Risk of Tuberculosis Disease in People With Chronic Kidney Disease Without Kidney Failure: A Systematic Review and Meta-analysis

Abstract

Background

Kidney failure is an established risk factor for tuberculosis (TB), but little is known about TB risk in people with chronic kidney disease (CKD) who have not initiated kidney replacement therapy (CKD without kidney failure). Our primary objective was to estimate the pooled relative risk of TB disease in people with CKD stages 3–5 without kidney failure compared with people without CKD. Our secondary objectives were to estimate the pooled relative risk of TB disease for all stages of CKD without kidney failure (stages 1–5) and by each CKD stage.

Methods

This review was prospectively registered (PROSPERO CRD42022342499). We systematically searched MEDLINE, Embase, and Cochrane databases for studies published between 1970 and 2022. We included original observational research estimating TB risk among people with CKD without kidney failure. Random-effects meta-analysis was performed to obtain the pooled relative risk.

Results

Of the 6915 unique articles identified, data from 5 studies were included. The estimated pooled risk of TB was 57% higher in people with CKD stages 3–5 than in people without CKD (adjusted hazard ratio: 1.57; 95% CI: 1.22−2.03; I² = 88%). When stratified by CKD stage, the pooled rate of TB was highest in stages 4–5 (incidence rate ratio: 3.63; 95% CI: 2.25–5.86; I² = 89%).

Conclusions

People with CKD without kidney failure have an increased relative risk of TB. Further research and modeling are required to understand the risks, benefits, and CKD cutoffs for screening people for TB with CKD prior to kidney replacement therapy.

This systematic review and meta-analysis demonstrates that people with chronic kidney disease without kidney failure have an elevated risk of tuberculosis versus people without kidney disease. Tuberculosis risk increases as kidney disease progressed and was highest in stages 4–5.

Chronic kidney disease (CKD) is a growing global health issue with an estimated prevalence of 10% worldwide [1]. CKD is associated with altered cellular immunity due to aging, hypoalbuminemia, malnutrition, uremia, vitamin D deficiency, and medical immunosuppression [2]. Conditions causing immune compromise, such as diabetes mellitus and human immunodeficiency virus (HIV), are more common in people with CKD [3–6]. Thus, people with CKD are at increased risk of infections [7–10], with the incidence of some infections increasing as CKD progresses [11]. People with CKD are also more often exposed to pathogens during frequent medical visits [12]. Both immune compromise and increased opportunity for infectious exposures appear to make people with CKD more susceptible to communicable diseases, including tuberculosis (TB) disease [13].

People with CKD requiring kidney replacement therapy are known to be at high risk for TB disease. A 2015 systematic review by Al-Efraij et al [14] demonstrated that the risk of TB disease was elevated in both kidney transplant and dialysis populations. Less is known about the risk of TB in people with CKD who do not require kidney replacement therapy (CKD without kidney failure). The above systematic review found no studies examining TB risk in people with CKD without kidney failure; however, several studies on this topic have since been published.

As the number of people with CKD globally continues to rise [1], a clearer understanding of TB risk in people with CKD without kidney failure is necessary to develop effective TB screening and prevention strategies. We, therefore, systematically reviewed the literature to improve our understanding of the association between CKD without kidney failure and TB disease.

Our primary objective was to estimate the pooled risk of TB disease in individuals with CKD stages 3–5 without kidney failure compared with people without CKD. Our secondary objectives were to describe the pooled risk of TB for all stages of CKD without kidney failure and by CKD stage.

METHODS

This study followed the Preferred Reporting Items for Systematic Review and Meta-Analyses (PRISMA) guidelines [15] and was prospectively registered (PROSPERO: CRD42022342499).

Identification of Studies

MEDLINE, Embase, the Cochrane Database of Systematic Reviews, and the Cochrane Central Register of Controlled Trials were systematically searched from 1 January 1970 to 4 July 2022. The full search strategy is shown in Supplementary Table 1.

