Risk of Incident Tuberculosis Disease in a Large Integrated Health Care System in California, 2004–2022

Abstract

Tuberculosis (TB) elimination, defined as TB disease incidence of <1/1 000 000 person-years, is a goal for both the state of California and the United States [1–3]. Over 85% of TB disease cases are due to reactivation among persons with latent tuberculosis infection (LTBI), and thus LTBI screening and treatment are key strategies to reduce TB incidence in the United States [4–6]. However, although LTBI screening and treatment for TB prevention are supported by >40 years of good quality evidence for both efficacy and effectiveness, are endorsed by both national and California-specific guidelines, and rely on readily available tests (eg, the tuberculin skin test, developed in 1907, and interferon-gamma release assay, approved in 2001) and treatments (eg, rifampin, introduced in 1968), implementation in most health systems has been poor [4–11].

LTBI screening and treatment are commonly incorporated into primary care settings, which deliver most recommended health prevention interventions (eg, cervical, colon, and breast cancer screening, diabetes and hypertension screening) [4]. However, most of our understanding of TB disease risk is from national and local public health surveillance systems, local tuberculosis control units, and special studies of immigrant populations; these data sources do not rely on longitudinal follow-up of well-defined populations and do not account for prior LTBI treatment [12, 13]. Few studies to date have defined the absolute risk of and risk factors for TB disease in a general population using a cohort design with longitudinal follow-up in a health care setting. These data are critical for primary care providers and health systems to understand TB disease risk for patients in primary care settings. Second, current guidelines define 2 main risk groups for LTBI screening: (1) persons born in TB-endemic countries (any country other than the United States, Canada, Australia, New Zealand, or a country in Western or Northern Europe) who have increased risk of TB infection due to exposure before immigrating to the United States; and (2) persons with medical conditions that increase risk of progression to TB disease (eg, current or planned immunosuppression). However, few studies have comprehensively assessed the potential overlap between these 2 non–mutually exclusive risk categories. Third, few longitudinal studies of TB disease risk have complete records of LTBI treatment and thus estimate the risk of incident TB disease in persons who have not been treated for LTBI.

To address these data gaps, we conducted a retrospective cohort study in a large integrated health care system with over 4.7 million persons and 35 million person-years of follow-up to (1) estimate the absolute incidence of TB disease among persons born in TB-endemic countries or with medical conditions that increase TB disease risk; (2) estimate the relative risk of TB disease among persons born in TB-endemic countries or with medical conditions; (3) define the overlap in TB disease risk among persons born in TB-endemic countries and those with medical conditions.

METHODS

Setting

Kaiser Permanente Northern California (KPNC) is an integrated health system that serves >4.7 million members in Northern and Central California and provides comprehensive preventive and curative care in inpatient and outpatient settings across 266 medical offices and 21 hospitals. Members receive all primary care services and most other clinical services, including laboratory testing, outpatient, and inpatient care in KPNC facilities. Members have similar sociodemographic characteristics to the diverse population of Northern and Central California [14].

Study Design

We conducted a retrospective cohort study of all KPNC members aged ≥18 years with at least 2 years of continuous membership between January 1, 2003, and September 30, 2022. We defined the start of follow-up (index date) as 1 year after date of first enrolling as a KPNC member during the study period. We excluded persons with any history of TB disease (microbiologically confirmed TB disease based on culture result or nucleic acid amplification test positive for Mycobacterium tuberculosis or International Classification of Diseases, 9th Revision [ICD-9], codes 010–018 before the index date or any treatment for LTBI or TB disease, defined as prescription fill for rifampin, isoniazid or rifapentine before the index date). The primary outcome was microbiologically confirmed TB disease based on a positive mycobacterial culture or nucleic acid amplification test for Mycobacterium tuberculosis. Persons were followed from index date until date of (1) diagnosis of TB disease (primary outcome); (2) any prescription fill for rifampin, isoniazid, or rifapentine as these medications are used for LTBI treatment and prevent progression to TB disease; (3) disenrollment from KPNC; (4) death; (5) end of study period on September 30, 2022. Data were obtained from the KPNC Virtual Data Warehouse, a common data model into which standardized data are extracted from clinical and administrative databases including an integrated electronic health record (EHR) database (Epic, Verona, WI, USA). The study was approved by the Kaiser Permanente Southern California and KPNC Institutional Review Boards with waivers of the requirement for informed consent.

