Incidence and Predictors of QTC Prolongation Among Drug-Resistant Tuberculosis Patients Receiving Bedaquiline-Based TB Regimens in Uganda: A Countrywide Retrospective Cohort Study Authors: Marble Nasasira, Kevin Ouma Ojiambo, Daniel Iraguha, Joseph Baruch Baluku Published: 2026 ABSTRACT AbstractBackgroundBedaquiline improves culture conversion in drug-resistant tuberculosis (DR-TB) treatment; its main safety concern is QTc prolongation, particularly when combined with other anti-TB drugs. This study assessed the incidence and predictors of QTc prolongation among DR-TB patients receiving bedaquiline under programmatic conditions in Uganda.MethodsWe conducted a retrospective cohort study using medical records from 14 DR-TB treatment sites across Uganda (January 2017–December 2022). Data on demographics, clinical characteristics, bedaquiline-based regimens, treatment duration, and QTc intervals were collected. Incidence rates with 95% CIs were calculated for QTc prolongation, while modified Poisson regression was used to identify predictors of QTc prolongation.ResultsData were abstracted from 532 patients, 70.9% of whom were male. The mean (SD) age was 41.4 (15.9) years, and 34.6% were coinfected with HIV. The baseline mean (SD) QTc was 426.8 (28.4) ms, with a median follow-up (interquartile range) of 10.3 (7.2–13.5) months. The overall incidence of QTc prolongation was 12.0 cases per 1000 person-months (95% CI, 10.8–16.8). Lower body mass index (BMI; incidence rate ratio [IRR], 0.924; 95% CI, 0.860; IRR, 0.993; P = .031) increased risk of QTc prolongation. Treatment success was similar among patients with and without QTc prolongation: 73.9% (51/69) vs 68.3% (316/463); P = .873.ConclusionsAn average of 12 new cases of QTc prolongation was observed for every 1000 person-months of accumulated observation time. The recommended monthly electrocardiogram monitoring should be emphasized among patients with low baseline BMI. QTc prolongation was not linked to poor treatment outcomes. The World Health Organization (WHO) estimates that the burden of drug-resistant tuberculosis (DR-TB) increased between 2020 and 2021, with 450 000 new cases of rifampicin-resistant TB (RR-TB) in 2021 [1]. Globally, there have been improvements in the treatment success rate for multidrug resistance (MDR)/RR-TB, which was reported at 60% in 2019, reflecting steady improvements from 50% in 2012 [1]. In Uganda, based on the recent DR-TB demographic survey, the overall incidence was 3.8/100 000, and the overall population incidence of RR-TB increased by 20% over the evaluation period (2014–2018). RR increased in new TB cases by 35% and in previously treated cases by 7% [2]. The WHO recommended use of all-oral regimens for MDR/RR-TB and pre-extensively drug-resistant tuberculosis (pre-XDR-TB), which reduces treatment duration to only 6 months [3]. Uganda is among the 124 countries that adopted the 2019 guidelines, which recommended a shorter all-oral bedaquiline (BDQ)- containing regimen of 9–12 months’ duration in eligible patients with confirmed MDR/RR-TB who have not been exposed to treatment with second-line TB medicines for more than a month and in whom resistance to fluoroquinolones has been excluded [4, 5]. In Uganda, patients with MDR-TB are managed through a mixed-model approach where they are hospitalized for the first 2–8 weeks before transitioning to outpatient care via directly observed therapy at a local health facility [5]. BDQ has a clinical efficacy with faster and lasting culture conversions for the first 24 weeks [6] and is therefore recommended as a class A drug for treating DR-TB. It is hepatically metabolized by the cytochrome p450 isoenzyme 3A (CYP3A) to its active M2 metabolite, which has reduced antimycobacterial activity [7]; however, M2 is associated with QTc prolongation because it blocks the hERG potassium channels in the heart. The heart's electrical cycle