Inclusion and Exclusion Criteria

Studies that met the following criteria were included: original research with a cohort or case-control study design, included humans of all ages with established CKD without kidney failure defined by Kidney Disease: Improved Global Outcomes (KDIGO) glomerular filtration rate (GFR) categories/stages G1–G5 or albuminuria categories/stages A1–3 with concurrent or subsequent assessments for TB disease through microbiologic diagnosis or clinical diagnosis; included an outcome estimate of TB disease in people with CKD without kidney failure compared with people without CKD, and included a comparison group without CKD accounting for least 1 demographic/medical risk factor either by matching or adjusting during statistical analysis. We defined a clinical diagnosis of TB to be diagnosis by a physician, International Classification of Diseases (ICD) code [16], or by medical database. Studies were excluded if they reported duplicate data; had a total population of less than 50 people with TB disease; used a cross-sectional, case series, or case report study design; or were unavailable in English.

Titles and abstracts were screened by 2 authors (P. L. and T. H.). Relevant studies underwent full-text screening. Any disagreements were settled by a third reviewer (J. C. J.). When articles reported duplicate data, the study with the most complete outcome reporting was included. A manual search of reference lists of selected articles was performed to identify other relevant studies.

Data Extraction

The items for data extraction were predefined. Variables were extracted and coded by P. L. and verified by T. H. We extracted the first author's surname, publication year, study location, source of TB and CKD diagnoses, and follow-up duration. We also extracted population characteristics, TB and CKD diagnoses' details, outcome measures and 95% confidence intervals (CIs), and the adjustment variables.

Quality Assessment

Two reviewers (P. L. and T. H.) independently assessed the quality of individual studies using a predetermined modified Newcastle-Ottawa Scale (NOS) [17]. Using this scale, stars were awarded based on the selection of study groups (0–4 stars), comparability of cohorts based on study design or analysis (0–2 stars), and ascertainment of the outcome of interest (0–3 stars). Complete details of the modified NOS assessment criteria are presented in the Supplemental Materials.

Outcomes

Our primary outcome was the relative risk of TB in people with CKD stages 3–5 without failure compared with people without CKD. Secondary outcomes included the relative risk of TB in people with CKD without kidney failure stages 1–5 and the relative risk by each CKD stage.

Statistical Analysis

For our primary analysis, we calculated the pooled adjusted hazard ratios (aHRs) of TB disease in individuals with CKD stages 3–5 without kidney failure compared with those without CKD. If an aHR was unavailable, we included an incidence rate ratio (IRR) [18]. A sensitivity analysis was performed excluding the IRR from the pooled outcome. We also conducted 2 supplementary analyses. First, we restricted our population to individuals with CKD stages 3–5 who had culture-confirmed TB. Then, we calculated the pooled unadjusted HR of TB disease in individuals with CKD stages 3–5 without kidney failure compared with people without CKD.

For our secondary analysis, we calculated the pooled aHR of TB disease in individuals with CKD stages 1–5 compared with people without CKD. If an aHR was unavailable, we included an IRR and sensitivity analyses were performed excluding the IRR. We also conducted a supplementary analysis calculating the pooled unadjusted HR of TB disease in individuals with CKD stages 1–5 compared with people without CKD.

Next, to assess the effect of CKD stage on TB disease, we stratified CKD stages in 3 categories: 1–2, 3, and 4–5, due to limitations in data reporting from included studies. For this analysis, we calculated the pooled IRRs of TB at each CKD stage compared with people without CKD. IRRs were used due to limited reporting of aHRs by stage. We also conducted 3 additional supplementary analyses. First, we calculated the pooled unadjusted HRs for each stage. Next, we compared the IRRs between each CKD category [19]. Last, we calculated the the incidence rate differences for TB in people with CKD without failure compared with those without CKD.

