Reply to Dobler

To the Editor—We thank Dr Dobler for her comments regarding the methods of how we calculated and compared cumulative incidence of tuberculosis (TB) in cancer patients and country-specific TB incidence in our study [1]. We estimated the cumulative incidence (proportion) of developing TB during a specific study period in cancer cohorts divided by the mean annual TB incidence in the general US population over the same study period. We were not able to estimate the incidence of TB per year or by unit of time of follow-up in cancer patients because neither time to TB diagnosis of all cases nor the amount of time cancer cohorts were followed was reported in the included studies. Although we adjusted for the change in TB incidence in the general population over the study period, we assumed that the majority of TB cases would have occurred early after the cancer diagnosis. We did not account for the fact that the denominator estimated the annual TB incidence in the United States (also a proportion) and thus all new cases occurring in the US population per year. On reexamination of the data, 2 of the included studies provide insight into the timing of onset of active TB after a cancer diagnosis and support our assumption [2, 3]. In the study by Kaplan et al, 41% of TB cases occurred concurrently with the cancer diagnosis and 90% of all cases occurred within 18 months of cancer therapy [2]. Similarly, in the study by Libshitz et al, 30% of TB cases were diagnosed concurrently with the cancer and 79% of all cases occurred within 18 months of completion of cancer therapy, which for most cancers is approximately 2 years after cancer diagnosis [3]. Extrapolating from studies in which follow-up time of cancer cohorts was reported (Table 1 of our manuscript), this was usually relatively short (2–5 years) whereas study periods were generally longer (up to 13 years) [1]. Thus, cancer patients in the Kamboj study were likely not followed for the entire 25-year study period and, if the timing of TB after cancer diagnosis followed the same pattern as the Kaplan and Libshitz studies, the majority of TB cases also would have occurred within 2 years of cancer diagnosis [2–4].