Exposures and Covariates

As our primary exposure, we defined 5 nativity categories. Data on place of birth, including country, are collected during registration events at KPNC, and data on preferred language are collected as part of clinical encounters. We used EHR data on country of birth and preferred language to define the following nativity categories (detailed mapping in Supplementary Table 1): (1) “born in TB-endemic country” was defined as EHR documentation of birth in any country other than the United States, Canada, Australia, New Zealand, or Northern and Western Europe; (2) “non-US-born by language only” was used if there was no EHR documentation of country of birth, but there was documentation of preferred language predominantly spoken in TB-endemic countries (eg, Arabic, Quechua, Swahili); (3) “non-US-born in non-TB-endemic country” was based on EHR documentation of birth in Canada, Australia, New Zealand, or Western or Northern Europe; (4) “US-born” was based on EHR documentation of birth in the United States; (5) “unknown” was defined as no EHR documentation of country of birth and preferred language unknown or English. Persons born in TB-endemic countries were grouped into geographic regions (Caribbean, Central America, South America, Africa, Eastern Europe, Eastern Asia, South Asia, Southeast Asia, Central and Western Asia, Oceania).

As our secondary exposure of interest, we defined high- and intermediate-risk medical conditions for progression to TB disease. High-risk conditions for progression to TB disease included (1) HIV infection based on ICD 9th and 10th edition (ICD-9: 079.53, 042, V08, 795.71; ICD-10: B20, B97.71, Z21, O98.7) codes; (2) solid organ transplantation ICD codes (ICD-9: V42.0, V42.1, V42.6, V42.7, V42.83, V42.84; ICD-10: Z94.0-Z94.4); (3) hematologic malignancy based on ICD codes (ICD-9: 200–208; ICD-10: C81–C96); (4) receipt of tumor necrosis factor alpha inhibitors (TNFα) based on pharmacy records, including adalimumab, certolizumab, etanercept, golimumab, infliximab; (5) receipt of high-dose corticosteroids, defined as ≥20 mg prednisone equivalents of oral or systemic corticosteroids daily for ≥30 days based on pharmacy records; (6) receipt of other immunosuppressants based on pharmacy records, including receipt of abatacept, anakinra, auranofin, azathioprine, baricitinib, canakinumab, cyclosporine, fingolimod, guselkmab, leflunomide, mycophenolate mofetil, riskankizumab, rituximab, secukinumab, sirolimus, tacrolimus, thalidomide, tofacitinib, tocilizumab, and ustekinumab. Intermediate-risk conditions for progression to TB disease included (1) diabetes mellitus using ICD codes (ICD-9: 249, 250; ICD-10: E08–E11, E13); (2) end-stage renal disease using ICD codes (ICD-9: 585.6; ICD-10: N18.6). Other covariates included demographic characteristics (age, sex, race/ethnicity) and Charlson comorbidity index score [15].

Statistical Analysis

We calculated the crude incidence rate of TB disease during the follow-up period by dividing the number of TB cases by the total number of person-years and used an exact method to estimate the 95% CIs [16]. Using Cox proportional hazards models with a Nelson-Aalen estimator, we produced cumulative hazard curves stratified by nativity categories and medical conditions. To assess the risk of incident TB disease among persons born in TB-endemic countries and persons with medical conditions, we estimated adjusted hazard ratios (aHRs) and CIs using Cox proportional hazards models adjusting for potential confounders defined a priori, including age in years and sex. High- and intermediate-risk medical conditions were included in all models as time-dependent covariates as the onset date could occur after the index date but before the date of TB disease or end of follow-up. Moreover, to assess possible effect modification between nativity and high/intermediate-risk categories for TB progression, we fit separate Cox models among persons born in TB-endemic countries and US-born persons. In all models, we assessed the proportional hazards assumption by testing for slopes in Schoenfeld residuals [17]. All analyses used SAS software, version 9.4 (SAS Institute Inc., Cary, NC, USA). All comparisons were 2-tailed, with P < .05 considered significant.