We only included studies without overlapping populations for our meta-analyses. If studies had overlapping populations, we included the study with the highest NOS score. If the NOS score was equal, we included the study with the largest sample population. When a study offered more than 1 adjusted outcome, we included the most-adjusted outcome. Once extracted, estimates were log-transformed, and the 95% CIs were used to calculate corresponding standard errors; data were then pooled using generic inverse variance random-effects models. Given the significant diversity and heterogeneity in observational studies, a random-effects model was chosen [20, 21]. The pooled log-effect estimates were then back-transformed for interpretation. We estimated the proportion of total variability due to between-study heterogeneity using I² [22, 23] and classified it as low level (<25%), moderate level (25–49%), substantial level (50–74%), or high level (>75%). All analyses were conducted in R (version 4.2.1; R Foundation for Statistical Computing) [24].

RESULTS

Literature Search and Study Characteristics

We identified 6915 unique articles in our search, of which 40 qualified for full-text review (Supplementary Figure 1). Our systematic review included 6 studies (Table 1): 5 retrospective cohort studies [25–29] and 1 prospective cohort study [30]. Five studies were included in the meta-analysis [26–30]. Cheng et al [25] was excluded from the meta-analysis due to a possible overlapping population with studies from Taiwan [28, 30]. All articles originated from high-income countries. The proportion of the study population who were immigrants was not reported, except for Yan et al [29], in which the population was exclusively immigrants. Three studies reported the relative risk of TB according to each CKD GFR stage [26, 28, 30]. None reported the risk by albuminuria category. Four studies described individuals with TB disease from national or regional databases [25, 26, 29, 30] and 2 studies from medical records [27, 28].

Table 1.

Characteristics of Included Studies for Systematic Review of Chronic Kidney Disease Without Failure and Tuberculosis

Study	Study Design	Country, WHO Region	Total No. of Participants	No. of Participants With TB	Source of CKD Diagnosis	Source of TB Diagnosis	Duration of Follow-up	Outcome Measure	Adjustment Variables
Cheng et al, 2018 [25]	Retrospective cohort	Taiwan, WPR	331 001	1028	NHI program research database of Taiwan (2000–2012)	NHI program research database of Taiwan (2000–2013)	Up to 13 y, mean 4.9 y in CKD and 6 y in controls	aHR	Age, sex, COPD, DM, pneumoconiosis
Cho et al, 2019 [30]	Prospective cohort	Taiwan, WPR	100 058	472	Community-based health screening program in northern Taiwan (2005–2008)	National TB Registry of the Taiwan CDC (2005–2013)	Up to 8 y, median 7.5 y	aHR	Age, sex, smoking, alcohol use, betel nut use, educational level, fasting glucose, BMI, corticosteroid use
Park et al, 2019 [26]	Retrospective cohort	Republic of Korea, WPR	817 746	3222	NHI Database of Korea (2012–2016)	NHI Database of Korea (2012–2016)	Up to 4 y, median 3 y	aHR	Age, sex, low-income status, smoking, place of residence (rural/urban), DM, HTN, cancer, COPD, immunosuppressant usage history, BMI
Ruzangi et al, 2020 [27]	Retrospective cohort	United Kingdom, EUR	477 026	246	UK Clinical Practice Research Datalink (2004–2014)	UK Clinical Practice Research Datalink (2004–2014)	Up to 10 y, median 3.81 y	aIRR	Age, gender, ethnicity, deprivation index, DM, COPD
Shu et al, 2020 [28]	Retrospective cohort	Taiwan, WPR	287 962	1000	National Taiwan University Hospital (2008–2013)	National Taiwan University Hospital (2008–2016)	Up to 8 y, mean 4.13 y	HR	Unadjusted
Yan et al, 2022 [29]	Retrospective cohort	Canada, AMR	714 717	1382	PROMIS registry, hospital records, and medical billing codes (1996–2015)	BC CDC TB registry (1996–2015)	Up to 19 y, median 9.4 y	aHR	Age at immigration, TB incidence in country of birth, sex, time since immigration, contact status, HIV, DM, cancer, solid-organ transplantation, immunosuppressive drugs, Gaussian frailty term for clustering by country of origin

