Association of CKD with Incident Tuberculosis

Ethical Considerations

The institutional review board of Seoul National University Hospital (institutional review board no. E-1801–027–913) approved the study. The attending government organization approved the usage of the National Health Insurance Service (NHIS, no.2018-1-148). This study was conducted in accordance with the Declaration of Helsinki.

Study Setting

The study was a nationwide population-based cohort study reviewing the National Health Insurance Database (NHID), which was provided by the NHIS of the Republic of Korea (14). From the year 2010, Korea’s NHIS has provided additional insurance coverage for all patients diagnosed with active TB to decrease the burden of TB in the nation, and specific insurance codes were obligatorily applied to patients with active TB after confirmation of their diagnosis. Because the NHID includes complete information on insured medical services throughout the country, the claims database enabled us to review all patients with active TB in the nation by means of their unique insurance codes (15).

In Korea, similar to in other nations, dialysis dependency and transplantation are considered as the risk factors for active TB, but no additional recommendation for active TB surveillance was present for the predialysis CKD population. Patients with latent TB did not receive the specific insurance codes and were not considered in this study.

The general health screening program is run by the NHIS, which is provided for office workers/nonworkplace subscribers every 2 years and nonoffice workers every year. The proportion of people receiving this charge-free health screening was >70% in the year 2011 among a target population of approximately 15 million people. Because general health screening measures serum creatinine values and urine dipstick albuminuria every time, CKD stages were identifiable by reviewing the information. The registered centers that provide the general health screening are quality controlled and regulated by a government organization; however, their laboratory equipment manufacturers are not unified.

Study Population

We screened the adult (age ≥19 years old) health examinees who received two or more examinations from 2012 to 2016 including non-isotope dilution mass spectrometry serum creatinine measurements by the Jaffe kinetic method, because this method was used for the majority of serum creatinine measurements in the health screening dataset. Patients excluded on the basis of the criteria were (1) those with a previous TB history identified by the specific insurance or International Classifications of Diseases, tenth revision (ICD-10) diagnostic codes; (2) those with a history of kidney replacement therapy or kidney transplantation at baseline or before diagnosis of TB; and (3) those who had their serum creatinine measured by a method other than the Jaffe kinetic method. To include those who had stable CKD or CKD-free state, (4) those who had transient or fluctuating laboratory results indicative of CKD or laboratory findings indicative of newly developed CKD during the study period were excluded. Because we intended to collect a control group free from CKD, (5) those who had CKD-related ICD-10 diagnostic codes were also excluded from the control group (Supplemental Material) (16). The examples of the study population, the index inclusion date, the follow-up duration, and the excluded patients are shown in Supplemental Figure 1.

Study Groups

The study group included people with predialysis CKD. The presence of predialysis CKD was determined when evidence of CKD (i.e., the presence of dipstick albuminuria [1+ or more] or eGFR<60 ml/min per 1.73 m² calculated from the Modification of Diet in Renal Disease equation) was observed for two or more consecutive health examinations and until the last examination. Among those sequential health examinations indicative of CKD, the first examination date was the inclusion date for the predialysis CKD group when baseline information was collected and follow-up was initiated. To stratify the predialysis CKD stages determined by the Kidney Disease Improving Global Outcomes guidelines (17), we defined four subgroups with the eGFR and albuminuria values from the first included health examination. CKD stage 1 (eGFR≥90 ml/min per 1.73 m²) and stage 2 (eGFR≥60 and <90 ml/min per 1.73 m²) included individuals who had successive albuminuria but had no reduced eGFR at baseline. Health examinees with CKD stage 3 (eGFR≥30 and <60 ml/min per 1.73 m²) and stage 4 or 5 without dialysis (eGFR<30 ml/min per 1.73 m²) were also identified.

The control group was 1:1 matched from a population of health examinees who did not have any evidence of CKD (reduced eGFR or albuminuria) during all included health examinations, and the first examination date was the inclusion date. The matching variables included age, sex, low-income status, and smoking history. Because we intended to construct a matched control group that included an age-, sex-, and social factors–matched general population, comorbidities (e.g., hypertension and diabetes) and body mass index (BMI) variables were not matched, because that would have included people with substantial comorbidities in the control group. However, because hypertension, diabetes, BMI, and other comorbidity variables are still important potential confounders, the variables were adjusted in additional multivariable analyses.

Study Outcomes

The main study outcome was incident active TB during the follow-up, and the follow-up was performed until the outcome, death, or censoring on December 31, 2016. Respiratory (A15 and A16), nonrespiratory (A17 and A18), and miliary TB (A19) events were separately identified with repetitive identification of the corresponding ICD-10 diagnostic codes.

Next, we investigated prognosis after active TB diagnosis. Because direct causes of death were unavailable in the NHID, we implemented an operational definition for TB-associated mortality, which was defined as 1-year all-cause mortality from a TB diagnosis.

Data Collection

Demographic and clinical characteristics at the inclusion dates were collected as baseline variables. The details are shown in Supplemental Material.

Statistical Analyses

We presented continuous variables as medians (interquartile ranges) and categorical variables as numbers (percentages). We used the chi-squared test and the Mann–Whitney U test or Kruskal–Wallis test to investigate the differences between the study groups. The Kaplan–Meier survival curve was plotted, with the P value calculated from the log–rank method. The Cox regression analyses were performed to investigate the risk of active TB within the study groups. The main multivariable model included matched variables (age, sex, low-income status, and smoking history) and other unmatched variables (place of residence, history of diabetes mellitus, hypertension, cancer, chronic obstructive pulmonary disease, immunosuppressants usage, and baseline BMI). After an interaction term analysis to confirm whether a significant difference of active TB risk was present according to CKD stages, risk of active TB of each CKD stage subgroup was compared with that of the matched controls. The risk of 1-year mortality was investigated with the same method as the multivariable Cox regression analysis above for those who were diagnosed with active TB. The risk factors for active TB among the predialysis CKD group were analyzed with a multivariable Cox regression analysis by using the backward elimination method. No missing values were present in the final group of studied people, because we excluded the patients with missing variables. The SAS 9.4 program (SAS Institute) was used to perform the statistical analysis. Two-sided P values of <0.05 indicated a statistically significant difference.