SUMMARY
BACKGROUND:
Loss to follow-up (LTFU) is common among patients with drug-resistant TB (DR-TB) receiving second-line TB treatment; however, little is known about outcomes after LTFU, including mortality.
OBJECTIVE:
To determine rates of and factors associated with all-cause mortality among patients with DRTB who were LTFU.
METHODS:
Retrospective cohort study of adult patients with DR-TB in Georgia who initiated second-line TB treatment during 2011–2014 and were LTFU. Survival analyses were used to estimate all-cause mortality rates and adjusted hazard ratios (aHR).
RESULTS:
During 2011–2014, 2,437 second-line treatment episodes occurred and 695 patients were LTFU. Among 695 LTFU patients, 143 (21%) died during 2,686 person-years (PY) post-LTFU (all-cause mortality rate 5.1%, 95% CI 4.3–6.0 per 100 PY). In multivariable analysis, low weight (BMI < 18.5 kg/m2) at treatment initiation (aHR 3.2, 95% CI 2.2–4.7), return to treatment after LTFU (aHR 3.1, 95% CI 2.2–4.4), <12 months of treatment (aHR 2.4, 95% CI 1.4–4.1) and a pre-LTFU positive culture (aHR 3.3, 95% CI 2.2–4.9) were associated with all-cause mortality.
CONCLUSION:
High all-cause mortality occurred among patients with DR-TB after LTFU despite a low HIV prevalence. Providing additional assistance for patients during DR-TB treatment to prevent LTFU and use of new and shorter treatment regimens may reduce mortality among LTFU.
Keywords: tuberculosis, drug-resistant tuberculosis, TB mortality, loss to follow-up
RÉSUMÉ
CONTEXTE:
La perte de vue (LTFU) est courante chez les patients atteints de TB pharmacorésistante (DR-TB) sous traitement antituberculeux de deuxième ligne. Cependant, peu de données sont disponibles sur le devenir des patients après LTFU, y compris sur la mortalité. Nous souhaitions determiner les taux et les facteurs associésà la mortalité toutes causes confondues des patients atteints de DR-TB perdus de vue.
MÉTHODES:
Étude de cohorte retrospective auprès de patients adultes atteints de DR-TB en Géorgie ayant commencé un traitement antituberculeux de deuxième ligne entre 2011 et 2014 et ayant été perdus de vue. Des analyses de survie ont été utilises pour estimer les taux de mortalité toutes causes confondues et les aHR.
RÉSULAS:
Entre 2011 et 2014, 2 437 traitements de deuxième ligne ont été instaurés et 695 patients ont été perdus de vue. Parmi les 695 patients perdus de vue, 143 (21%) sont décedés pendant une observation de 2 686 personnes-années (PY) post-LTFU (taux de mortalité toutes causes confondues 5,1%, IC 95% 4,3–6,0 pour 100 PY). Dans l’analyse multivariée, un faible poids (indice de masse corporelle <18,5 kg/m2)à l’instauration du traitement (aHR 3,2, IC 95% 2,2–4,7), une reprise du traitement après LTFU (aHR 3,1, IC 95% 2,2–4,4), <12 mois de traitement (aHR 2,4, IC 95% 1,4–4,1) et une culture positive pré-LTFU (aHR 3,3, IC 95% 2,2–4,9) ont été associés à la mortalité toutes causes confondues.
CONCLUSION:
La mortalité toutes causes confondues était élevée chez les patients atteints de DR-TB après LTFU malgré une faible prevalence du VIH. Une aide supplémentaire apportée aux patients pendant le traitement de la DR-TB afin de prévenir la LTFU et l’utilisation des nouveaux schemas thérapeutiques plus courts pourraient permettre de réduire la mortalité chez les patients perdus de vue.
Drug resistant TB (DR-TB) is a major public health problem worldwide. In 2018, there were nearly 500,000 new cases of rifampicin-resistant TB (RR-TB), of which 78% had confirmed multidrug-resistant TB (MDR-TB).1 DR-TB treatment regimens are more complex than those for drug-susceptible TB (DS-TB), and traditionally require longer duration of treatment with more toxic drugs that often have disruptive side effects.2 In part due to side effects and longer treatment regimens, patients with DR-TB have a higher risk of treatment interruption and loss to follow-up (LTFU) during care.3
The WHO defines LTFU as a treatment interruption of ≥2 consecutive months.4 LTFU contributes to lower treatment success rates and is a critical challenge for TB control worldwide.1 Characteristics associated with LTFU among TB patients include male sex, drug use, tobacco and alcohol use, and previous history of TB treatment.5–9 Among DR-TB patients who are LTFU, long-term health outcomes, including all-cause mortality rates, are understudied. Mortality rates among patients with DR-TB are generally high (15–39%).10–18 However, few studies have evaluated mortality specifically among DR-TB patients after LTFU. One study reported higher post-LTFU mortality among DRTB patients with consistent positive cultures during treatment, low education and comorbid psychiatric disorder.19
The objectives of this study were 1) to estimate the all-cause mortality rate among patients with DR-TB who were LTFU, and 2) to identify socio-demographic, clinical, treatment-related factors, and comorbidities associated with death after LTFU. Identification of factors associated with mortality among DR-TB patients with LTFU will inform clinical and public health prevention policies to reduce mortality, and ultimately, achieve End TB Strategy goals.