Abbreviations: aHR, adjusted hazard ratio; aIRR, adjusted incidence rate ratio; AMR, Region of the Americas; BC, British Columbia; BMI, body mass index; CDC, Centers for Disease Control and Prevention; CKD, chronic kidney disease; COPD, chronic obstructive pulmonary disease; DM, diabetes mellitus; EUR, European Region; HIV, human immunodeficiency virus; HR, unadjusted hazard ratio; HTN, hypertension; NHI, National Health Insurance; PROMIS, Patient Records and Outcome Management Information System; TB, tuberculosis; WHO, World Health Organization; WPR, Western Pacific Region.

Further details about population characteristics, CKD diagnosis and stages, and TB diagnosis are summarized in Supplementary Table 2, Supplementary Table 3, and Supplementary Table 4, respectively.

Quality Assessment

The quality of the studies ranged from moderate to high (Supplementary Table 5). All were found to have representative CKD cohorts with a non-CKD comparison group from the same population. The diagnosis of CKD was made by GFR estimation or diagnostic codes in all studies, and only 1 study did not exclude individuals with a history of kidney replacement therapy [30]. One study did not have adequate controls or adjustment variables [28], and another did not comment on loss to follow-up [27].

Tuberculosis Risk in CKD Stages 3–5 Without Failure

Four studies were included in our primary analysis to calculate the pooled aHR of TB disease in individuals with CKD stages 3–5 without kidney failure compared with people without CKD [26, 27, 29, 30]. The estimated pooled risk of TB was 57% higher in people with CKD stages 3–5 than in those without CKD (aHR: 1.57; 95% CI: 1.22–2.03; I² = 88%) (Figure 1). Excluding the study reporting an IRR yielded similar results (aHR: 1.61; 95% CI: 1.18–2.20; I² = 90%) (Supplementary Figure 2). When restricted to individuals with culture-confirmed TB, the aHR was 1.81 (95% CI: 1.15–2.84; I² = 85%) (Supplementary Figure 3).

Figure 1.

Meta-analysis of the aHR for TB disease among people with CKD without kidney failure stages 3–5 compared with people without CKD. Abbreviations: aHR, adjusted hazard ratio; aIRR, adjusted incidence rate ratio; CI, confidence interval; CKD, chronic kidney disease; TB, tuberculosis. *Ruzangi et al [27] used an aIRR.

The pooled unadjusted HR of TB in individuals with CKD stages 3–5 without kidney failure compared with people without CKD was 2.33 (95% CI: 1.70–3.19; I² = 92%) (Supplementary Figure 4). The incidence rate differences are reported in Supplementary Table 6.

Tuberculosis Risk in All Stages of CKD Without Kidney Failure

Four studies were included in our secondary analysis to calculate the pooled aHR of TB disease in individuals with all stages of CKD without kidney failure (stages 1–5, when available) compared with people without CKD [26, 27, 29, 30]. Of note, Shu et al [28] included stage 1 CKD in their comparison group. The estimated pooled risk of TB was 55% higher in people with all CKD stages without failure than in those without CKD (aHR: 1.55; 95% CI: 1.06–2.27; I² = 88%) (Supplementary Figure 5). Sensitivity analysis excluding the IRR (aHR: 1.61; 95% CI: .94–2.78; I² = 92%) (Supplementary Figure 6) and the pooled unadjusted HR (HR: 1.69; 95% CI: 1.31–2.18; I² = 95%) (Supplementary Figure 7) resulted in similar findings.

