Association of Mortality and Years of Potential Life Lost With Active Tuberculosis in the United States

This cohort study examines the risk of mortality and years of potential life lost among patients with active tuberculosis disease compared with individuals without active tuberculosis.


Introduction
In low-incidence countries, such as the United States, most active tuberculosis (TB) disease is due to reactivation of latent TB infection (LTBI) 1 and thus is preventable through screening for and treatment of LTBI. 2 However, despite LTBI screening being an established recommendation by numerous guidelines, 2 many studies have documented low screening and treatment rates. 3 For example, in California, only 12% of the estimated 2 million people with LTBI have received LTBI treatment. 4 As active TB incidence has declined in the United States, the perceived value of TB prevention has declined for clinicians and health systems. As we have shown previously, active TB disease continues to be associated with high health care use and mortality within the first year after diagnosis. 5 Many studies have documented early mortality in the first year after TB diagnosis, 1 but few have explored delayed mortality or residual mortality risk more than 1 year after TB diagnosis. 6,7 Recognizing delayed mortality risk is important to understanding the full burden of active TB disease and the potential value of LTBI screening and treatment programs.
Persons who develop active TB disease are at high risk of mortality in the first year after diagnosis, yet those who survive their initial episode of active TB disease are considered cured and are not expected to have increased residual mortality. Some models of cost-effectiveness rely on this assumption. 8 Unfortunately, health deficits might persist even after successful treatment and lead to increased risk of death. [9][10][11][12][13][14][15][16][17][18][19][20][21] Thus far, only 1 study in the United States has assessed delayed mortality and years of potential life (YPL) lost among TB survivors and has shown an increased risk of mortality and YPL lost among TB survivors compared with patients with LTBI. 6,7 Smaller studies from the Netherlands; Estonia; Liverpool, the United Kingdom; and Barcelona, Spain have also found increased mortality among persons with active TB disease. [22][23][24][25] However, all of the studies used different methods, and only 2 studies addressed delayed mortality among TB survivors. 22,24 The present study expands on previous work as we (1) assess the risks of early and delayed mortality in a single, large retrospective cohort and thus provide data on the relative contributions of early vs delayed mortality; (2) assess the possible association of different risk factors with early vs delayed mortality; (3) compare the incidence of early and delayed mortality among patients with active TB compared with a matched cohort and adjust for key comorbidities to ascertain whether excess mortality risk among patients with active TB disease is due to active TB disease or underlying risk factors that predispose patients to develop active TB disease; and (4) estimate YPL gained or lost among patients with active TB disease compared with a comparison cohort adjusting for demographic and clinical characteristics.

Methods
This cohort study was approved by the Kaiser Permanente Northern California (KPNC) Institutional Board Review with a waiver of the requirement for informed consent as a data-only study using information collected as part of routine care.

Statistical Analysis
Data were analyzed from January 2019 to January 2020. All analyses were performed using SAS, version 9.4 (SAS Institute Inc). Bivariate and multivariable analyses were conducted to examine the relationships between patients' demographic and clinical characteristics and death. Clinically relevant and statistically significant variables in the bivariate analysis were included in a multivariable Cox proportional hazards model to assess factors associated with early and delayed mortality among patients with active TB disease. Similarly, clinically relevant and statistically significant variables in the bivariate analysis were included in a multivariable Cox proportional hazards model to estimate adjusted incidence rates for early and late mortality among patients with active TB disease and among the comparison cohort. Adjusted incidence rate ratios were calculated to estimate differences between patients with active TB disease and the comparison cohort. The Kaplan-Meier method was used to calculate the survival curves for patients with active TB disease and the comparison cohort.
We calculated YPL lost or gained for both the active TB and comparison cohorts compared with the United States Life Tables, 2015. 26 The YPL represents the potential life expectancy for each individual, and differences in the average YPL between the active TB and comparison cohorts represent potential disparities in life expectancy between the 2 groups. We adapted the life expectancy method 27 Tables, 2015 26 ; this value could be either a loss (died before average life expectancy) or a gain (died after average life expectancy). We then conducted linear regression modeling to estimate the differences in YPL between the active TB and comparison cohorts adjusting for demographic variables and comorbidities. A 2-sided P < .05 was considered statistically significant, and 95% CIs were presented for all model estimates.

