Between April 2022 and April 2023, 185 patients diagnosed with rifampin-susceptible pulmonary TB through microbiological tests, including gene amplification techniques or both molecular and phenotypic drug susceptibility tests, were admitted to Masan National Tuberculosis Hospital (MNTH) in Changwon, South Korea. None of these patients had been diagnosed with CDI at the time of admission. Considering that the average duration from antibiotic exposure to CDI onset is approximately 13 days, 168 patients who stayed more than 2 weeks were ultimately enrolled in our study (Figure 1) [13]. These patients received standard treatment with first-line drugs, including isoniazid (INH) (300 to 400 mg once daily), rifampin (600 to 750 mg once daily), ethambutol (1,000 to 1,400 mg once daily), and pyrazinamide (1,250 to 1,750 mg once daily). Where rifampin administration was contraindicated due to adverse drug reactions, rifabutin (300 mg once daily) was used as an alternative. Drug doses were determined based on the patient’s body weight and therapeutic drug monitoring (TDM) results. Those displaying adverse drug reactions or having contraindications to standard therapy received additional treatment with quinolones and cycloserine. Newly diagnosed patients underwent a 6-month treatment regimen, while those with extensive lesions, as assessed by attending physicians, and those diagnosed with relapsed TB received treatment for 9 months. Patients with both relapsed TB and extensive lesions were treated for 12 months.

In our study, we employed the classification of diagnosis based on the history of previous TB treatment as described by the World Health Organization (WHO) in 2013 [14]. ‘New patients’ were defined as those never treated for TB or who have taken anti-TB drugs for less than 1 month. ‘Relapsed TB’ was considered as those with a recurrent episode of TB after being declared ‘cured’ or ‘treatment completed’ at the end of their most recent TB treatment. Patients who were lost during follow-up or who experienced treatment failure during their latest TB treatment were defined as ‘lost to follow-up’ and ‘treatment failure,’ respectively. Patients with a history of TB treatment whose outcome after their most recent course was unknown or undocumented were classified as ‘others.’

The definition of treatment outcomes introduced by WHO in 2021 was utilized in describing treatment outcomes [15]. Treatment success was defined as the sum of ‘cured’ and ‘treatment completed’ categories. Patients who died were classified as those who expired during TB treatment. Those who discontinued TB treatment due to adverse drug reactions were categorized as ‘lost to follow-up’ group. Patients transferred out for other medical reasons or whose outcomes were not reported in any hospital were categorized as ‘not evaluated.’

All attending physicians for each patient at our center systematically monitored stool patterns from the initiation of TB treatment. Diarrhea was defined as stool corresponding to Bristol stool scale type 7, and loose stools were defined as those corresponding to Bristol stool scale types 5 and 6 [16,17]. The VIDAS C. difficile Toxin A & B Assay (bioMérieux, Marcy-l'Étoile, France) was conducted in the patients clinically suspected of CDI with stool patterns characterized by Bristol stool scale types 5 to 7. The onset of CDI was recognized at the time when the toxin assay was performed. Isolation was promptly enforced for confirmed cases. TB treatment continued unchanged, regardless of CDI occurrence. An initial episode of CDI was treated with four daily doses of oral vancomycin 125 mg for 10 to 14 days. For recurrent episodes, treatment included tapered and pulsed dosing of vancomycin, beginning with four daily doses for the first 2 weeks, followed by bi-daily doses for the next week, daily doses for the subsequent week, doses every other day for the next week, and doses every 3 days during the final week.

Baseline characteristics of the enrolled patients were examined, including age, sex, body mass index (BMI) at the baseline visit, history of smoking and alcohol use, underlying diseases, the classification of diagnosis when initiating TB treatment, presence of miliary TB, TB pleurisy, extrapulmonary TB, or paradoxical responses, length of hospital stay, follow-up periods, types of anti-TB drugs, and treatment outcomes. The number of patients with INH-resistant TB was separately calculated. The length of hospital stay was defined as the duration from admission to discharge, and the follow-up period was defined as the interval from the initiation of TB treatment to the date of the last clinical visit, whether at discharge or during an outpatient visit. Additionally, other characteristics such as the use of other antibiotics for any combined infections during TB treatment and the presence of combined bacterial pneumonia at the baseline visit were also documented.