is measured on an electrocardiogram (ECG), and the QT interval is a measure of the time between the start of the Q wave and the end of the T wave, representing the electrical depolarization and repolarization of the left and right ventricles. A prolonged QT interval is a biomarker for ventricular tachyarrhythmia, which could progress to torsades de pointes and sudden death [8]. Increased QTc prolongation was observed when BDQ was co-administered with other drugs (fluoroquinolones and clofazimine) that prolong the QTc interval [9]; these drugs have also been repurposed to be used as preferred drugs in DR-TB management [10]. Therefore, WHO and Ugandan guidelines recommend an ECG at baseline (before treatment) and during follow-up: at 2 weeks and monthly until the DR-TB treatment is completed, while on BDQ [3]. However, a retrospective cohort study in Indonesia among 73 patients concluded that there were no significant differences on QTc prolongation associated with BDQ [11]. Therefore, more real-world data are needed to guide programmatic implementation of BDQ regimens, particularly in low-resource settings where ECG measurements are not readily available. Uganda has 17 treatment sites delivering DR-TB treatment, and each is required to have an ECG machine and expertise to read and interpret the ECG. This technical capacity is not always available at these facilities, imposes additional costs, and may act as a barrier to BDQ rollout. The incidence and predictors of QTc prolongation among patients on BDQ are not well documented in Uganda, and therefore the justification for universal ECG monitoring is unclear. This study determined the incidence of QTc prolongation for a large cohort of patients receiving BDQ in combination with other DR-TB agents under programmatic conditions in Uganda. METHODS Study Design and Setting This was a countrywide retrospective cohort study among DR-TB patients who were initiated on a BDQ-based regimen from January 2017 to December 2022 from 14 DR-TB treatment sites in Uganda. The study utilized secondary data from the DR-TB registers and medical charts of patients, which included 1 urban national referral hospital, 11 rural regional referral hospitals, and 2 rural general district hospitals. The DR-TB treatment centers are as follows: Mulago National Referral Hospital, Arua Regional Referral Hospital, Jinja Regional Referral Hospital, Iganga District Hospital, Mbale Regional Referral Hospital, Soroti Regional Referral Hospital, Lira Regional Referral Hospital, Matany District Hospital, Moroto Regional Referral Hospital, Mbarara Regional Referral Hospital, Kabale Regional Referral Hospital, Hoima Regional Referral Hospital, Fortportal Regional Referral Hospital, and Mubende Regional Referral Hospital. Study Population and Procedures All patients initiated on a BDQ-based regimen, regardless of their background DR-TB regimen, with laboratory-confirmed DR-TB between January 2017 and December 2022 were included in the study. DR-TB was diagnosed using Xpert MTB/RIF assay or conventional Mycobacterium culture on sputum samples. The common regimen consisted of 9–12 months of treatment with bedaquiline, clofazimine, linezolid, levofloxacin, and cycloserine (BDQ-CFZ-LZD-LFX -Cs) [5]. Patients were excluded if they did not have a baseline ECG recorded before BDQ initiation and/or if they did not have at least 1 follow-up ECG in the months of BDQ treatment, had a missing medical chart, and had no DR-TB confirmation record. This was a nationwide census study that considered all eligible patients initiated on a BDQ-based regimen in the 14 DR-TB treatment sites out of the 17 DR-TB treatment sites from 2017–2022 in Uganda. We did not include data from 2 regional referral hospitals (Gulu Regional Referral Hospital and Masaka Regional Referral Hospital) and 1 district hospital (Kitgum District Hospital). This is because these sites were