METHODS
Setting and study design
This study took place in the country of Georgia from 2011 to 2018. Georgia was one of the first low- and middle-income countries to have universal access to both diagnosis and treatment for DR-TB. In 2018, TB incidence in Georgia was 80 per 100,000 persons, and 12% of new and 31% of previously treated patients had RR/MDR-TB.20 Patients with DR-TB receive directly observed treatment (DOT) in line with WHO treatment guidelines2 at TB clinical facilities throughout the country. DOT covers 100% of patients with TB (both drug-susceptible and drug-resistant). During the study period, standard second-line drug regimens included prothionamide, kanamycin, capreomycin, levofloxacin, moxifloxacin, cycloserine, para-aminosalicylic acid (PAS), amoxicillin/clavulanate, clarithromycin and clofazamine. New drugs were not routinely used during the study period; treatment for extensively drug-resistant TB (XDR-TB) using linezolid and imipenem-cilastatin began in mid-2014 and with bedaquiline in 2015.21
We used retrospective cohort information from patients with DR-TB who started second-line treatment from January 2011 to December 2014 under the Georgian National Tuberculosis Program (NTP), and subsequently had a clinical outcome of LTFU. TB data were cross-linked with patients’ mortality status using the Georgian National Death Registry in 2018. This study is a follow-up to a previous study that included adult patients with culture-confirmed pulmonary TB who started second-line TB treatment during 2011–2014 and were LTFU. The study assessed the bacteriologic status of patients after LTFU and determined factors associated with culture positivity after LTFU.22
Participants
Eligible participants included adult patients (≥18 years) with culture-confirmed or clinically diagnosed TB (pulmonary and extrapulmonary TB, or both) who began second-line treatment during the study period. Only the last episode of treatment was considered among patients with >1 outcome of LTFU during 2011–2014. If a patient started a new treatment episode after 2014, they were retained in the study regardless of the outcome of the last treatment episode.
Definitions and data sources
The primary outcome was all-cause mortality after LTFU. Patients’ vital status was determined by cross-referencing their national ID (unique personal identifying link) with the Georgian National Death Registry. We obtained patients’ name, date of birth, and individual code from National Center for Tuberculosis and Lung Diseases (NCTLD) medical records; we then determined the patients’ vital status (death from any cause) before July 27, 2018. Cause of death was not available from the death registration.
Additional characteristics were abstracted from the Georgian National TB Surveillance database and patient medical charts at NCTLD, including demographic information, TB history (previous treatment episodes and outcomes), hospitalization admission and discharge dates, comorbidities, incarceration history, tobacco use, alcohol and injecting drug use (IDU); a person was considered as injection drug user if they currently or ever used injection drugs. NCTLD routinely screens for diabetes, hepatitis C virus (HCV), and HIV among DR-TB patients. Low body mass index (BMI) was defined as ≤18.5 kg/m2. Diagnostic sputum and culture results, including date of sputum and culture conversion, were obtained from the National TB Surveillance database, and sputum culture conversion dates were obtained from the National TB Reference Laboratory database. Number of previous treatment episodes were categorized as 1) one treatment episode (current treatment episode that was included in the study); 2) 2–3 treatment episodes; and 3) ≥4 treatment episodes.
Statistical methods
All-cause mortality rates were estimated using survival analysis. Person-time (measured per 100 person-years [PY]) was defined from LTFU date until date of death or date of censorship. Participants were censored if there was no record of death on the date of death registry inquiry (July 2018). We used incidence rates (IRs) and proportional hazards models to estimate hazard ratios (HRs) and 95% confidence intervals (CIs) for factors associated with mortality after LTFU. Proportional hazards assumptions were evaluated statistically and graphically; covariates that did not satisfy the proportional hazards assumption were adjusted for in a stratified Cox model.23 In order to estimate associations between the study exposure domains of interest, we used three separate multivariable Cox models to estimate adjusted HRs (aHRs). The purpose of the separate models was to quantify the relationship between sociodemographic, clinical, treatment-related factors, and comorbidities associated with all-cause mortality after LTFU. Model A included treatment-related factors (time on treatment, previous TB treatment history either with first- or second-line drugs, pre-LTFU culture status); Model B included comorbidities (HIV, HCV, diabetes, IDU) and Model C included clinical factors (BMI, return to treatment, past TB history). Analyses were performed using SAS 9.4 (SAS Institute, Cary, NC, USA). A two-sided P < 0.05 was considered significant for all analyses.