We specifically targeted patients suspected of CDI, surveying the detailed abdominal symptoms and signs they initially reported. Patients were asked five types of questions about whether they experienced only diarrhea, only nausea and vomiting, only abdominal pain, nausea with loose stool, or only fever. All these symptoms and signs were considered present if they persisted for at least 48 hours, regardless of frequency, due to inaccuracy in assessing the changes in bowel habits. Fever was defined as a tympanic temperature over 37.5°C. Several dichotomous variables were established for the analyses of risk factors and clinical outcomes. The old-aged group was defined as patients aged over 70.

BMI under 16 kg/m², hypoalbuminemia, and vitamin D deficiency were assessed at the baseline visit and included as nutritional factors indicating undernutrition. Body weight loss over 5% of usual body weight (UBW) was evaluated 1 month after admission and also considered a marker of undernutrition [18,19]. Hypoalbuminemia was defined as a serum albumin level less than 2.5 g/dL, and similarly, vitamin D deficiency was defined as less than 10.0 ng/mL [20-22].

The primary outcome was all-cause mortality during TB treatment, analyzed in all enrolled patients. The secondary outcomes included delayed conversion of AFB smear and AFB culture, defined as taking more than 8 weeks (56 days). Sputum assays were performed biweekly following admission. Successful conversion was regarded as consistent culture or smear negativity following the initial conversion until the end of follow-up or TB treatment. Negative conversion of AFB culture was defined as the initial simultaneous negative conversion of both solid and liquid media.

Three types of subgroup analyses were conducted. Subgroup A’s analysis involved 86 patients suspected of CDI undergoing C. difficile toxin assay to identify initially manifested clinical features of CDI among TB patients. Subgroup B’s analysis was on 114 AFB culture positive patients, and subgroup C’s on 100 AFB smear positive patients, to examine predictive factors for delayed conversion. Smear and culture positivity were initially evaluated at treatment initiation; culture positivity was defined as a positive result in either liquid or solid culture.

Descriptive data for baseline characteristics were calculated for all variables. Data normality was assessed by visually inspecting the distribution. Categorical variables were presented as numbers and proportion with analysis using either a chi-square test or Fisher’s exact test. Continuous data were reported as mean±standard deviation, depending on the data distribution, and were assessed using an independent t-test. Variance equality was evaluated with Levene’s test.

A chi-square test or Fisher’s exact test was performed for subgroup A analysis. Logistic regression analysis was used to evaluate risk factors for CDI occurrence across the entire patient cohort, estimating odds ratio (OR) and 95% confidence interval (CI) using both univariable and multivariable effects.

The analysis of the primary outcome was conducted using the Fine-Gray model, where CDI occurrence was treated as a time-varying covariate. Cumulative incidence functions were presented to illustrate the impact of CDI on all-cause mortality, as determined through Fine-Gray model analysis. Subgroup analyses B and C utilized the time dependent Cox-regression analysis, with CDI occurrence as a time-varying covariate, aligning with the primary outcome analysis. Hazard ratio (HR) and corresponding 95% CI were assessed using both the univariable and multivariable effects, and multivariable analyses were performed solely with factors that showed statistical significance in the univariable analysis.

All analyses were conducted utilizing R Studio 2024.04.0 (R Foundation for Statistical Computing, Vienna, Austria) and SAS version 9.4 (SAS Institute Inc., Cary, NC, USA). Statistical significance was assessed based on a p-value less than 0.05. The Institutional Review Board of MNTH (IRB-398837-2023-E03) approved this study. Written informed consent was obtained from each patient and guardian for the use of their data in statistical analyses, with no refusals reported.

The baseline characteristics of the 168 patients finally enrolled, categorized into CDI and non-CDI groups, are presented in Table 1. Likewise, the administered anti-TB drugs and treatment outcomes are outlined in Table 2. The mean age was 74.4±14.1 in the CDI group and 67.9±17.1 in the non-CDI group, highlighting a significant age difference between the groups (p=0.023). Comparisons of BMI, stratified at 16.0, 18.5, and 23.0 kg/m², showed no significant differences between the two groups (p=0.757). There were no significant differences observed for any anti-TB drugs, including quinolones, or other antimicrobial agents such as intravenous piperacillin/tazobactam, vancomycin, and meropenem.