Results
We identified 2522 persons with 17 166 persons-years of follow-up who had microbiologically confirmed active TB disease between 1997 and 2017 and included 100 880 persons with 735 726 person-years of follow-up in the comparison cohort. The active TB relapse rate in KPNC from 1997 to 2017 was low, with 8 of 2522 active TB patients experiencing relapse in 17 166 person-years of follow-up, for a rate of 0.047 per 100 person-years. In the active TB and comparison cohorts, similar percentages of persons were male (56.3% vs 55.6%), aged 45 to 64 years (33.7% vs 33.7%), and aged 65 years or older (24.7% vs 24.7%) ( Table 1). The 2 groups had notably different race/ethnicity distributions: in the active TB cohort, 57.4% were of Asian/Pacific Islander race/ethnicity compared with 14.2% of the comparison cohort. In the active TB cohort, 7.0% had early mortality and 16.3% had delayed mortality compared with 1.1% and 12.0% in the comparison cohort, respectively. Overall survival probability was significantly lower (log-rank P < .001) in the active TB cohort compared with the comparison cohort ( Figure). Compared with the comparison cohort, persons in the active TB group were more likely to have diabetes (25.5% vs 9.4%), HIV infection (2.5% vs 0.3%), end-stage  Table 3 presents aHRs between the active TB and the comparison cohort. Mortality probability was higher among TB patients compared with the comparison cohort for both within year 1 after diagnosis or matching date (early mortality) and after year 1 (delayed mortality). The adjusted risks of early mortality (aHR, 7.29; 95% CI, 6.08-8.73; P < .001) and delayed mortality (aHR, 1.78; 95% CI, 1.61-1.98; P < .001) were significantly higher among patients with active TB disease compared with the comparison cohort after adjusting for age, sex, race/ethnicity, and selected comorbidities. Thus,