RESULTS

In all, 4 761 427 persons with 35 591 565 person-years of follow-up were included in this analysis over the 18-year study period, and 1463 persons were diagnosed with TB disease (incidence rate, 4.11/100 000PY) (Tables 1 and 2). In the overall study population, 53.0% were age >40 years (median age [interquartile range {IQR}], 41 [29–53] years), 51.2% were female, 45.4% were non-Hispanic White, 6.08% were Black, 18.3% were Hispanic, and 17.5% were Asian. In all, 12.3% were born in TB-endemic countries, 4.43% were non-US-born by language only, 33.6% were US-born, and 48.5% had unknown nativity. Only 5.52% had a high-risk condition, and 15.6% had an intermediate-risk condition. Persons with incident TB disease were significantly more likely to be older (median age [IQR], 51 vs 41 [26 vs 24] years), male (58.4% vs 48.7%), Asian (67.1% vs 17.5%), born in a TB-endemic country (61.2% vs 12.3%), and to have a high-risk (20.4% vs 5.52%) or intermediate-risk (41.5% vs 15.5%) medical condition (P < .0001 for all comparisons). Only 13.9% of all TB cases were among US-born persons. Persons with unknown nativity comprised 48.5% of the population and accounted for 19.0% of all persons with TB disease.

The incidence of TB disease increased with age, from 2.94 (95% CI, 2.55–3.33) per 100 000PY among persons aged 30–39 years to 7.54 (95% CI, 5.40–9.67) per 100 000PY among persons aged 80–89 years (Table 2). Asian persons had the highest incidence rate among all race/ethnicity groups (IR, 15.8/100 000PY; 95% CI, 14.8–16.8/100 000PY). Persons born in TB-endemic countries (IR, 17.6/100 000PY; 95% CI, 16.4–18.7/100 000PY) or non-US-born by language only (IR, 6.43/100 000PY; 95% CI, 4.98–7.89/100 000PY) had substantially higher rates of TB disease than US-born persons (IR, 1.27/100 000PY; 95% CI, 1.09–1.44/100 000PY) or persons with unknown nativity (IR, 2.19/100 000PY; 95% CI, 1.93–2.45/100 000PY). TB disease incidence rates were higher for persons with high-risk (IR, 11.3/100 000PY; 95% CI, 10.0–12.6/100 000PY) medical conditions than for persons who had no high- or intermediate-risk medical conditions (IR, 2.63/100 000PY; 95% CI, 2.43–2.82/100 000PY).

The adjusted hazard ratio of TB disease was 15.3 (95% CI, 13.2–17.9) times higher for persons born in TB-endemic countries and 5.83 (95% CI, 4.46–7.63) times higher for non-US-born persons by language only compared with US-born persons (Table 2, Figure 1). Persons with HIV infection (aHR, 3.77; 95% CI, 2.7–3.89), diabetes (aHR, 2.85; 95% CI, 2.53–3.20), end-stage renal disease (aHR, 2.84; 95% CI, 2.07–3.89), and those who had received high-dose corticosteroids (aHR, 1.39; 95% CI, 1.10–1.77) or other immunosuppressants (aHR, 2.37; 95% CI, 1.73–3.24) had significantly increased risk of TB disease compared with persons without those conditions (Table 2, Figure 2).

Figure 1.

Cumulative hazard of TB disease by nativity, Kaiser Permanente Northern California, 2004–2022. Abbreviation: TB, tuberculosis.

Open in new tabDownload slide

Figure 2.

Cumulative hazard of TB disease by high- and intermediate-risk medical conditions for progression to TB disease, Kaiser Permanente Northern California, 2004–2022. Abbreviations: TB, tuberculosis; TNFα, tumor necrosis factor alpha.

Open in new tabDownload slide

Among persons born in TB-endemic countries, TB incidence rates were substantially higher among persons with high-risk (IR, 49.7/100 000PY; 95% CI, 42.8–56.6/100 000PY) and intermediate-risk (IR, 27.5/100 000PY; 95% CI, 24.9–30.1/100 000PY) medical conditions compared with persons without any high-risk or intermediate-risk conditions (IR, 11.4/100 000PY; 95% CI, 10.2–12.5/100 000PY) (Table 3, Figure 3). Similarly, among US-born persons, TB incidence rates were substantially higher among persons with high-risk (IR, 3.00/100 000PY; 95% CI, 2.20–3.81/100 000PY) and intermediate-risk (IR, 1.85/100 000PY; 95% CI, 1.42–2.27/100 000PY) medical conditions compared with persons without any high- or intermediate-risk conditions (IR, 0.910/100 000PY; 95% CI, 0.732–1.09/100 000PY) (Figure 4). Among persons born in TB-endemic countries, HIV infection (aHR, 4.73; 95% CI, 2.25–9.96), high-dose corticosteroid use (aHR, 1.53; 95% CI, 1.15–2.03), other immunosuppressant use (aHR, 2.67; 95% CI, 1.86–3.89), diabetes mellitus (aHR, 2.01; 95% CI, 1.74–2.31), and end-stage renal disease (aHR, 2.35; 95% CI, 1.66–3.33) were independent risk factors for TB disease. Among US-born persons, only HIV infection (aHR, 6.91, 95% CI, 2.56–18.7) and diabetes mellitus (aHR, 1.53; 95% CI, 1.09–2.15) were independent risk factors for TB disease. Persons born in TB-endemic countries accounted for the majority of all TB cases among persons with high-risk (200/253; 79.1%) or intermediate-risk medical conditions (424/495; 85.7%).