The study protocol was approved by the institutional review boards at NCTLD, Tbilisi, Georgia, and Emory University, Atlanta, GA, USA. Data collection and record linkages took place at NCTLD during June–September 2018.
RESULTS
Study population
During the study period, a total of 2,437 second-line TB treatment episodes occurred in the country of Georgia (Figure). Of these treatment episodes, 782 (32%) had a final clinical outcome of LTFU. After removing duplicate treatment episodes, 695 patients were included in data analyses (Figure). Among these, 87% were male; the median age was 35 years (interquartile range [IQR] 28.4–45.1), 56% were previously treated for TB, and 30% resumed TB treatment after LTFU (Table 1). Among 211 patients who returned to treatment after LTFU, 26% (n=56) had successful treatment outcomes, 20% (n = 42) were LTFU, 6% (n=12) had outcome as failure, 26% (n = 55) patients had died and 22% (n=46) was missing outcome (these patients were probably still on treatment). Mean duration of TB treatment was 320 days (standard deviation [SD] 187) before LTFU. Median follow-up time (time from LTFU to event/censored) was 1,563 days (IQR 1058–1891) (available for 686 patients). During DR-TB treatment, 435 (63%) patients achieved sputum culture conversion prior to LTFU. Mean time to culture conversion was 143 days (SD 91) (available for 421 patients). HIV test result was available for 631 study patients, and 32 (5%) were HIV-positive.
Figure.
Flowchart of TB patients from Georgia who started second-line treatment during 2011–2014 and were LTFU. LTFU = lost to follow-up.
Table 1.
Crude estimates of association between patients’ demographic and clinical characteristics and death among patients who started second-line TB treatment in 2011–2014 and were lost to follow-up in the country of Georgia (n = 695)
| Variable | Total* (n = 695) n | Died (n = 143, 21%) n (%) | IR† (95% CI) | cHR‡ (95% CI) |
|---|---|---|---|---|
| Sex | ||||
| Male | 606 | 122 (20) | 5.2 (4.3–6.1) | Reference |
| Female | 89 | 21 (24) | 6.1 (3.8–9.1) | 1.2 (0.8–2.0) |
| Age at LTFU, years | ||||
| 19–30 | 224 | 31 (14) | 3.1 (2.1–4.4) | Reference |
| 31–45 | 300 | 61 (20) | 5.2 (4.0–6.7) | 1.6 (1.0–2.5) |
| 46–65 | 156 | 44 (28) | 7.6 (5.6–10.1) | 2.3 (1.4–3.6) |
| >65 | 15 | 7 (47) | 14.6 (6.4–28.9) | 4.2 (1.9–10.0) |
| BMI at treatment start, kg/m2 (n = 649) | ||||
| Underweight (<18.5) | 124 | 53 (43) | 13.6 (10.2–17.7) | 3.2 (2.2–4.5) |
| Normal weight (18.5–24.9) | 473 | 79 (17) | 4.0 (3.2–4.9) | Reference |
| Overweight (25–29.9) | 43 | 6 (14) | 3.6 (1.4–7.4) | 0.9 (0.4–2.0) |
| Obese (>30) | 9 | 1 (11) | 2.7 (0.1–13.3) | 0.7 (0.1–4.7) |
| Returned after LTFU | ||||
| Yes | 211 | 77 (36) | 10.8 (8.6–13.4) | 3.1 (2.2–4.3) |
| No | 484 | 66 (14) | 3.2 (2.5–4.0) | Reference |
| Region (n = 692) | ||||
| Tbilisi | 311 | 54 (17) | 4.1 (3.1–5.4) | Reference |
| Other | 381 | 89 (23) | 6.2 (5.0–7.6) | 1.4 (1.0–2.0) |
| Employment (n = 642) | ||||
| Employed | 55 | 11 (20) | 5.2 (2.7–9) | Reference |
| Unemployed | 559 | 117 (21) | 5.2 (4.3–6.3) | 1.0 (0.6–2.0) |
| Student | 7 | 0 (0) | 0 (0) | N/A |
| Retired | 20 | 10 (50) | 15.2 (7.7–27) | 2.9 (1.2–6.9) |
| Other | 1 | 0 (0) | 0 (0) | N/A |
