OBJECTIVE—To investigate whether self-rated health profiles compiled using the EuroQol group’s visual analog scale (EQ VAS) are independent predictors of vascular events and major complications in people with type 2 diabetes after controlling for standard clinical risk factors.
RESEARCH DESIGN AND METHODS—The study is based on 7,348 individuals with a mean follow-up of 2.4 years after completing the EQ-5D questionnaire. We used Cox proportional hazards modeling to estimate hazard ratios associated with EQ VAS scores after controlling for baseline covariates: age, sex, smoking status, diabetes duration, A1C, systolic blood pressure, BMI, plasma lipids, and prior clinical history.
RESULTS—A 10-point higher EQ VAS score was associated with a 6% (95% CI 1–11) lower risk of vascular events and a 22% (95% CI 15–28) lower risk of diabetesc complications.
CONCLUSIONS—Self-rated health profiles compiled using the EQ VAS provide valuable information on patient risk in addition to that determined from clinical risk factors alone.
Several risk models are available for people with diabetes (1–3) based on clinical risk factors such as systolic blood pressure and A1C. These provide a means of identifying high-risk patients to target the management of diabetes care (4) and can be used for economic evaluation of diabetes therapies (5). To date, little attention has been given to the role of subjective measures of health in risk assessment; few studies have considered outcomes other than mortality that are particularly relevant to people with diabetes, whose risk of cardiovascular and other complications is elevated (6).
Studies on self-rated health have consistently shown that people who report their health status as “poor” or “fair” have higher mortality than those reporting their health as “excellent” or “good” (7). Self-rated health may also be measured with a visual analog scale (VAS), e.g., the EuroQol (EQ) VAS from the EQ-5D (8), a thermometer-like scale with zero representing the worst and 100 the best imaginable health state. In this study, we examined whether self-rated health measured by the EQ VAS is an independent predictor of cardiovascular events and major diabetes complications in individuals with type 2 diabetes.
RESEARCH DESIGN AND METHODS—
The study is based on Australian and New Zealand participants of the Fenofibrate Intervention and Event Lowering in Diabetes (FIELD) study (9) who participated in a quality-of-life substudy. We focused on two end points: cardiovascular events and other diabetes-related complications (see Table 1 and the online appendix, which is available at http://dx.doi.org/10.2337/dc07-1391).
Baseline values of clinical risk factors were calculated as an average of all available measures between trial randomization and first administration of the EQ-5D questionnaire. Mean values from the whole sample were imputed for patients (<1% of the sample) with any missing baseline risk factors.
Log-rank methods without adjustment for covariates were used to test for significant differences in risk among patients categorized into approximate quartiles according to their EQ VAS scores.
We used Cox proportional hazards modeling to estimate hazard ratios (HRs) associated with EQ VAS scores after controlling for baseline covariates of age, sex, smoking status, diabetes duration, A1C, systolic blood pressure, BMI, plasma lipids, and prior clinical history. We investigated any heterogeneity in the HR of the VAS by an interaction term with prior events. Risk factors were dropped from the models through stepwise elimination if their HRs were not significantly different from 1.0 (P > 0.1). The proportional hazards assumption was tested using Schoenfeld residuals (10). Stata/SE (version 9.1; Stata, College Station, TX) was used for all statistical analyses.
RESULTS—
A total of 7,348 patients (87.5% of all Australian and New Zealand patients in the FIELD trial) consented and correctly answered a written EQ-5D questionnaire at an average of 2.9 years after trial randomization, generating a mean follow-up of 2.4 years. Mean ± SD patient age was 66 ± 6.9 years; 91.8% were white and 38% women. Mean EQ VAS score was 77.8 ± 15.9; other baseline risk factors were similar to those reported in the FIELD trial (11). The cumulative hazard of vascular events and complications was stratified by patients’ baseline EQ VAS score (P < 0.001), with differences most pronounced for patients scoring below 70 (see online appendix).
After adjustment for other risk factors, EQ VAS scores remained significant independent predictors of cardiovascular events, with a 10-point increase in EQ VAS score associated with a 6% (95% CI 1–11) decrease in risk of a future event (Table 1). Patients with a history of any prior event were three times more likely to experience a vascular event.
A 10-point change in EQ VAS score was associated with a 22% (95% CI 15–28) reduction in the risk of complications for patients with no known clinical history of any events. For individuals with a history of prior events, the EQ VAS score was not significantly associated with future complications (P = 0.054).
CONCLUSIONS—
This study demonstrates that self-rated health reported on the EQ VAS provides additional valuable information on the risk of vascular events and complications in individuals with type 2 diabetes over and above that determined from clinical history and established risk factors alone.
Possible reasons for patients with lower self-rated health profiles having poorer outcomes are that self-rated health profiles are markers of disease severity or could indicate diabetes-related complications at a subclinical stage. Interestingly, the increase in relative risk associated with the EQ VAS is less for acute events such as mydocardial infarction than for more chronic complications.
