Dr. Hoenig et al highlight some of the intrinsic limitations of our retrospective cohort study. Although no control group was used, the annual incidence and prevalence of diabetes in our study group over the 1 year of follow-up was dramatically higher than expected in the general population. According to the Centers for Disease Control and Prevention, the average age-adjusted annual incidence rate of diabetes in adults from 2005 to 2007 was 11.1 per 1,000 individuals in Texas and 9.1 per 1,000 individuals for the entire United States population. The observed incidence rate for diabetes (301.6 per 1,000 subjects) in our cohort was markedly higher. Furthermore, the prevalence of diabetes in adults aged 65 to 74 years from 2005 to 2007 was 21.2% for Texas and 18.5% for the United States population. Although our patient population had a somewhat higher prevalence of diabetes at baseline (38%), at the end of 1 year of follow-up, the prevalence of diabetes (57%) was >2 times their Texan aged-matched counterparts and >3 times that of similar United States individuals, a magnitude suggesting a true finding.

As Dr. Hoenig et al correctly point out, the mean response rate among mail surveys published in medical journals is 60%. The general response rates for postal questionnaires, however, is 31.5%, a number similar to our response rate. Interestingly, a much larger study surveying Medicare-aged patients in the same catchment area received only a 12% response rate to a very short postal survey (4,653 respondents of 39,222 questionnaires). To assess whether our survey respondents differed from nonrespondents, we compared glycemic effects and found that blood glucose increased similarly in the 2 groups over the year of follow-up and that respondents actually trended toward more new diagnoses of diabetes (32% vs 11%, p = 0.056) compared with nonrespondents. Because the survey results were only a small aspect of our study, we do not believe that the low response rate should diminish the overall study findings.

Although we agree that our study design limitations preclude the attribution of causality, we believe that the findings do support our conclusions that further study in this area is warranted and that discussion with patients to ensure maintenance of tolerated physical activity after scooter prescription is important. We also believe very strongly that physicians and third-party payers should not be biased by our small, limited study against providing equipment that may substantially improve overall quality of life.