| Incarceration history (n = 685) | ||||
| No | 342 | 75 (22) | 5.7 (4.5–7.1) | Reference |
| Yes | 343 | 67 (20) | 4.8 (3.7–6.1) | 0.9 (0.6–1.2) |
| Tobacco use (n = 692) | ||||
| No | 282 | 56 (20) | 4.9 (3.7–6.3) | Reference |
| Yes | 410 | 87 (21) | 5.5 (4.4–6.7) | 1.1 (0.8–1.6) |
| Alcohol use (n = 651) | ||||
| No | 389 | 72 (19) | 4.7 (3.7–5.8) | Reference |
| Yes | 262 | 63 (24) | 6.2 (4.8–7.9) | 1.3 (0.9–1.9) |
| IDU (n = 582) | ||||
| No | 540 | 111 (21) | 5.1 (4.2–6.2) | Reference |
| Yes | 42 | 17 (40) | 12.4 (7.2–20.0) | 2.2 (1.3–3.7) |
| Past TB history | ||||
| No | 310 | 53 (17) | 4.1 (3.1–5.3) | Reference |
| Yes | 385 | 90 (23) | 6.1 (5.0–7.5) | 1.4 (1.0–2.0) |
| Past DR-TB history (n = 385) | ||||
| No | 268 | 54 (20) | 5 (3.8–6.5) | Reference |
| Yes | 117 | 36 (31) | 9.4 (6.7–13.0) | 1.7 (1.1–2.6) |
| Previous treatment outcome (n = 265) | ||||
| Cured | 80 | 9 (11) | 2.3 (1.1–4.4) | Reference |
| Completed | 45 | 12 (27) | 7.6 (4.1–12.3) | 3.1 (1.3–7.5) |
| Failure | 28 | 9 (32) | 9.1 (4.4–16.7) | 3.6 (1.4–9.4) |
| Lost to follow-up | 112 | 33 (29) | 8.8 (6.2–12.3) | 3.4 (1.6–7.4) |
| Total number of treatment episodes | ||||
| Only 1 | 310 | 53 (17) | 4.1 (3.1–5.3) | Reference |
| 2–3 | 332 | 71 (21) | 5.5 (4.3–7.0) | 1.3 (0.9–1.9) |
| ≥4 | 53 | 19 (36) | 10.8 (6.6–16.7) | 2.4 (1.4–4.1) |
| Previously had TB treatment | ||||
| New case | 254 | 41 (16) | 3.8 (2.7–5.1) | Reference |
| First-line drugs only | 319 | 67 (21) | 5.3 (4.1–6.7) | 1.4 (0.9–2.0) |
| Second-line drugs | 122 | 35 (29) | 8.5 (6.0–11.8) | 2.1 (1.3–3.3) |
| TB type (n = 683) | ||||
| Pulmonary | 623 | 128 (21) | 5.2 (4.4–6.2) | Reference |
| Extrapulmonary | 25 | 3 (12) | 2.8 (0.7–7.5) | 0.5 (0.2–1.7) |
| Both | 35 | 9 (26) | 6.2 (2.9–11.7) | 1.2 (0.6–2.5) |
| Time on treatment, months (n = 686) | ||||
| <6 | 185 | 52 (28) | 7 (5.3–9.1) | 3.1 (1.9–5.0) |
| 6–11 | 243 | 60 (25) | 6.6 (5.1–8.4) | 2.7 (1.7–4.4) |
| 12–24 | 248 | 24 (10) | 2.4 (1.6–3.5) | Reference |
| >24 | 10 | 3 (30) | 8.1 (2.1–22.1) | 3.3 (1.0–10.6) |
| Sputum smear conversion | ||||
| Yes | 495 | 83 (17) | 4.2 (3.3–5.1) | Reference |
| No | 200 | 60 (30) | 8.0 (6.1–10.3) | 1.9 (1.4–2.7) |
| Culture conversion | ||||
| Yes | 435 | 59 (14) | 3.3 (2.5–4.2) | Reference |
| No | 260 | 84 (32) | 8.8 (7.0–10.8) | 2.7 (1.9–3.7) |
| Last culture result (n = 646) | ||||
| Positive | 179 | 72 (40) | 12.2 (9.5–15.3) | 4.3 (3–6.2) |
| Negative | 467 | 53 (11 | 2.7 (2.0–3.5) | Reference |
| Fasting blood glucose level, mmol/L | ||||
| <4 | 95 | 24 (25) | 6.7 (4.4–9.9) | 1.4 (0.9–2.2) |
| 4–5.6 | 337 | 66 (20) | 4.8 (3.7–6.1) | Reference |
| 5.7–6.4 | 58 | 10 (17) | 4.6 (2.3–8.1) | 1.0 (0.5–1.9) |
| >6.5 | 91 | 17 (19) | 4.5 (2.7–7.0) | 1.0 (0.6–1.6) |
| On diabetes treatment | ||||
| Yes | 43 | 10 (23) | 5.7 (2.9–10.1) | Reference |
| No | 652 | 133 (20) | 5.1 (4.3–6.1) | 0.9 (0.5–1.7) |
| HCV (n = 543) | ||||
| Negative | 381 | 74 (19) | 4.9 (3.9–6.1) | Reference |
| Positive | 162 | 33 (20) | 4.8 (3.3–6.8) | 1.0 (0.7–1.5) |
| HIV (n = 631) | ||||
| Negative | 599 | 117 (20) | 4.9 (4–5.8.0) | Reference |
| Positive | 32 | 11 (34) | 10.4 (5.5–18.0) | 2.1 (1.1–3.9) |
The numbers under each variable might not sum up to total due to the missing and unknown values.
