This is to the best of our knowledge the first research that evaluates prospectively the association of starting insulin therapy with HRQoL and depressive symptoms specifically in elderly with T2D. Given the unique characteristics of this population (elevated prevalence of cardiovascular disease, frailty, increased risk of falls, more deleterious consequences from a hypoglycemic episode), it is of value that studies evaluating this specific population are conducted. Indeed, this population is more vulnerable to lose their independence and the introduction of insulin therapy increases treatment complexity, requiring some cognitive and visual skills [2].
We observed that starting insulin therapy in elderly participants with T2D was associated with significant improvement in depressive symptoms, as indicated by an important reduction in BDI score. We also found that the higher the improvement in depressive symptoms, the higher the improvement in physical functioning and role physical, as well as in pain domain. The influence of depressive symptoms in HRQoL of people with T2D is known and has been published elsewhere [16]. Recent data indicated a higher prevalence of depression in a Brazilian population with T2D and a consequent reduction in its HRQoL, most notably in physical functioning and physical role. However, this study evaluated subjects aged between 40–60 years. Therefore, we can not extrapolate these data [17]. Another study examined the relationship between severity of depressive symptoms and diabetes-related emotional distress in patients with T2D and evaluated whether this relationship is independent of demographic and clinical characteristics, such as sex, age, duration of diabetes, glycemic control, and diabetic complications. This study used the same questionnaires used in our research (SF-36, PAID and BDI). The results indicated that diabetes-related emotional distress (PAID) was significantly related to the severity of depressive symptoms. On multivariate analysis by regression model, it was found that this relation was independent of the demographic, including age, clinical characteristics, and therapeutic regimens [18].
We found a direct correlation between the BDI and PAID ranges, which signifies the higher the improvement on depressive symptoms, the bigger the improvement on HRQoL complaints registered in PAID score. We also found an inverse correlation between baseline BDI score and BDI range, which means the more severely depressed the individual at baseline, the smaller the improvement on depressive symptoms.
We did not find a clear change in HRQoL scores. However, when gender and educational level have been taken into account, we also observed that men had a greater improvement in physical functioning domain than women. This gender interaction with HRQoL has also been showed previously by Eljedi et al. His group have studied a sample of patients with T2D and evaluated HRQoL in this population in a cross-sectional design. Their results indicated that HRQoL of patients with T2D was strongly reduced and indeed identified that women were specially affected [19]. In Zhang’s study, it has also been identified an interaction between HRQoL and gender, and women were more severely affected [20]. Likewise, women did worse than men in Wandel’s review and in Akinci’s study [21, 22].
Educational level seems also to have a different meaning on our population. Participants with higher educational levels had a greater improvement in mental health when compared with those with lower educational levels, as previously shown [23]. Our patients with nephropathy had a worse outcome in physical functioning when compared to those without this comorbidity. Although these data have to be reassured, there is some available evidence indicating that proteinuria per se plays a role in HRQoL of patients with diabetic nephropathy [24].
Finally, we observed a trend of direct correlation between the range of BDI score and HbA1c levels, which means that the higher the reduction in HbA1c levels, the higher the improvement in depressive symptoms. However, this did not achieve statistical significance, possibly because of relatively small sample.
Indeed our data showed reduction in HbA1c levels, although of smaller magnitude than in non-elderly adult population studies, presumably because of the lower doses used in our sample (related to less strict glycemic targets) and the relatively short follow-up [4–7]. We did not observe an increase in body weight as described in other studies possibly for the same reasons [25].
The major limitation of our study is its relatively small sample size that could have influenced the statistical power to show a significant change in HRQoL scores. However, the number of individuals included is similar to the studies published previously [16, 26]. We adopted a convenience sample and have taken into account the study of Kamoi and cols, in which a prospective comparison of different types of insulin scheme has been performed, as a model. They have studied a smaller number of subjects than our study and have showed that HRQoL of patients treated by continuous subcutaneous insulin therapy is superior to that treated by multiple daily insulin injections (p < 0.05). The results in SF-36 and PAID scores were very similar, when whole sample was analyzed, which makes improbable that if the sample was greater we would find significant differences. However, it is important to note that the exclusion of the case of serious hypoglycemia may have influenced the results. The participants lost in the follow up maybe could have been those in whom insulin therapy had a different meaning. It could be also considered a bias. Furthermore, the design of the study (lack of a control-group and the open-label format) does not guarantee that the increase in BDI score was related to insulin therapy or to the improvement in glycemic control. A possible improvement in depressive symptoms related to the starting of insulin therapy in elderly diabetic subjects could be relevant in clinical setting in which starting insulin faces some barriers, especially in elderly subjects. However, these data must be investigated in a randomized controlled study and further research is warranted.