missing a baseline or follow-up ECG in the patient file. Data were retrieved using a data abstraction form from patients’ registers and patients’ charts by principal investigator assisted by trained research assistants. Data were checked for any inconsistencies, coding errors, out-of-range values, completeness, accuracy, clarity, and missing values, and appropriate corrections were made on a daily basis by the principal investigator. In case of inconsistencies such as treatment start date, we considered the 1 documented during directly observed therapy by the treatment partner and not that observed by the health worker. Laboratory values such as renal function tests (RFTs), liver function tests (LFTs), and DR-TB classification were verified using laboratory results as source documents. Study Variables and Measurements Data collected included sociodemographic and clinical characteristics: age, sex, category of DR-TB (new, relapse, treatment failure, and loss to follow-up), and drug sensitivity test profiles. MDR/RR-TB was defined as TB caused by Mycobacterium tuberculosis strains that are resistant to at least rifampicin and isoniazid, pre-XDR-TB was defined as MDR- TB plus resistance to at-least 1 fluoroquinolone (either levofloxacin or moxifloxacin), and extensively DR-TB was defined as MDR-TB plus resistance to at least 1 fluoroquinolone and at least 1 other “Group A” drug (bedaquiline or linezolid) [4, 12]. We computed and reported the participants’ body mass index (BMI) as weight in kg divided by height in m2. Other data collected were the history of smoking, history of alcohol use, and clinical information: HIV status, antiretroviral therapy (ART) regimen, other concomitant medication, and other comorbidities such as diabetes mellitus, hypertension, asthma or chronic obstructive pulmonary disease (COPD), cardiac morbidity, and malignancies. Laboratory data included baseline hemoglobin, aspartate aminotransferase (AST), alanine aminotransferase (ALT), alkaline phosphatase (ALP), gamma-glutamyl transferase (GGT) creatinine, sodium, potassium, and bilirubin. A patient was considered to have hepatic impairment if ALT >55 U/L or AST >48 U/L or GGT >61 U/L or ALP >129 U/L, or albumin <3.5 umol/L. A patient was considered to have renal impairment if serum creatine >1.25 mg/dL. Patients were prescribed the longer MDR regimen (15–20 months) or the shorter standardized MDR regimen (9–12 months). The duration on treatment largely depended on the patient's response to therapy, with BDQ being prescribed for >6 months in some patients [4, 5]. Study Outcome Computerized interpretation of the electrocardiogram was used, and the QTc values were abstracted directly from the printed ECG paper. The machines used in the 14 health facility sites use the Bazett QT correction formula. A significant elevation in QTc during BDQ use was defined as either a peak QTc >500 ms or an increase of QTc from a baseline of >60 ms. For each patient, QTc values were abstracted at different time points provided a patient had a baseline ECG and a follow-up ECG. Data Processing and Analysis Data were double-entered into REDCap and later exported to Stata, version 17, for analysis. Descriptive statistics were used to describe patient demographic, clinical, and social characteristics. The categorical variables were summarized using percentages, and continuous variables were summarized using mean and SD, or median and interquartile range (IQR) depending on the distribution. We cross-tabulated the covariates with the outcome (QTc prolongation) and tested differences in the outcome using the 2-sided chi-square test if the cell frequencies were large (≥5) or the Fisher exact test if they were small (<5). For continuous variables, mean differences between outcomes were tested using the 2-sided Student t test if the data were normally distributed, and the Wilcoxon rank-sum test was used if