In the CDI group, 52.2% of patients were successfully treated and 30.4% were classified as ‘not evaluated.’ Among the six deaths, five were due to worsening pneumonia and one was due to septic shock associated with CDI-related acute respiratory distress syndrome (ARDS). In the non-CDI group, 73.0% were successfully treated and 23.0% were classified as ‘not evaluated.’ Among the four deaths, three were due to worsening pneumonia and one was due to coronavirus disease 2019 (COVID-19) pneumonia. No deaths were attributed to pulmonary TB. A total of three patients discontinued TB treatment due to adverse reactions such as anorexia and cognitive impairment. There were no observed treatment failures in either group.

The cumulative incidence of CDI among TB patients was 46 (27.3%) out of 168 patients, with an incidence rate of 15.0 cases per 10,000 patient-days. The median time from the initiation of TB treatment to the occurrence of CDI was 57 days (interquartile range, 32.0 to 98.3). Twelve patients (26.0%) were identified as having recurrent CDI. Among them, one patient experienced a second recurrence, totaling two recurrences for that individual. There were no cases of fulminant colitis, and all CDI patients, except for one who died from CDI-related ARDS, were successfully treated according to the treatment protocol.

Among the 86 patients screened, 46 (53.4%) were diagnosed with CDI (toxin-positive group). In the toxin-positive group, the most common initial symptom was diarrhea only, reported in 69.6% of patients, which was more frequent than in the toxin-negative group. Nausea with loose stool and nausea and vomiting only were observed in 19.6% and 6.5% each. Isolated abdominal pain was reported only in this group, with a proportion of 4.3%. In the toxin-negative group, 42.5% of patients reported diarrhea only. Nausea with loose stool was observed in 20.0%, while 25.0% experienced nausea and vomiting only. Isolated fever was only reported in this group, with a proportion of 12.5%. The overall symptom distribution between the two groups was statistically significant (p=0.005), as shown in Table 3.

Risk factors for CDI occurrence are outlined in Table 4. In the univariable analysis, being over the age of 70 was identified as a significant factor, with an OR of 2.21 (95% CI, 1.08 to 4.49; p=0.029). Nutritional factors such as a BMI under 16 kg/m² and hypoalbuminemia also emerged as significant, demonstrating ORs of 2.23 (95% CI, 1.01 to 4.92; p=0.046) and 8.35 (95% CI, 2.11 to 33.07; p=0.002) respectively. Weight loss exceeding 5% of UBW a month after admission, vitamin D deficiency, underlying renal diseases, and the administration of quinolones including levofloxacin and moxifloxacin, as well as other antibiotics, were not significant. In the multivariable analysis, age over 70 and hypoalbuminemia were noted as significant factors, with ORs of 2.94 (95% CI, 1.27 to 6.78; p=0.011) and 8.35 (95% CI, 1.90 to 36.60; p=0.004), respectively.

The results of the analyses on the primary outcome are presented in Table 5 and Figure 2. In the unadjusted analysis, CDI emerged as a predictive factor with a HR of 4.63 (95% CI, 1.33 to 16.08; p=0.020). The age group over 70 also proved significant, exhibiting an HR of 8.37 (95% CI, 1.13 to 64.81; p=0.042). The presence of underlying respiratory diseases such as asthma, chronic obstructive pulmonary disease, or interstitial lung disease, as well as TB pleurisy and combined pneumonia at baseline, demonstrated HRs of 4.84 (95% CI, 1.25 to 18.71; p=0.022), 4.12 (95% CI, 1.16 to 14.60; p=0.028), and 4.63 (95% CI, 1.33 to 16.08; p=0.016), respectively. Nutritional factors such as a BMI under 16 kg/m², hypoalbuminemia, and vitamin D deficiency were not identified as predictive factors. In the adjusted analysis, CDI did not emerge as significant, while old age, underlying respiratory diseases, and combined pneumonia exhibited HRs of 10.43 (95% CI, 2.05 to 53.21; p=0.001), 4.33 (95% CI, 1.11 to 16.89; p=0.038), and 3.57 (95% CI, 1.00 to 12.73; p=0.053), respectively.

The results of the analyses on the secondary outcomes are detailed in Table 6. All factors evaluated using unadjusted time dependent Cox-regression analysis, including CDI, were determined to be non-significant; these factors include the older age group, undernutrition, underlying respiratory diseases, TB pleurisy, or combined pneumonia at the baseline visit, and paradoxical response.