Under the status quo and considering only acute TB consequences, we projected 927 (95% UR 764–1,123) persons to develop TB and 30 (95% UR 24–36) to die from TB; we projected 312 (95% UR 241–386) QALYs lost. We projected TB-related health system expenditures to total $25.3 million (95% UR $7.6–$61.9 million) over the 25-year time horizon (Table 2; Appendix Table 3). Inclusion of post-TB death increased the estimated number of QALYs lost because of TB 1.6-fold (95% UR 1.4–1.7-fold), representing a 60% increase compared with acute TB consequences only. Similarly, inclusion of post-TB illness increased the estimated number of QALYs lost because of TB by 1.5-fold (95% UR 1.2–2.0-fold). Although including excess healthcare costs post-TB did not affect QALYs lost, it increased estimated TB-related health system expenditures to $27.6 million (95% UR $9.7–$64.5 million). When all post-TB consequences were included, TB was projected to result in 657 (95% UR 479–876) QALYs lost, a 2.1-fold (95% UR 1.7–2.5-fold) increase compared estimates excluding post-TB consequences (Table 2; Appendix Table 3). This result suggests post-TB consequences accounted for 52% (95% UR 41%–60%) of QALYs lost because of TB.

Across all scenarios, a postarrival TB infection screening intervention was estimated to result in QALY gains, increased costs, and reduced TB episodes and deaths (Table 3). Compared with a scenario including only acute TB consequences, including post-TB death resulted in a slightly greater increase in estimated incremental QALYs gained by the intervention than including post-TB disease. When all 3 post-TB consequences were incorporated, estimated QALY gains increased by 2.4-fold (95% UR 1.6–4.4-fold), estimated TB deaths averted increased by 2.2-fold (95% UR 1.6–3.0-fold), and incremental costs were projected to be 4% (95% UR 1%–13%) lower compared with the exclusion of post-TB consequences.

When only acute consequences of TB were included, the estimated ICER of a postarrival TB infection screening intervention was $235,088 per QALY gained, compared with the status quo. When all 3 post-TB consequences were included, the ICER was reduced to $100,742. In that scenario, 53% of simulations had an ICER below a WTP threshold of $100,000, compared with only 8% when only acute consequences were included (Table 3; Appendix Figure 2).

Figure

Figure. PRCC for incremental net monetary benefit as drivers of cost-effectiveness of tuberculosis prevention among immigrants, Canada. Incremental net monetary benefit refers to the difference in net monetary benefit between the...

When all post-TB consequences are included, the PRCC analysis showed that post-TB illness and death were positively correlated with iNMB, although excess post-TB healthcare costs were not (Figure). TB infection test specificity, cost of TB disease treatment, TPT effectiveness, annual TB progression rates, and health disutility from post-TB illness were the 5 parameters most positively correlated with iNMB. Post-TB death was the 11th most positively parameter correlated with iNMB. Similarly, we found estimates of effectiveness and cost-effectiveness of postarrival TB infection screening were most sensitive to the disutility associated with post-TB disease and minimally sensitive to post-TB death and excess post-TB healthcare costs (Appendix Table 4).

Estimates of the relative contribution of post-TB consequences to QALYs lost because of TB disease were inversely correlated to the acute consequences of TB (Appendix Table 5), but overall ICERs were more favorable when post-TB consequences were included (Appendix Table 6). Substitution of an interferon-γ release assay for the tuberculin skin test did not substantially affect estimated ICERs (Appendix Table 7) because specificity gains were offset by increased costs. Relative increases in QALYs gained after incorporating post-TB consequences and PRCC estimates were largely unchanged across discount rates (Appendix Tables 8, 9, Figures 3, 4). Unsurprisingly, ICERs were more favorable with no discounting and more unfavorable with 3% discounting. When all post-TB consequences were included, the estimated ICER was $71,714 per QALY gained, and 76% of simulations were <$100,000 per QALY gained with no discounting, whereas the corresponding outcomes with 3% discounting were $137,407 and 27%.

A more conservative parameterization of post-TB death substantially reduced the effect on QALYs lost because of TB and QALYs gained by a screening program. With a more conservative parameterization (9% vs. 69%), including post-TB death only increased the number of QALYs lost because of TB by 5% (95% UR 2%–9%). As a result, overall post-TB consequences accounted for a smaller proportion of QALYs lost because of TB than our base analysis at 44% (95% UR 23%–64%). In addition, the more conservative post-TB death risk reduced its influence on QALYs gained through the screening program and on the ICER (Appendix Tables 10, 11). Reducing the TB progression rate did not affect long-term TB consequence effect on QALYs gained through a screening program, but it did reduce the estimated cost-effectiveness (Appendix Tables 12, 13).