Discussion
Active TB disease is preventable, but screening and treatment for LTBI is underutilized in the United States. 3 Engagement in TB prevention by clinicians, health systems, and policy makers is low, partially owing to an underappreciation of the full burden of active TB disease. In this cohort study, we evaluated the risk of mortality among persons with active TB disease in an integrated health system who were insured and had access to high-quality health care. We found that compared with a comparison cohort of KPNC members, persons with active TB disease had an increased risk of early and delayed mortality and significant YPL loss even after adjusting for known factors that increase the risk of developing active TB disease (eg, diabetes, HIV infection, end-stage kidney disease, and solid organ transplantation). Expanding on prior studies, we calculated both the early and delayed mortality risk associated with TB and provide additional data on the potential value of investing in TB prevention.
Patients with active TB disease within KPNC were overall similar to persons with active TB disease in California; a high percentage were Asian/Pacific Islander and 25 years or older, and a substantial percentage had comorbidities that are known to increase risk of TB reactivation, such as HIV infection. 4 Compared with the comparison cohort, persons with TB were more likely to be Asian/ Pacific Islander and, as expected, were more likely to have comorbidities that increase the risk of LTBI reactivation (eg, HIV infection). Although many factors are associated with increased risk of progression to active TB disease in the literature (eg, use of tumor necrosis factor alpha inhibitors), 2,28 we were only able to include variables that could be extracted from our EHR databases and have high enough prevalence to be meaningfully included in this analysis; for example, we did not include persons who have been exposed to tumor necrosis factor alpha inhibitors, as in the 20-year period included in this analysis, we found only 24 persons with active TB who had been prescribed a tumor necrosis factor alpha inhibitor. Thus, we recognize the inherent limitation of not being able to completely account for all covariates that might meaningfully influence the association between active TB disease and mortality compared with a comparison cohort. However, to our knowledge, our study is the only one to date to assess early and delayed mortality risk adjusting for comorbidities.
Interestingly, we found different factors associated with early vs delayed mortality among persons with active TB disease. For example, HIV infection was only associated with early mortality,  whereas diabetes was only associated with delayed mortality (Table 2). This exploratory analysis suggests that certain comorbidities might increase early and delayed mortality risk differentially among patients with TB; this finding should be explored further.
We found increased risk of both early and delayed mortality among patients with active TB disease compared with a comparison cohort after adjusting for age, sex, race/ethnicity, and selected comorbidities (Table 3). Patients in the active TB cohort had an increased aHR in all subgroups compared with the comparison cohort, suggesting that the increased hazard is associated with the early and delayed effects of active TB disease rather than the key comorbidities that increase risk of TB reactivation. This finding suggests that active TB, even when treated, may lead to increased  mortality and YPL loss. There are several possible mechanisms for increased delayed mortality.
International studies of posttreatment mortality among TB patients suggest high rates of relapse leading to death, 29 but this does not seem to be the case in the United States 30 and more specifically in our cohort. Post-TB disease pulmonary sequelae, decreased pulmonary function, and other sequelae have been noted. 9,11,12,14,15,17,18,20 In addition, persons with some severe forms of TB disease (eg, TB meningitis) have poor survival even after completion of treatment. 19 Alternatively, among patients with comorbidities, TB disease might complicate management of the underlying comorbidity and thus lead to increased risk of death; for example, drug-drug interactions between TB and antiretroviral medications can lead to use of second-line or alternative agents; modification or reduction in immunosuppression among posttransplant patients to improve TB treatment outcomes can lead to organ rejection. This study was not designed as a cause-of-death study, and thus we cannot delineate a clear cause among those with early or delayed mortality. However, the increased risk of delayed mortality among persons with active TB disease should prompt additional work into better understanding the long-term sequelae of active TB disease. In addition, a re-evaluation of the potential benefits of LTBI screening and treatment may be warranted, especially among persons at high risk of TB reactivation, such as those with immunosuppression, diabetes, or end-stage kidney disease.

Limitations
This study has several limitations. First, although we tried to account for all variables that might be associated with both an increased risk of developing TB and an increased risk of mortality, we were limited by either inability to extract certain data or insufficient sample size. Although a fully adjusted analysis that includes all variables that might contribute to mortality is ideal, it is not feasible in either our EHR data set or most other data systems. Thus, there might be residual confounding due to medical comorbidities we did not examine in this analysis or other factors, including unmeasured confounding such as due to differences in socioeconomic status. Despite these limitations, our analysis expands on the work of others to adjust for known medical comorbidities. Second, data on country of birth are not systematically captured in our EHR and could not be extracted as an analytic variable. We recognize this is an important variable to consider for the future. Similarly, certain variables, such as smoking, alcohol use, and other substance use are not available for streamlined extraction within our EHR. Third, KPNC members belong to a large, integrated health system and are thus insured; previous analyses comparing KPNC members to the general adult population of Northern California suggest that the 2 populations are similar but have some differences in socioeconomic status. 31 These differences could impact the generalizability of our results to all of California or other regions of the United States. Fourth, KPNC members have higher life expectancy than the general US population, and thus we used an internal comparison cohort for all comparisons to minimize introducing bias into the analysis.

Conclusions
In summary, active TB disease was associated with increased early and delayed mortality and more than 7 YPL lost even after adjusting for demographic characteristics and comorbidities. As most active TB disease can be prevented with screening for and treatment of LTBI, these data provide critical estimates to appropriately value LTBI programs. By improving screening practices and early treatment of LTBI, we may be able to reduce much of the early and delayed mortality associated with active TB.

JAMA Network Open | Infectious Diseases
Association of Mortality and Years of Potential Life Lost With Active Tuberculosis