The prevalence and associations of low testosterone levels and erectile dysfunction in a male diabetic urban population

Abstract

Introduction: According to the literature, low serum testosterone levels are associated with diabetes mellitus. Minimal data exist for its prevalence or predictors in South Africa. Erectile dysfunction is a common condition in diabetic patients. The prevalence and predictors in our patient population is unknown. Methods: An observational, cross-sectional study was performed in 150 consecutive male diabetic patients over the age of 50 years in the Diabetic clinic of Steve Biko Academic Hospital. These patients were evaluated for diabetes control and complications, the presence of erectile dysfunction and for hypogonadism symptoms. Morning serum testosterone levels were done. Subjects with low testosterone levels were compared to those with normal levels. Results: The mean age of the patients was 62 years (standard deviation (SD) 7.87), 91.3% had type 2 diabetes, and 84.7% were on insulin. The mean duration of diabetes was 15 years (SD 8.65). The mean body mass index was 30.7 (SD 5.37), the mean waist circumference was 112.4cm (SD 16.42), the median creatinine was 96μmol/L (interquartile range (IQR) 79-133) and the median HbA1C was 7.85% (IQR 6.80-9.30). Ischaemic heart disease was previously diagnosed in 40.7% of patients. Some degree of erectile dysfunction was reported in 95.3% of the patients with 51.3% reporting serious dysfunction. The prevalence of androgen deficiency symptoms was 94.7%. Fifty percent of the men had low total testosterone levels; 40.7% had low modified calculated bioavailable testosterone levels, and in 27.3% both were low. With multivariate logistic regression the significant factors associated with low total testosterone were waist circumference and known cardiovascular disease. For a low modified calculated bioavailable testosterone level significant variables were age, diabetes duration and body mass index and for an outcome defined as both the above the significant factors were diabetes duration, body mass index, and known cardiovascular disease. With multivariate logistic regression the significant factors associated with erectile dysfunction were age, body mass index, peripheral neuropathy score, and diuretic therapy. The prevalence of symptoms of androgen deficiency was very high with 94.7% of all patients reporting a significant amount of symptoms on the Androgen Deficiency in Adult Males (ADAM) questionnaire. If only the total serum testosterone level was evaluated instead of the modified calculated bioavailable testosterone, the sensitivity was 69%, the specificity was 63%, with a poor positive predictive value of only 56%. The negative predictive value was better at 75%. Differences in quality of life scores were only seen for some erectile dysfunction subgroups but not for low testosterone levels. Conclusion: This study confirms the high prevalence of low testosterone levels and of erectile dysfunction in diabetic male patients in a tertiary setting, and argues in favour of universal screening of this population group. Multiple predictors of low testosterone levels and of erectile dysfunction were identified. The ADAM questionnaire was not useful in identifying subjects with a low testosterone level. Total testosterone testing alone performed poorly in comparison with modified calculated bioavailable testosterone and is not the recommended test of choice. Erectile dysfunction negatively affected the quality of life.