Figure 3.

Cumulative hazard of TB disease by high- and intermediate-risk conditions for progression to TB disease among persons born in TB-endemic countries, Kaiser Permanente Northern California, 2004–2022. Abbreviations: TB, tuberculosis; TNFα, tumor necrosis factor alpha.

Open in new tabDownload slide

Figure 4.

Cumulative hazard of TB disease by high- and intermediate-risk conditions for progression to TB disease among US-born persons, Kaiser Permanente Northern California, 2004–2022. Abbreviations: TB, tuberculosis; TNFα, tumor necrosis factor alpha.

Open in new tabDownload slide

Among the 584 801 persons born in TB-endemic countries, we assessed the incidence and risk of TB disease among persons from selected TB-endemic countries and geographic regions (Table 4). Persons born in Ethiopia (IR, 63.9/100 000PY; 95% CI, 34.0–109/100 000PY), Burma (IR, 48.2/100 000PY; 95% CI, 26.3–80.8/100 000PY), the Philippines (IR, 39.6/100 000PY; 95% CI, 35.9–43.7/100 000PY), and Vietnam (IR, 33.0/100 000PY; 95% CI, 27.1–39.8/100 000PY) had the highest incidence of TB disease. By region of birth, persons born in Africa and Eastern, South, or Southeast Asia had the highest incidence of TB disease, while persons born in the Caribbean, Central or Western Asia, or Eastern Europe had the lowest incidence of TB disease. Adjusted Cox models produced aHRs similar to these incidence rates.

DISCUSSION

This large US-based retrospective cohort study with >4.7 million persons and 35 million PY of follow-up adds to our understanding of the risk of incident TB disease in California in several ways. First, using available data on country of birth and preferred language in the EHR, we have identified populations at high risk (born in a TB-endemic country based on recorded country of birth or preferred language) and low risk (US-born; unknown country of birth) of incident TB disease. Second, we report that risk of TB disease is significantly higher in persons born in TB-endemic countries and among persons with medical conditions associated with progression to TB disease than among persons without those risk factors. Moreover, we report that >79% of all TB cases among persons with high- or intermediate-risk medical conditions for progression to TB disease occur among non-US-born persons at TB risk. Third, we report that risk of TB disease is substantially higher in persons born in Africa as well as East, South, and Southeast Asia.

LTBI screening guidelines recommend LTBI screening of persons who were born in, or are former residents of, countries with high TB disease incidence, but few US-based studies in a general health system have assessed the incidence of TB disease among persons born in TB-endemic countries [4–6]. In this analysis, we demonstrate that persons born in TB-endemic countries based on country of birth or preferred language in the EHR have significantly higher risk of TB disease than US-born persons and might have substantial benefit from screening and treatment for LTBI. As many health systems do not routinely collect data on country of birth, demonstrating that language as a proxy for country of birth is also associated with high risk of progression has practical utility for identifying persons who are at increased risk of progression to TB disease [18]. The absolute risk of TB disease was 13.88 times higher among persons born in TB-endemic countries compared with US-born persons, and this group accounted for 61% of all incident TB disease cases. However, risk of incident TB disease differed substantially by country as well as region, with persons from Eastern, Southeast, and South Asia and Africa having the highest rates of TB disease, as has been shown in other studies [19]. Our analysis supports current guidelines specifying that persons born in TB-endemic countries are at highest risk of TB disease and should be prioritized for LTBI screening and treatment, but it offers insights into differences among persons born in TB-endemic countries that could be used to prioritize particular subgroups for screening and treatment.

