Type-2 diabetes has been described to be associated with a reduced health related quality of life [3, 4]. Using the EQ-5D questionnaire we found that this was mostly related to the presence of co-morbidity and episodes of hypoglycaemia but not to average glycaemic control. With a median score of 0.9 diabetic patients scored high on the EQ-5D in general while there were only few patients with substantially lowered scores (<0.5). On the other hand changes as little as 0.05 have been reported as the minimal important difference for the EQ-5D , illustrating that there are clinically relevant differences in HRQoL between those in the lowest and highest tertile of the EQ-5D range seen in DiaRegis.
We found no association between glycaemic control (defined as HbA1c, fasting- or postprandial glucose levels) and changes in the EQ-5D scoring, which remained robust even in multivariable analyses. While the findings are in agreement with a number of other analyses [4, 23, 24], they are in contrast to data reported by Testa et al., who found that improved glycaemic control was associated with substantial improvements in QoL . Quality-of-life treatment differences (SD units) for symptom distress (+0.59; p < 0.001), general perceived health (+0.36; p = 0.004), cognitive functioning (+0.34; p = 0.005), and the overall visual analog scale (VAS) (+0.24; p = 0.04) were significantly more favorable for active therapy. In that study patients with established type-2 diabetes were washed out from prior antidiabetic pharmacotherapy for 3 weeks and then randomized to glipizide GITS or placebo for 12 weeks. HbA1c increased to 9.3% and fasting blood glucose to 168 mg/dl in the placebo group while it was reduced to 7.5% and 126 mg/dl in the glipizide group. In another analysis by Klein et al. diabetic patients after 14 years of follow-up were subjected to the Short Form 36 (SF-36), which demonstrated an improved quality of life with low HbA1c values , this was however not multivariable adjusted.
On this background the lack of an association between glycaemic control and HRQoL in our study might be related to the fact that the group of patients was rather homogenous, selected by restricting recruitment to those with oral antidiabetic therapy, and showing a rather narrow range of HbA1c values (median 7.4; interquartile range 6.8–8.2) that appeared to be stable over time. Taken together this might suggest that average contemporary glycaemic control in a stable environment might prevent detecting differences in QoL that may arise in patients with extreme differences in glycaemic control.
Differences in HRQol with the selection of oral antidiabetic pharmacotherapy are, on the background of a largely absent impact of glycaemic control, related to the side effect profile which might include weight gain and/or hypoglycaemia. In the analysis of our dataset which included patients being largely treated with metformin (84.6%) and/or sulfonylureas (28.1%) and also to a lesser extent with other oral antidiabetic drugs, we found no impact of antidiabetic pharmacotherapy in univariable or multivariable regression analyses. This is compatible with other analyses of observational registries and larger outcomes studies which also identified no such association [4, 26, 27].
Hypoglycemia is a frequent treatment related complications that might be reduced by properly selecting antidiabetic pharmacotherapy and considering specific patient characteristics use [28, 29]. There are only few reports demonstrating that hypoglycaemia with oral antidiabetic drugs might have an impact. Alvarez-Guisasola et al. reported for example from a large observational, multicenter, cross sectional study including 1,709 patients with type-2 diabetes (mean age 63 years, 45% female, mean HbA1c 7.1%, 38% hypoglycaemic symptoms within the last 12 months), that QoL was substantially reduced with hypoglycaemia . This is well compatible with our own observations. Patients received either sulfonylureas or thiazolidinediones on top of metformin monotherapy. Those reporting hypoglycaemic symptoms had significantly lower EQ-5D VAS scores (mean difference -4.33, p < 0.0001) in adjusted linear regression analyses. Relative to those not reporting symptoms, the adjusted decrement to quality of life increased with severity of hypoglycaemia (mild: -2.68, p = 0.0039; moderate: -6.42, p < 0.0001; severe: -16.09, p < 0.0001). They established no link however between the choice of either sulfonylureas or glitazones and the frequency of hypoglycaemia. Further research is clearly needed to link antidiabetic treatment options to hypoglycaemia and their impact on health related quality of life. It is however plausible that the incidence rates of (symptomatic) hypoglycaemia with single agents in our cohort is not sufficient to establish such a link from a statistical perspective.
Co-morbidity and patient related variables
Patients in our cohort had a substantial co-morbidity which ranged however from none to more than 6 co-morbidities. After multivariable analysis hypertension (OR 1.62; 95%CI 0.93–2.84), peripheral neuropathy (OR 1.73, 95%CI 1.03–2.93) and clinically relevant depression (OR 11.01; 95%CI 3.97–30.50) but not heart failure or coronary artery disease (CAD) remained significant predictors. This is in partial agreement to previous findings [10–14] that have repeatedly shown, that symptomatic more than asymptomatic co-morbidities are responsible for this observation. Miksch for example illustrated that the impact of osteoarthritis as a disabling and painful condition reduces QoL more than hypertension which is largely asymptomatic . Wexler et al. also found that patients with symptomatic co-morbidities such as microvascular complications and heart failure had a substantially reduced QoL, while those without symptoms showed no reduction . These findings are also consistent with our observation that peripheral neuropathy had a higher impact on HRQoL than coronary artery disease and even heart failure. It was surprising however for us to find that heart failure did not reduce quality of life in our multivariable model while it did in the univariable analyses, because it is in contrast to previous reports . This might have been due to multiple adjustments in the multivariable model, which reduced the impact of heart failure.
We were not able to show a direct impact of physical activity on HRQoL, which was elegantly demonstrate by Daniele et al. , who found an improved HRQoL in physically active patients with diabetes. This might provide, beyond improvements of metabolic control and body weight, an opportunity to improve HRQoL.