The study by Bishu et al.1 estimated the cost of missed workdays associated with diabetes among individuals aged 18 years or above using nationally representative 2011 Medical Expenditure Panel Survey (MEPS) data. The authors found that having diabetes was associated with 0.84 more of the number of missed work days, as well as with higher costs of missed work days, compared to those without diabetes; individuals with diabetes had $120 higher cost of missed workdays annually. Generalized linear model (GLM) methods have been proposed as better for heavily skewed cost data; in another study using GLM methods from MEPS data (2004–2006), Fu, Radican and Wells2 found that diabetic patients with macrovascular comorbid conditions cost more than diabetic patients without macrovascular comorbidity. Interestingly, there was considerable difference in the estimate of days lost between the two studies. Bishu et al. found that diabetic patients averaged 1.9 work days lost (2011 MEPS data), compared to 6.7 days estimated from 2004 and 2006 MEPS data. Both groups of investigators used the same estimating approach and both took into account the sampling weights, so it is hard to understand why the results would be so discrepant.

Assessing the economic burden of diabetes from the employer’s perspective is important because individuals of working age receive employment benefits that include sick days and disability coverage from their employers. For example, Ramsey et al.3 estimated the costs associated with work loss due to diabetes using 1998 administrative claims data where work loss costs were determined based on productivity costs and measured by “medically related absences”; this is more accurate than just using missed workdays. Because claims data provide information on actual dates of medical care and disability, including only missed workdays, this likely reflects a more accurate measure of productivity costs due to diabetes. The authors found that workers with diabetes had 8 days of work loss on average, compared to 5 days for non-diabetic workers. This finding is more consistent with the Fu, Radican and Wells study than with the Bishu et al. study. It is possible that diabetic patients in 2011 missed fewer work days than in 1998 or even in the 2004/2006 time period.