Tuberculosis Risk by CKD Stage

Three studies were included in our secondary analysis to calculate the pooled IRR of TB by CKD stage [26, 28, 30]. The relative rate of TB was lowest in CKD stages 1–2 (IRR: 1.09; 95% CI: .81–1.48; I² = 85%), with the point estimate increasing in stage 3 (IRR: 2.5; 95% CI: 1.41–4.42; I² = 98%) and stages 4–5 (IRR: 3.63; 95% CI: 2.25–5.86; I² = 89%) (Figure 2). The pooled unadjusted HR of TB by CKD stage demonstrated similar results (Supplementary Figure 8).

Figure 2.

Meta-analysis of the incident rate ratios for TB disease among people with CKD without kidney failure by CKD stage compared to people without CKD. Abbreviations: CI, confidence interval; CKD, chronic kidney disease; IRR, incidence rate ratio; TB, tuberculosis. *Shu et al [28] included stage 1 CKD in their comparison group.

The IRR of TB was significantly greater between CKD stages 4–5 than CKD stages 1–2 (difference of IRR: 2.54; 95% CI: .37–1.5), whereas the differences between the IRR of TB disease between CKD stages 1–2 compared with stage 3 (difference of IRR: 1.4; 95% CI: −.31, .98) and CKD stages 3 versus 4–5 (difference of IRR: 1.14; 95% CI: −.61, .87) were not significant. The incidence rate differences by CKD stage are reported in Supplementary Table 7.

DISCUSSION

This systematic review and meta-analysis found that people with CKD stages 3–5 without failure have an increased relative risk of TB disease when compared with people without CKD. Elevated risk was also present when pooling CKD stages 1–5. Findings were consistent in the sensitivity analyses. In our analysis by CKD stage, the point estimate for the relative rate of TB increased as CKD stage progressed and was highest in CKD stages 4–5. The level of heterogeneity was high across all pooled risk estimates and in subgroup and sensitivity analyses.

To our knowledge, this is the first systematic review and meta-analysis to publish pooled risk estimates in people with CKD without kidney failure. A prior systematic review found that people receiving dialysis had at least a 3-fold higher rate of TB than the general population [14]. Compared with people with kidney failure, the point estimate of TB risk in people with CKD stages 3–5 appears to be less, at approximately 1.5 times that of people without CKD. The risk of TB in people with CKD without failure is likely driven by similar factors as the general CKD population, such as aging and immune compromise [2–6]. People with CKD without failure also have frequent medical visits where they could be exposed to TB, but likely fewer exposures than those receiving dialysis [12].

The World Health Organization recommends that people initiating chronic dialysis and preparing for kidney transplantation be screened for TB infection [31]. In contrast, no clear guidance exists for TB screening in people with CKD without kidney failure. Our findings suggest that CKD stages 3–5 without kidney failure is a significant risk factor for developing TB. However, in our analysis by CKD stage, there is no clear cutoff in the progression of kidney impairment where the risk of TB increases substantially, making it challenging to determine the optimal timing to screen for TB infection in people with CKD without kidney failure.

Because the number of people with CKD without kidney failure far outnumbers those with kidney failure, screening this population would require a substantial increase in resources. Further research is needed to determine the benefit of screening in different populations with CKD without kidney failure. Based on our findings, we suggest that clinicians maintain a high degree of suspicion for TB disease in people with CKD without kidney failure but feel there is insufficient evidence to recommend systematic screening for TB infection in all CKD populations.

The main strength of this study was the robust methodology, which included a detailed study protocol and thorough search strategy. This allowed us to provide pooled relative risk estimates for both CKD without kidney failure overall and by CKD stage. Additionally, we communicated with study authors to determine important points of potential bias.