Calculated among patients with known survival time (n = 139).
Calculated among patients with known survival time (n = 686). Proportional hazard. assumptions were assessed using goodness of fit and time-dependent covariate tests.
IR=incidence rate; CI=confidence interval; cHR=crude hazard ratio; LTFU = lost to follow-up; BMI = body mass index; IDU = injection drug use; DR-TB = drug-resistant TB; HCV = hepatitis C virus.
All-cause mortality rate
Among LTFU patients, there were 143 (21%) deaths during 2,686 PY post-LTFU. Survival time of 4 patients was missing: date of death was not available for 1 patient and 3 patients did not have date of LTFU indicated. The all-cause mortality rate (among 139 patients with full survival information) was 5.2 per 100 PY (95% CI 4.4–6.1). The median time to death was 385 days (IQR 174–662) after LTFU.
Risk factors of all-cause mortality
Mortality rates (per 100 PY) were highest among patients aged ≥65 (IR 14.6, 95% CI 6.4–28.9), persons who injected drugs (IR 12.4, 95% CI 7.2–20), patients with low BMI (<18.5 kg/m2) at treatment initiation (IR 13.6, 95% CI 10.2–17.7), and patients with ≥4 treatment episodes (IR 10.8, 95% CI 6.6–16.7). In unadjusted analyses, other factors associated with all-cause mortality included returning to care after LTFU, being culture-positive before LTFU, and having HIV (Table 1).
In a model adjusted for age and sex, the hazard rate of all-cause mortality was greater in participants with HIV than in those without HIV (aHR 2.0, 95% CI 1.0–3.7). Patients with a recorded positive sputum culture before LTFU (aHR 4.3, 95% CI 3.0–6.2), history of returning to treatment after LTFU (aHR 3.2, 95% CI 2.3–4.5), treatment <6 months (aHR 3.0, 95% CI 1.8–4.9), and low BMI (,18.5 kg/m2) (aHR 3.5, 95% CI 2.4–5.0) at treatment initiation were associated with increased rates of all-cause mortality after LTFU (Table 2).
Table 2.
Associations between patient characteristics and hazard of all-cause mortality among patients with DR-TB who started treatment in 2011–2014 and were lost to follow-up in the country of Georgia
| Model | HR (95% CI) | Age- and sex-adjusted HR (95% CI) | Fully adjusted HR (95% CI) |
|---|---|---|---|
| A: Treatment related factors | |||
| Time on treatment, months | |||
| <6 | 3.1 (1.9–5.0) | 3.0 (1.8–4.9) | 2.4 (1.3–4.2)* |
| 6–11 | 2.7 (1.7–4.4) | 2.6 (1.6–4.2) | 2.4 (1.4–4.1)* |
| 12–24 | Reference | Reference | Reference |
| >24 | 3.3 (1.0–10.6) | 3.3 (1.0–11.2) | 2.4 (0.5–10.7)* |
| Last culture–positive | 4.3 (3.0–6.2) | 4.3 (3.0–6.2) | 3.3 (2.2–4.9)* |
| Previously had TB treatment | |||
| New | Reference | Reference | Reference |
| First–line drugs | 1.4 (0.9–2) | 1.5 (1.0–2.2) | 1.2 (0.7–1.9)* |
| Second–line drugs | 2.1 (1.3–3.3) | 2.2 (1.4–3.5) | 1.4 (0.8–2.4)* |
| B: Comorbidities | |||
| HIV | 2.1 (1.1–3.9) | 2.0 (1.0–3.7) | 0.9 (0.3–2.9)† |
| HCV | 1.0 (0.7–1.5) | 0.9 (0.6–1.4) | 0.8 (0.4–1.3)† |
| On diabetes treatment | 0.9 (0.5–1.7) | 0.8 (0.4–1.5) | 1.0 (0.4–2.4)† |
| Injection drug use | 2.2 (1.3–3.7) | 2.2 (1.2–3.7) | 2.8 (1.3–5.9)† |
| C: Other factors | |||
| BMI, kg/m2 | |||
| <18.5 | 3.2 (2.2–4.5) | 3.5 (2.4–5.0) | 3.2 (2.2–4.7)‡ |
| 18.5–24.9 | Reference | Reference | Reference |
| 25–29.9 | 0.9 (0.4–2.0) | 0.8 (0.3–1.8) | 0.7 (0.3–1.7)‡ |
| >30 | 0.7 (0.1–4.7) | 0.4 (0.1–3.0) | 0.4 (0.1–2.7)‡ |
| Returned to treatment after LTFU | 3.1 (2.2–4.3) | 3.2 (2.3–4.5) | 3.1 (2.2–4.4)‡ |
| Having past TB history | 1.4 (1.0–2.0) | 1.4 (0.9–2.3) | 1.2 (0.8–1.7)‡ |
Adjusted for age, sex, BMI, returned after LTFU, last-culture result, having previous TB treatment and treatment duration.