LTBI and disease-specific treatment guidelines recommend LTBI screening and treatment for persons with medical factors associated with progression to TB disease such as a medical condition (eg, HIV infection, solid organ transplantation, hematologic malignancy) or treatment (eg, receipt of TNFα inhibitors, corticosteroids, or other immunosuppressants) [5, 20]. In a stratified analysis, persons born in TB-endemic countries with medical risk factors for TB progression had significantly higher TB incidence than persons born in TB-endemic countries without medical risk factors, and >79% of all TB cases among those with medical risk factors were concentrated among persons born in TB-endemic countries. In addition, we highlight the increased risk of TB disease among persons with diabetes and end-stage renal disease among persons born in TB-endemic countries. Although the number of TB cases among US-born persons was small, US-born persons with high- or intermediate-risk medical conditions associated with progression to TB disease did not have significantly higher risk of TB disease compared with persons without medical risk factors, except among persons with HIV infection and diabetes. These data reinforce that birth in a TB-endemic country is the main risk factor for TB disease and TB disease risk is especially high among persons born in TB-endemic countries with high or intermediate risk of progression to TB disease. Unfortunately, even this very high-risk group is often not screened appropriately for LTBI [11].

Many rigorously conducted epidemiologic studies have suggested that the risk of incident TB disease is greatest in the first 2 years after primary infection and thus time since primary infection is a critical risk factor for incident TB disease [21]. Thus, proxies for time since primary infection such as time since initial TB exposure (eg, close contact with a TB case), immigration to the United States, or most recent travel to a TB-endemic country might be useful in defining overall TB risk [22]. However, these data are often not collected routinely within health systems in the United States and are often hard to ascertain in routine clinical practice. Our analysis suggests that for persons residing in a low-incidence setting such as California, the risk of TB disease appears to remain stable over an 18-year observation period and suggests that reactivation TB from remote infection is the primary driver of incident TB in California.

Our study has several limitations. We only included persons with TB disease who were diagnosed within our health system and thus might have missed some persons with incident TB disease. We did not assess all possible previously reported risk factors for TB disease, including travel to a TB-endemic country, close contact with a known TB case, homelessness, injection drug use, and incarceration. Data on these risk factors were not available in our EHR and thus could not be included in this analysis. In addition, we did not explore other possible risk factors for TB disease, and thus some residual confounding might remain. Lastly, we only assessed risk in persons who have not been previously treated for LTBI as treatment reduces the risk of progression to TB disease, and we did not explore differential access to treatment among the different risk groups.

In summary, this study provides a detailed report of risk of TB disease in a general health system and highlights the need to focus our prevention efforts on patients at highest risk of developing TB disease, persons born in TB-endemic countries, and especially those with medical factors associated with progression to TB disease.

Supplementary Data

Supplementary materials are available at Open Forum Infectious Diseases online. Consisting of data provided by the authors to benefit the reader, the posted materials are not copyedited and are the sole responsibility of the authors, so questions or comments should be addressed to the corresponding author.

Acknowledgments

The authors are grateful to all Kaiser Permanente members, without whom this study would not have been possible.

Author contributions. All contributed to conceptual ideas. All contributed to methodology. J.S. and Y.N. contributed to formal analysis. J.S. and Y.N. contributed to writing and original draft preparation. All authors contributed to writing, reviewing, and editing the manuscript. J.S. supervised the work.

Disclaimer. The content is solely the responsibility of the authors and does not necessarily represent the official views of the funders or of Kaiser Permanente.

Data sharing. Anonymized data that support the findings of this study may be made available from the investigative team in the following conditions: (1) agreement to collaborate with the study team on all publications, (2) provision of external funding for administrative and investigator time necessary for this collaboration, (3) demonstration that the external investigative team is qualified and has documented evidence of training for human subjects protections, and (4) agreement to abide by the terms outlined in data use agreements between institutions.

Patient consent. The study was approved by the Kaiser Permanente Southern California and the KPNC Institutional Review Boards with waivers of the requirement for informed consent.

Financial support. This work was supported by the National Institutes of Health (1 R01 AI151072-01 to S.Y.T., J.S.) and the Physician Researcher Program of The Permanente Medical Group Delivery Science and Applied Research Program (J.S.). The funders had no role in the study design, data collection, data analysis, interpretation, or writing of the report.

References

Author notes