data were not normally distributed. Person-time was calculated from the start of BDQ therapy until QTc prolongation. Participants were censored if there was no QTc prolongation for the observation period; that is, the participant died or was lost to follow-up or was still on treatment or completed treatment during the observation treatment period. We used the modified Poisson regression model (through generalized linear models identity log and family Poisson) with robust standard errors and time as an offset variable for analysis at both the bivariable and multivariable levels to determine the factors associated with QTc prolongation among patients receiving DR-TB treatment. The modified Poisson technique with robust standard errors was used instead of logistic regression, the preferred technique for a binary outcome because the prevalence of the outcome in this study was high [13]. Where the prevalence of the outcome is not rare, that is, >10%, the odds ratios produced from a logistic regression model overestimate the effects of the covariates on the outcome [14]. However, the incidence rate ratios (IRRs) produced from the modified Poisson regression model are a better alternative for approximating risk ratio [15]. We selected variables with a P value <.2 on bivariate analysis to include in the multivariate model; however, in some instances we also considered clinical relevance and importance of the variable for QTc prolongation. We assessed for confounding using an absolute relative difference of 10% between the crude IRR and the adjusted IRR. We assessed for possible interaction using the likelihood ratio (LR) test, Bayesian information criterion (BIC), and Akaike information criterion (AIC), comparing the full model with all interaction terms and the reduced model without the interaction terms and found no significant interaction. All the predictors in the final model were considered significant at a level of P < .05, and adjusted rate ratios and 95% CIs were reported. Ethics Approval and Participant Consent The study was approved by the School of Medical and Health Sciences Research and Ethics Committee (SMHSREC; No.SBS-2023-302) and the Uganda National Council of Science and Technology (UNCST; No. HS2802ES), and administrative clearance was obtained from the respective DR-TB hospitals. The SMHSREC provided a waiver of participant consent. RESULTS Study Profile Based on the National TB and Leprosy Program data, of the 1957 DR-TB patients initiated on a BDQ-based regimen from January 2017 to December 2022, 72.8% (1425/1957) were excluded from the study for the following reasons: 49.3% (425, 702/1) had no baseline ECGs, 47.4% (425, 676/1) had no follow-up ECGs, 1.2% (17/1425) had missing files, 1.4% (20/1425) had no DR-TB confirmation by MTB/RIF assay or conventional Mycobacterium culture on sputum samples, and 0.7% (10/1425) had a baseline ECG of >500 ms. Only 27.2% (532/1957) were included in the study (see Figure 1 for details). Study profile of participants. Abbreviation: DR-TB, drug-resistant tuberculosis. Baseline Characteristics Among the 532 participants enrolled in the study, 377 (70.9%) were male, with a mean (SD) age of 41.4 (15.9) years. The median (IQR) BMI was 18.2 (16.3–20.6) kg/m2. Notably, the median (IQR) BMI was 17.7 (15.6–19.5) kg/m2 among those with QTc prolongation, while it was 18.3 (16.5–20.7) kg/m2 among those without QTc prolongation. The participants who reported smoking and alcohol use were 147 (27.6%) and 230 (43.2%), respectively. A positive culture at baseline was observed in 49.6% of participants. The mean baseline QTc interval was 426.8 ± 28.4 ms: 9.0% (n = 11) females and 11.7% (n = 44) males had borderline prolonged QTc (>450 ms for males and >460 ms for females) at baseline. More than a third of the participants (34.6%) were HIV positive, and 146 (27.4%) were on a dolutegravir (DTG)-based regimen. Generally, 