We found excluding post-TB consequences of illness, death, and excess healthcare costs can lead to substantially undervaluing TB prevention strategies in a low-incidence, high-income setting like Canada. Post-TB consequences account for a large fraction of QALYs lost because of TB. Including post-TB consequences in our analysis of a post-arrival TB infection screening and preventive treatment intervention among persons immigrating to Canada more than doubled the estimated QALYs gained and reduced the estimated ICER by >50%.

One of the most studied use cases for TB infection screening and preventive treatment in low-incidence countries is among persons who recently immigrated (4,30,31). Yet, the literature on that topic is largely divided when it comes to cost-effectiveness (8). Most of the literature evaluating cost-effectiveness has not incorporated post-TB consequences. A recent systematic review (8) found only 3 (32–34) of 14 included studies evaluating postarrival TB infection screening in immigrants considered post-TB consequences. None of those 3 studies considered increased healthcare cost, and only 1 study (34) considered elevated risk for death after TB. Although all 3 studies considered persistent illness after TB, its inclusion was highly variable across studies, which unsurprising because of the paucity of prospective cohort data estimating health utility decrements associated with post-TB disease. Improving our understanding of the long-term effect of TB on health utility should be a priority, considering our finding that the cost-effectiveness estimates for TB prevention were sensitive to post-TB disease parameterization.

A global analysis of the health burden of incident TB estimated that 47% (95% UR 37%–57%) of TB-related disability-adjusted life years were attributed to post-TB disease and death (9). In our study, post-TB consequences accounted for 52% (95% UR 41%–60%) of TB-related QALYs lost. Our higher estimate might reflect a lower number of deaths and lower health disutility during the acute phase of TB in Canada but higher estimated post-TB death. When we used a conservative estimate of post-TB death, like the previous global study, our estimated contribution was only 44% (95% UR 23%–64%).

By neglecting the contribution of post-TB consequences on health disutility, most literature to date has likely undervalued the potential benefits of TB prevention strategies. Because ICERs for TB prevention among recent immigrants are typically above common WTP thresholds (30,31), reductions in the ICER might have meaningful policy implications. Also, ensuring that cost-effectiveness is accurately and comprehensively measured is necessary for policymakers seeking to prioritize scarce healthcare funds.

The first major strength of our study is that we parameterized post-TB consequences in our analysis by using high-quality data from Canada or from other published literature, with appropriate consideration of uncertainty. Second, we leveraged a common use case for our analysis, with our model structured to capture necessary elements of the care cascade and TB natural history. Finally, we also evaluated several implications of ignoring post-TB consequences relevant to both researchers and decision-makers.

The first limitation of our study is that we did not model secondary TB transmission beyond the cost of routine contact investigations. Including future TB cases resulting from secondary transmission would likely improve the estimated cost-effectiveness of the intervention. Second, we assumed that the observed increases in death, disease, and healthcare costs after TB were attributable to TB only and therefore preventable through TPT. Although we carefully adjusted the analyses of our underlying parameterization of those estimates, it is not possible to determine if TB was the only cause of those outcomes (35), and there might have been residual confounding factors. Third, we parameterized the elevated risk for illness and death post-TB to be lifelong. Our estimate of post-TB death risk was derived from a population-based study of permanent residents of British Columbia with a mean follow-up of 13 years. The elevated death risk persisted throughout the study period, suggesting it might be long-term. However, data on post-TB illness are less certain. Post-TB respiratory disease is heterogeneous (35), and some data suggest that the increase in post-TB respiratory disease might dissipate over time (36). However, post-TB illness is not limited to the respiratory system, and nonrespiratory disease might be substantial (37,38). Those 2 considerations might lead to inaccurately assessing the effect of post-TB illness. Finally, we only considered health system costs related to TB. We did not include costs incurred by patients and caregivers during or after treatment. No comprehensive data on those costs are available in Canada (39). It is unclear how including patient-incurred costs would affect cost-effectiveness because patients incur costs during both TB disease and TPT (40,41).

In summary, neglecting post-TB consequences underestimated the effectiveness of a postarrival infection screening and treatment intervention among persons who recently immigrated to Canada by more than 2-fold. Including all post-TB consequences reduced the estimated ICERs from $235,088 to $100,742 per QALY gained. Our findings suggest the long-term health and economic burden of TB substantially affects how TB prevention is valued. Those findings should motivate TB preventive efforts and prompt the continued collection of relevant cohort data to fill critical knowledge gaps. We encourage the consistent inclusion of post-TB consequences in analyses of TB prevention to guide policy in low-incidence settings moving forward.