Adjusted for age, sex, BMI, returned after LTFU, HIV, HCV, diabetes treatment and injection drug use.
Adjusted for age, sex, BMI, returned after LTFU and having past TB history.
DR-TB = drug-resistant TB; HR = hazard ratio; CI = confidence interval; HCV = hepatitis C virus; BMI = body mass index; LTFU = lost to follow-up.
Low BMI was associated with increased mortality (aHR 3.2, 95% CI 2.2–4.7) after controlling for multiple factors (age, sex, returned after LTFU and having past TB history). Association was greater in the fully adjusted model (aHR 3.6, 95% CI 2.0–6.4; Supplementary Table S1). In multivariable analysis, other factors significantly associated with increased mortality rate included having a positive culture at time of LTFU (aHR 3.3, 95% CI 2.2–4.9), receiving TB treatment for <6 months (<6 months vs. 12–24 months: aHR 2.4, 95% CI 1.3–4.2), receiving treatment for 6–11 months (6–11 months vs. 12–24 months: aHR 2.4, 95% CI 1.4–4.1), history of returning to treatment after LTFU (aHR 3.1, 95% CI 2.2–4.4), and IDU (aHR 2.8, 95% CI 1.3–5.9) (Table 2).
DISCUSSION
This retrospective cohort study investigated post-LTFU mortality among a large cohort of patients with DR-TB. We reported that nearly one quarter of patients with DR-TB who were LTFU subsequently died during an average of 4 years of follow-up. Factors associated with increased hazard of all-cause mortality after LTFU included low BMI (<18.5 kg/m2) at the start of TB treatment, returning to treatment after LTFU, positive sputum culture before LTFU, receiving second-line treatment for <12 months, and history of IDU. These findings demonstrate a frequently overlooked mortality burden attributable to DR-TB, and highlight the importance of preventing LTFU during second-line TB treatment.
Overall, TB is associated with increased risk of mortality after treatment. A 2019 review evaluating long-term, all-cause mortality among people treated for TB found increased risk of mortality compared to the general population (standardized mortality ratio (SMR) 2.9, 95% CI 2.2–3.8).24 Our study suggests that mortality rates among DR-TB patients after LTFU are significantly higher (SMR 15.9, 95% CI 12.4–20.2; Supplementary Table S2) than among the general population. Among Georgian DR-TB patients with LTFU, the all-cause mortality rate was 51.7/1000 people, while in the general population (country of Georgia) the 2018 all-cause mortality rate was 12.5/1000.25
We reported that 43% of LTFU patients with BMI <18.5 kg/m2 at the time of TB diagnosis died on average 4 years after LTFU. The mortality rate among those with low BMI was three times that of LTFU patients with normal weight (after adjusting for age and sex). Previous studies in the context of DR-TB also reported increased mortality among underweight patients, but these studies examined mortality during TB treatment only. A study conducted in Latvia among pulmonary MDR-TB patients identified being underweight at treatment initiation as a significant risk factor for mortality during treatment (aHR 1.9, 95% CI 1.1–3.5).26 Another study from China among XDR-TB patients reported low BMI was associated with mortality (aHR 4.5, 95%CI 1.3–15.7) during treatment.18 Low BMI can be an indicator of other factors such as malnutrition that may directly impact mortality risk. A study conducted in Malawi between July 1999 and December 2000 among patients with pulmonary TB found that low BMI was independently associated with severity of TB disease.27 A systematic review that included 36 studies reported that low BMI (odds ratio 0.4, 95% CI 0.2–0.7) was associated with poor outcomes among patients with MDR-TB.28 Taken together with findings from previous studies, our results suggest that DR-TB patients who are underweight at anti-TB treatment initiation should be prioritized for additional support and receive increased observation during treatment.
We reported that LTFU patients who returned to TB treatment after LTFU had higher relative rates of all-cause mortality. Those patients who returned to TB care after LTFU may have had persistent TB symptoms and thus sought additional medical care. It is likely that some LTFU patients who did not return to care achieved cure (i.e., achieved microbial clearance) prior to TB treatment completion, although they had been categorized as LTFU. Another risk factor associated with mortality in our study was having positive last culture before LTFU. Consistent with our findings, a study conducted in Peru among patients with laboratory-confirmed MDR-TB who were LTFU reported that having last culture positive before LTFU was a strong predictor of mortality (aHR 9.9, 95% CI 2.5–38.8).19 Positive cultures at the time of LTFU may indicate continued active disease, and therefore, increased risk of death.