26 (4.9%) had other comorbidities, for example, diabetes, hypertension, cardiac comorbidity, COPD. See Table 1 for further details. Table 1.Baseline Characteristics of the 532 Patients Initiated on a BDQ-Based Regimen From 2017–2022 From the 14 DR-TB SitesVariableCategoryNo. (%)QTc Prolongation P ValueYes (n = 69)No (n = 463)SexMale377 (70.9)45 (65.2)332 (71.7)Age, yMean (SD)41.4 (15.9)38.3 (17.3)41.8 (15.7).131BMI, kg/m2Median (IQR)18.2 (16.3–20.6)17.7 (15.6–19.5)18.3 (16.5–20.7).010BMI categorizedUnderweight (<18.5)271 (50.9)41 (59.4)230 (49.7).405Normal (18.5–24.9)207 (38.9)24 (34.8)183 (39.5)Overweight (25.0–29.9)11 (2.1)0 (0.0)11(2.4)Obese (≥30.0)12 (2.3)0 (0.0)12 (2.6)Unknown31 (5.8)4 (5.8)27 (5.8)Alcohol useYes230 (43.2)24 (34.8)206 (44.5).152SmokingYes147 (27.6)15 (21.7)132 (28.5).312HIV statusPositive184 (34.6)29 (42.0)155 (33.5).176HIV-based regimen (n = 184)DTG based146 (27.4)25 (36.2)121 (26.1).227Microbial culture sensitivityPositive264 (49.6)33 (47.8)231 (49.9).861Negative230 (43.2)32 (46.4 .7)198 (42.8)Unknown38 (7.1)4 (5.8)34 (7.3)Baseline QTcMean (SD)426.8 (28.4)414.0 (35.1)426.3 (22.3).001Baseline QTc categorizeda>460, sex is female (n = 155)14 (9.0)3 (12.5)22 (8.4).510>450, sex is male (n = 377)44 (11.7)6 (14.0)38 (11.5).636Serum potassiumMean (SD)5.1 (18.1)4.3 (0.3)5.2 (19.4).354Serum sodiumMean (SD)137.9 (2.9)137.9 (1.7)137.9 (3.0).996Hepatic impairmentYes192 (36.1)23 (33.3)169 (36.5).894No202 (38.0)27 (37.1)175 (37.8)Unknown138 (25.9)19 (27.5)119 (25.7)Renal impairmentYes36 (6.8)7 (10.1)29 (6.3).299No329 (61.8)38 (55.1)291 (62.9)Unknown167 (31.4)24 (34.8)143 (30.9)Any other comorbidity, eg, diabetes, hypertension, cardiac comorbidity, asthma, COPDYes26 (4.9)1 (1.5)25 (5.4).285Abbreviations: ALP, alanine phosphatase; ALT, alanine transferase; ASP, aspartate transferase; BMI, body mass index; COPD, chronic obstructive pulmonary disease; GGT, gama glutamyl transferase; IQR, interquartile range; TB, tuberculosis. aBaseline QTc categorized was categorized as borderline QTc prolonged in males if >450 ms and in females if >460 ms. Baseline Characteristics of the 532 Patients Initiated on a BDQ-Based Regimen From 2017–2022 From the 14 DR-TB Sites Abbreviations: ALP, alanine phosphatase; ALT, alanine transferase; ASP, aspartate transferase; BMI, body mass index; COPD, chronic obstructive pulmonary disease; GGT, gama glutamyl transferase; IQR, interquartile range; TB, tuberculosis. aBaseline QTc categorized was categorized as borderline QTc prolonged in males if >450 ms and in females if >460 ms. Clinical Characteristics More than half (55.3%) were newly initiated on TB treatment, and the majority (90.0%) had RR/MDR-TB. Most of the participants were co-administered levofloxacin (96.1%) and clofazimine (93.4%) alongside BDQ. The number of patients co-administered macrolides, metronidazole, amitriptyline, fluconazole, or ketoconazole and haloperidol was 12 (2.3%), 9 (1.7%), 17 (3.2%), 3 (0.6%), and 3 (0.6%), respectively (see Table 2 for details). Table 2.Clinical Characteristics of the 532 Patients Initiated on a BDQ-Based Regimen From 2017–2022 From the 14 DR-TB SitesVariableCategoryNo. (%)QTc Prolongation P ValueYes (n = 69)No (n = 463)TB treatment historyNew294 (55.3)38 (55.1)256 (55.3)1.00Retreatment227 (42.7)30 (43.5)197 (42.6)Unknown11 (2.1)1 (1.5)10 (2.2)Resistance profileRR/MDR479 (90.0)62 (89.9)417 (90.1).571Others (pre-X, XDR, H)53 (10.0)7 (10.1)46 (9.9)Drugs that prolong the QTC intervalMacrolidesYes12 (2.3)4 (5.8)8 (1.7).057MetronidazoleYes9 (1.7)0 (0.0)9 (1.9).613MoxifloxacinYes3 (0.6)1 (1.5)2 (0.4).341LevofloxacinYes511 (96.1)67 (97.1)444 (95.9).754ClofazimineYes506 (93.4)78 (98.7)428 (92.4).070FluconazoleYes3 (0.6)0 (0.0)3 (0.7)1.00HaloperidolYes3 (0.6)1 (1.5)2 (0.4).341AmitriptylineYes17 (3.2)3 (3.3)14 (3.2)Other drugs in the BDQ-based regimenEthionamideYes63 (11.8)6 (8.7)57 (12.3).548PyrazinamideYes90 (16.9)9 (13.0)81 (17.5).490IsoniazidYes6 (1.1)1 (1.5)5 (1.1).567Kanamycin/capreomycin/amikacinYes7 (1.3)1 (1.5)6 (1.3)1.00LinezolidYes477 (89.7)62 (89.9)415 (89.6)1.00CycloserineYes522 (98.1)68 (98.5)454 (98.1)1.00EthambutolYes20 (3.8)1 (1.5)19 (4.1).495Treatment outcomesDR-TB treatment outcomesCured/completed367 (69.0)51 (73.9)316 (68.3).873Still on treatment117 (22.0)13 (18.8)104 (22.5)Died25 (4.7)2 (2.9)23 (5.0)Lost to follow-up/treatment failure11 (2.1)1 (1.4)10 (2.2)Not evaluated12 (2.3)2 (2.9)10 (2.2)Abbreviations: IQR, interquartile range; MDR, multidrug resistance; RR, rifampicin resistance. Clinical Characteristics of the 532 Patients Initiated on a BDQ-Based Regimen From 2017–2022 From the 14 DR-TB Sites Abbreviations: IQR, interquartile range; MDR, multidrug resistance; RR, rifampicin resistance. Time to Onset of QTc Prolongation Of the 69 new cases of QTc prolongation recorded, 29 (42.0%), 27 (39.1%), 5 (7.2%), and 8 (11.6%) new cases were recorded at 3 months, 6 months, 9 months, and >9 months, respectively. QTc Measures and Incidence of QTc Prolongation The overall incidence risk was 12 (95% CI, 10.8–16.8) cases per 1000 person-months. Of the 69 patients who experienced QTc prolongation, 7.4% (5) had a follow-up QTc >500 ms only, and 84.1% (58) had a QTc change >60 ms from baseline only, while 8.7% (6) had both a follow-up QTc >500 ms and >60 ms from baseline (see Table 3 for details). Table 3.QTc Measures and Incidence of QTc ProlongationVariablesFrequenciesNo. at risk at baseline532QTc >500 ms11QTc >60 ms from baseline64Both QTc >500 ms and QTc >60 ms from baseline6No. with ≥1 prolonged QTc69Crude incidence rate0.012Cumulative incidence0.130 QTc Measures and Incidence of QTc Prolongation Treatment Outcomes Among Patients With and Without QTc Prolongation The treatment success rate was not different among patients with QTc prolongation vs those without (51 [73.9%] vs 316 [68.3%]; P = .873). Death occurred in 2 (2.9%) vs 23 (5.0%; P = 1.000) patients with and without QTc prolongation, respectively (see Table 2 for details). Factors Associated With QTc Prolongation Among Patients on BDQ-Based Regimens Every unit decrease in baseline BMI (IRR, 0.924; 95% CI, 0.860–0.993; P = .031) is associated with an 8.2% increased risk of QTc prolongation (see Table 4 for details). Table 4.Multivariate Analysis of the Patients Initiated on a BDQ-Based Regimen From 2017–2022 From the 14 DR-TB SitesVariableYes,No. (%)No, No. (%)Crude IRR (CI) P ValueAdjusted IRR (CI)Adjusted P ValueSexMale45 (65.2)332 (71.7)0.011 (0.008–0.015)<.0010.897 (0.510–1.580).708Age, yMedian (IQR)38.3 (16.6)41.8 (15.7)0.879 (0.869–0.889)<.0010.990 (0.971–1.009).287BMIMedian (IQR)17.3 (15.4–19.4)18.3 (16.5–20.7)0.778 (0.765–0.790)<.0010.924 (0.860–0.993) .031 HIV statusPositive46 (58.2)308 (66.5)0.014 (0.001–0.021)<.0011.175 (0.687–2.009).556MacrolidesYes5 (6.3)8 (1.7)0.036 (0.012–0.110)<.0012.984 (0.832–10.692).093SodiumMean (SD)4.2 (0.3)5.2 (19.4)0.968 (0.967–0.970)<.0010.987 (0.971–1.002).092PotassiumMean (SD)138.0 (1.5)137.9 (3.0)0.351 (0.330–0.373)<.0010.830 (0.572–1.203).325Baseline QTca>460 ms or >450 ms9 (13.4)49 (10.61)0.016 (0.008–0.033)<.0011.283 (0.530–3.106).581Baseline QTC was categorized based on sex >460 ms for females and >450 ms for males.Abbreviations: IQR, interquartile range; IRR, incidence rate ratio; MDT, multidrug resistance. aValues are mean (SD) or median (IQR) or No. (%). Multivariate Analysis of the Patients Initiated on a BDQ-Based Regimen From 2017–2022 From the 14 DR-TB Sites Baseline QTC was categorized based on sex >460 ms for females and >450 ms for males. Abbreviations: IQR, interquartile range; IRR, incidence rate ratio; MDT, multidrug resistance. aValues are mean (SD) or median (IQR) or No. (%). DISCUSSION From this study, 12 new cases of QTc prolongation were observed for every 1000 person-months of accumulated observation time. Decrease in BMI was associated