Limitations of our study are primarily related to the use of published, observational data. First, we included both prospective and retrospective studies, which may affect the validity of our results. Second, the increased risk of TB disease in the population with CKD without kidney failure may be confounded by unmeasured variables. For example, included studies did not uniformly control for factors that may influence TB or CKD risk [31]. While we extracted data that adjusted for factors such as diabetes mellitus, HIV, and age, low sample sizes precluded stratified analyses. Studies also did not account for hospitalization or medical visits, which are known to increase risk of infectious exposures [12]. Next, the majority of studies did not report immigration status. The unknown proportion of immigrants in most studies in our analyses is a key limitation, because this population is likely to have overlapping risk factors for both CKD and TB [13].

Although all studies excluded people with a history of TB disease, none stated TB-infection status and only 1 study explicitly stated that people with a history of TB infection were excluded [29]. The unknown TB-infection status is a potential confounder as TB infection has been reported to be more prevalent in people with CKD [32] and increases the risk of progressing to TB disease. Three of the 6 studies in our systematic review did not state the location of TB [27–29] and 1 study excluded extrapulmonary TB disease [25]. Extrapulmonary TB occurs more frequently in people with CKD receiving dialysis [33]. If the same is true for the population with CKD without failure, this may have caused an underestimation of the risk of TB in our analysis. Additionally, Shu et al [28] did not exclude people with a history of kidney replacement therapy. However, from correspondence with the authors, the probability that people with a history of kidney replacement therapy were enrolled was extremely low as the cohort was derived from a voluntary community-based health screening service with healthy participants at baseline.

Next, heterogeneity was high across all analyses, potentially influencing the accuracy of our pooled estimates. We were unable to explain the high heterogeneity; thus, our results should be interpreted with caution. However, since there were no inconsistencies in the direction of effect, we determined that pooling through a random-effects model was appropriate [34]. In our study, the high heterogeneity may represent true variation in TB risk, but may also result from differences in methods, such as diagnostic criteria and study quality.

Fourth, the studies included in our secondary analysis did not all provide adjusted risk estimates for the relationship between CKD and TB. For this reason, we could not calculate an adjusted pooled outcome for TB risk by individual CKD stage. Due to limitations in data reporting, we instead pooled incidence ratio estimates by ranges of CKD stages. The pooled estimates using IRRs were notably higher than the corresponding estimates using aHRs, which is likely due to confounding and censoring.

Last, the generalizability of our findings is limited because our analyses were limited to studies primarily from high-income, low-TB-incidence countries. As CKD is a growing health issue in low- and middle-income countries, it will be important to study the risk of TB in CKD without failure in populations with a higher TB prevalence [13].

In summary, our results demonstrate that people with CKD without kidney failure have an elevated risk of TB disease compared with those without CKD, and the relative rate of TB increases as CKD stage progressed. However, there was no clear cutoff where TB risk increases during the progression of CKD, making it challenging to determine when and if this population should be systematically screened for TB infection. Prospective studies and cost-effectiveness modeling are required to guide TB-infection screening policy in people with CKD.

Supplementary Data

Supplementary materials are available at Clinical 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.

Notes

Author contributions. J. C. J. initiated the project. J. C. J., E. G. C., and P. L. were responsible for the design of the protocol. P. L. and T. H. conducted the literature search, extracted the data, and assessed the quality of the studies. O. A. A. and K. R. analyzed the data. J. C. J., E. G. C., V. J. C., O. A. A., K. R., and P. L. interpreted the data. P. L. and T. H. wrote the initial draft of the manuscript. J. C. J., E. G. C., V. J. C., O. A. A., and K. R. were responsible for critical revisions of the manuscript. All authors approved the final version submitted for publication.

Acknowledgments . The authors gratefully acknowledge the guidance of Dean Giustini, Health Sciences librarian, University of British Columbia (UBC), who helped design the search strategy.

Disclaimer. The researchers were independent of their sources of support, which had no role in this study.

Financial support. K. R. is supported by the Canadian Institutes for Health Research (CIHR) Frederick Banting and Charles Best Doctoral Award (2020–2023). J. C. J. is supported by a Michael Smith Foundation for Health Research Scholar Award and CIHR (No. PJT- 153213).