Patients in our study who had history of IDU had higher risk of all-cause mortality after LTFU. A prospective cohort study conducted in Vietnam between 2005 and 2007 among 894 male injection drug users reported that previous TB diagnosis increased mortality rates among injection drug users (aHR 10.0, 95% CI 4.1–24.3).29 Mechanisms underlying the higher mortality among patients with a history of IDU is unknown. The association is likely due to a combination of factors that often accompany IDU and are known risk factors for death, such as lower socioeconomic status and comorbidities like HIV and HCV.
This study had limitations. First, tobacco, alcohol and IDU were self-reported and may have been underreported. Second, in some cases, culture conversion date was not recorded, and such cases would be misclassified as “not converted”. Third, we did not have key clinical measures (e.g., changes in blood glucose, adverse events, liver function) during treatment before LTFU or when patient returned after LTFU. Also, we did not have information about drug susceptibility profile, including resistance to fluoroquinolones. These measures could signal severity of TB episode which could partially explain the risk of all-cause mortality after LTFU. Fourth, as we were unable to determine cause of death, this study does not report information about what proportion of the patients died from TB-attributable complications, and only evaluates all-cause mortality. Fifth, although the death registry is highly sensitive in capturing deaths within the country, it may not contain information about persons who died outside of Georgia. As we did not have information about study patients’ immigration status, our measure of all-cause mortality may be an underestimate of the true number of deaths.
Despite limitations, our study had several strengths. First, the Georgian NTP data used in this study is representative of the entire country. Our study population included all patients who were enrolled into DR-TB treatment under the NTP, and are thus broadly generalizable to the entire country. Second, we successfully linked TB patient data with the Georgian National Death Registry, a nationwide system that captures timely information on death with high sensitivity. Third, we had follow-up information on a large cohort of lost to follow-up patients (695 TB patients for an average of 3.9 years of person-time after LTFU). This extensive amount of person-time information provided detailed estimation of hazard rates of mortality.
In conclusion, all-cause mortality rates in the country of Georgia were high among patients with DR-TB who were lost to follow-up. Our findings indicate the importance of preventing treatment interruption in order to reduce TB-attributable mortality. TB programs should tailor interventions to reduce LTFU rates and/or mortality among lost to follow-up patients by focusing on patients with risk factors such as those with positive culture before LTFU and being underweight. These risk groups should be targeted with novel approaches that are effective at reducing LTFU, such as video-observed therapy30 and psychosocial support,31 during treatment. These interventions, in combination with new short-treatment regimens, may reduce mortality in patients with DR-TB by increasing retention in treatment.
Supplementary Material
Acknowledgements
The authors thank National Center for Tuberculosis and Lung Diseases database manager, M Chincharauli, who helped us access to NTP surveillance database and relevant variables, and M Kvashali, who assisted in the data collection process.
This work has been supported in part by the National Institutes of Health/Fogarty International Center Global Infectious Diseases Grant D43TW007124. ADS was supported by a VECD Global Health Fellowship, funded by the Fogarty International Center (FIC) of the NIH (D43 TW009337). The work was also supported by NIH grants R01AI153152 and R21TW011157.
Footnotes
Conflicts of interest: none declared.