Self-rated health has received little attention in the literature on risk assessment for diabetes and was not listed in a recent editorial of emerging risk factors for cardiovascular disease (12) or in a statement on risk factors for cardiovascular disease in diabetes (13). One recent study indicated that measures of social functioning were significant predictors of mortality and disability in people with diabetes (14), and the Wisconsin Epidemiologic Study of Diabetic Retinopathy showed that health-related quality of life can predict mortality in people with type 2 diabetes (15). Our study extends these findings by demonstrating that self-reported health profiles can predict which patients are at higher risk for major complications of diabetes even after taking into account established risk factors.
The EQ VAS could potentially be used in a clinical setting to target primary prevention in diabetes patients. Our results suggest that a 10-point difference in EQ VAS score stratifies risk of complications to a greater degree than either a 1-unit change in the ratio of total cholesterol to HDL cholesterol or a 10-year longer duration of diabetes. A key feature of the VAS in this setting is its simplicity: it only requires patients to indicate a score on a 0 to 100-point scale to represent their current health status.
The models presented here focus on a single measure of self-rated health and demonstrate how variations in this measure across a population affect subsequent outcomes. Whether change in health status at an individual level can also predict risk is unknown. Addressing this question would require repeated measures of EQ VAS on the same individual.
Cox proportional hazard ratios (HRs) for fatal and nonfatal vascular events and complications of diabetes
| . | HR for vascular events (453 events)* . | P . | HR for diabetes complications (193 events)† . | P . |
|---|---|---|---|---|
| EQ VAS (per 10-point increase) | 0.94 | 0.02 | — | — |
| No prior events | — | — | 0.78 | <0.001 |
| Prior events | — | — | 0.93 | 0.054 |
| Women | 0.77 | 0.013 | 0.57 | 0.001 |
| BMI | — | — | 1.05 | <0.001 |
| A1C per % | 1.20 | <0.001 | 1.45 | <0.001 |
| Age (per 10 years) | 1.49 | <0.001 | 1.79 | <0.001 |
| Diabetes duration (per 10 years) | — | — | 1.04 | <0.001 |
| Current smoker | 1.58 | 0.002 | 2.27 | <0.001 |
| Systolic blood pressure (per 10 mmHg) | 1.16 | <0.001 | — | — |
| Total–to–HDL cholesterol ratio | 1.14 | 0.003 | 1.13 | 0.077 |
| Prior events‡ | 3.31 | <0.001 | — | — |
| . | HR for vascular events (453 events)* . | P . | HR for diabetes complications (193 events)† . | P . |
|---|---|---|---|---|
| EQ VAS (per 10-point increase) | 0.94 | 0.02 | — | — |
| No prior events | — | — | 0.78 | <0.001 |
| Prior events | — | — | 0.93 | 0.054 |
| Women | 0.77 | 0.013 | 0.57 | 0.001 |
| BMI | — | — | 1.05 | <0.001 |
| A1C per % | 1.20 | <0.001 | 1.45 | <0.001 |
| Age (per 10 years) | 1.49 | <0.001 | 1.79 | <0.001 |
| Diabetes duration (per 10 years) | — | — | 1.04 | <0.001 |
| Current smoker | 1.58 | 0.002 | 2.27 | <0.001 |
| Systolic blood pressure (per 10 mmHg) | 1.16 | <0.001 | — | — |
| Total–to–HDL cholesterol ratio | 1.14 | 0.003 | 1.13 | 0.077 |
| Prior events‡ | 3.31 | <0.001 | — | — |
A 10 point increase in EQ VAS is associated with a 6% lower risk of vascular events and a 22% lower risk of major complications (patients with no prior events). Separate HRs are presented where there are significant differences in the terms for EQ VAS interacted by prior events.
Vascular events included vascular death, myocardial infarction, stroke, and hospitalized angina. During follow-up, 453 patients had one or more vascular events, including vascular death (n = 89), myocardial infarction (n = 129), stroke (n = 83), and hospitalized angina (n = 152).
Diabetes complications included heart failure, lower extremity ulcers, amputation, and renal dialysis. During follow-up, 193 patients experienced diabetes complications, the most prevalent of which was heart failure (n = 95), followed by ulcers (n = 44), amputations (n = 30), ulcer plus amputation (n = 7), and renal dialysis (n = 17).
A total of 1,454 patients had a history of any prior events (nonfatal vascular events or diabetes complications defined above) at the time of administration of the EQ-5D questionnaire.
Article Information
A detailed list of acknowledgments can be found in the online appendix.
References
Published ahead of print at http://care.diabetesjournals.org on 9 January 2008. DOI: 10.2337/dc07-1391.
Additional information for this article can be found in an online appendix at http://dx.doi.org/10.2337/dc07-1391.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C Section 1734 solely to indicate this fact.