with increased risk of QTc prolongation. Our study findings show a cumulative incidence of 13%, which is lower than the 14%, 18.9%, 19.5%, and 28% reported by other cohort studies [16–19] though higher than the 2.5% reported by a multicenter prospective study [20], 0%–6% by clinical trials [21, 22], and 7.8% by a systematic review and meta-analysis [23]. The lower prevalence could be because of a higher exclusion criterion of 70% of the participants with missing baseline and follow-up ECGs. In addition, ECGs in our study were computer-interpreted, yet in these other cited studies ECGs were manually interpreted by a clinician. This could have led to underreporting of the outcome in our study [24]. We have also noted that most cases with QTc prolongation were observed during the first 3 or 6 months post–treatment initiation, which is similar to findings from other retrospective studies [19, 25, 26], a prospective study [20], and clinical trials [21] conducted. Our study calls for health workers to be vigilant and conduct routine ECG monitoring during the first 6 months of BDQ initiation, such that all the cases are identified and monitored. It is of concern that >70% of potential participants missed baseline or follow-up ECGs for reasons such as a nonfunctional ECG machine, lack of ECG paper, and human resource constraints, which also demonstrates that ECG monitoring is currently not feasible or practical in low-resource settings and argues for a comprehensive approach to support the availability and utilization of ECG services in resource-limited settings. In our study, every unit increase in baseline BMI was associated with decreased risk in QTc prolongation, which is consistent with results from other studies conducted [11, 27]. This is likely because higher BMI is associated with decrease of serum concentration of BDQ [28], which further decreases the M2 metabolite concentration in serum and hence decreases QTc prolongation [29]. Our study is among the first large retrospective studies conducted in multiple hospitals to investigate the incidence and predictors of QTc interval prolongation among patients with DR-TB. Therefore, the findings could be generalizable to other settings in Sub-Saharan Africa. QT interval monitoring was done using ECG machines in the 14 hospitals using the Bazett QT correction method, which directly reported the QTc interval as recommended by the WHO [10]. To ensure reliability, ECG values were double-entered by 2 independent research assistants, and in case of disagreement by the abstractors, the principal investigator verified with the source ECG document. There are some limitations to our study that require discussion. Because of the retrospective nature of the study, follow-up data regarding some risk factors such as baseline electrolytes and renal and liver function tests were missing in the patient files. Missing data, with electrolytes registering the highest missingness at 23.4%, were handled using simple mean or median imputation methods. Also of concern was the missing baseline and follow-up ECGs, which led to 70% exclusion criteria, which was a potential source of selection bias. Some variables such as history of alcohol consumption and smoking and other drugs used by the patients for comorbidities were based on records in the files, which could have been underreported. Lastly, given that it is a retrospective review of records, a potential source of misclassification would be clinicians recording in the source document. CONCLUSIONS From this study, 12 new cases of QTc prolongation were observed for every 1000 person-months of observation period. A decrease in BMI was associated with increased incidence of QTc prolongation. The WHO recommended monthly ECG monitoring for patients with low baseline BMI. There was no indication that QTc prolongation was associated with adverse TB treatment outcomes in this cohort.