE, Michael E. DeBakey VA Medical Center. AN is also an Assistant Professor, Department of Medicine, Section of Health Services Research, Baylor College of Medicine and Adjunct Assistant Professor, Division of Health Promotion and Behavioral Sciences, School of Public Health, The University of Texas Health Sciences Center at Houston. AB is a Programmer in the Design and Analysis Program, Houston VA Health Services Research and Development CoE, Michael E. DeBakey VA Medical Center and an Instructor in the Department of Medicine, Section of Health Services Research, Baylor College of Medicine. JS is a Professor of Health Economics, at the University Texas School of Public Health and at the University of Texas Medical School, Center for Clinical Research Evidence-Based Medicine, and Adjunct Professor, Department of Economics, Rice University. RS is a Research Professor in Medicine at Texas A&M University and Director, Health Communication and Decision-Making Program in the Houston Center for Quality of Care and Utilization Studies, Baylor College of Medicine. MP is the Associate Director of Evaluation for the University of Texas Prevention Research Center and an Assistant Professor at The University of Texas School of Public Health.
Competing interests
The authors declare that they have no competing interests.
Authors’ contributions
AK and AN conceived of the study and participated in its design. AK drafted the manuscript. AB conducted statistical analysis and helped draft the manuscript. AN participated in study conception and design, helped draft the manuscript, and conducted the parent study from which data for this study was derived. JS, RS, and MP contributed to interpretation of data, revising manuscript critically, and final approval of the version submitted to the journal. All authors read and approved the final manuscript