References
- 1.World Health Organization. Global tuberculosis report, 2019. Geneva, Switzerland: WHO, 2019. [Google Scholar]
- 2.World Health Organization. Guidelines for the programmatic management of drug-resistant tuberculosis, 2011. Geneva, Switzerland: WHO, 2011. Accessed January 2020. www.who.int/hiv/topics/mtct/grc_handbook_mar2010_1.pdf [Google Scholar]
- 3.Lan Z, et al. Drug-associated adverse events in the treatment of multidrug-resistant tuberculosis: an individual patient data meta-analysis. Lancet Respir Med 2020; 8(4): 383–394. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Gandhi NR, et al. Multidrug-resistant and extensively drug-resistant tuberculosis: a threat to global control of tuberculosis. Lancet 2010; 375(9728): 1830–1843. [DOI] [PubMed] [Google Scholar]
- 5.Kuchukhidze G, et al. Risk factors associated with loss to follow-up among multidrug-resistant tuberculosis patients in Georgia. Public Health Action 2014; 4: S41–S46. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Holtz TH, et al. Risk factors associated with default from multidrug-resistant tuberculosis treatment, South Africa, 1999–8211. Int J Tuberc Lung Dis 2001; 10(6): 649–655. [PubMed] [Google Scholar]
- 7.Wohlleben J, et al. Risk factors associated with loss to follow-up from tuberculosis treatment in Tajikistan: a case-control study. BMC Infect Dis 2017; 17(1): 543. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Moyo S, et al. Loss from treatment for drug resistant tuberculosis: risk factors and patient outcomes in a community-based program in Khayelitsha, South Africa. PLoS One 2015; 10(3): e0118919. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Tupasi TE, et al. Factors associated with loss to follow-up during treatment for multidrug-resistant tuberculosis, the Philippines, 2012–2014. Emerg Infect Dis 2016; 22(3): 491–502. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Frank M, et al. Long-term follow-up reveals high posttreatment mortality rate among patients with extensively drug-resistant tuberculosis in the country of Georgia. Open Forum Infect Dis 2019; 6(4): ofz152. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Kliiman K, Altraja A. Predictors and mortality associated with treatment default in pulmonary tuberculosis. Int J Tuberc Lung Dis 2010; 14(4): 454–463. [PubMed] [Google Scholar]
- 12.Javaid A, et al. Risk factors for high death and loss-to-follow-up rates among patients with multidrug-resistant tuberculosis at a programmatic management unit. Am J Infect Control 2017; 45(2): 190–193. [DOI] [PubMed] [Google Scholar]
- 13.Blöndal K, et al. Overall and cause-specific mortality among patients with tuberculosis and multidrug-resistant tuberculosis. Int J Tuberc Lung Dis 2013; 17(7): 961–968. [DOI] [PubMed] [Google Scholar]
- 14.Mohd Shariff N, Shah SA, Kamaludin F. Predictors of death among drug-resistant tuberculosis patients in Kuala Lumpur, Malaysia: a retrospective cohort study from 2009 to 2013. J Glob Antimicrob Resist 2016; 6: 102–107. [DOI] [PubMed] [Google Scholar]
- 15.Chung-Delgado K, et al. Mortality among MDR-TB cases: comparison with drug-susceptible tuberculosis and associated factors. PLoS One 2015; 10(3): e0119332. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Chingonzoh R, et al. Risk factors for mortality among adults registered on the routine drug resistant tuberculosis reporting database in the Eastern Cape Province, South Africa, 2011 to 2013. PLoS One 2018; 13(8): e0202469. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Salinas JL, et al. Factors associated with all-cause mortality among patients with multidrug-resistant tuberculosis—United States, 1993–2013. Clin Infect Dis 2017; 65(11): 1924–1926. [DOI] [PubMed] [Google Scholar]
- 18.Bei C, et al. Mortality and associated factors of patients with extensive drug-resistant tuberculosis: an emerging public health crisis in China. BMC Infect Dis 2018; 18(1): 261. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Franke MF, et al. Risk factors and mortality associated with default from multidrug-resistant tuberculosis treatment. Clin Infect Dis 2008; 46(12): 1844–1851. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.World Health Organization. Global tuberculosis report, 2019. Geneva, Switzerland: WHO, 2020. [Google Scholar]
- 21.World Health Organization. Interim guidance on the use of bedaquiline to treat MDR-TB. Geneva, Switzerland: WHO, 2013. [Google Scholar]
- 22.Kuchukhidze G, et al. Long-term mortality and active TB disease among patients who were lost to follow-up during second line tuberculosis treatment in 2011–2014: population-based study in the country of Georgia. Open Forum Infect Dis 2021; 8(6): ofab127. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Kleinbaum D, Klein M. Survival analysis: a self-learning text. 3rd ed. New York, NY, USA: Springer, 2012. [Google Scholar]
- 24.Romanowski K, et al. Long-term all-cause mortality in people treated for tuberculosis: a systematic review and meta-analysis. Lancet Infect Dis 2019; 19(10): 1129–1137. [DOI] [PubMed] [Google Scholar]
- 25.National Statistics Office of Georgia. Deaths. Tbilisi, Georgia: NSO, 2020. https://www.geostat.ge/en/modules/categories/320/deaths [Google Scholar]
- 26.Podewils LJ, et al. Impact of malnutrition on clinical presentation, clinical course, and mortality in MDR-TB patients. Epidemiol Infect 2011; 139(1): 113–120. [DOI] [PubMed] [Google Scholar]
- 27.van Lettow M, et al. Micronutrient malnutrition and wasting in adults with pulmonary tuberculosis with and without HIV co-infection in Malawi. BMC Infect Dis 2004; 4: 61. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Johnston JC, et al. Treatment outcomes of multidrug-resistant tuberculosis: a systematic review and meta-analysis. PLoS One 2009; 4(9): e6914. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Quan VM, et al. Mortality and HIV transmission among male Vietnamese injection drug users. Addiction 2011; 106(3): 583–589. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Ravenscroft L, et al. Video-observed therapy and medication adherence for tuberculosis patients: randomised controlled trial in Moldova. Eur Respir J 2020; 56(2): 2000493. [DOI] [PubMed] [Google Scholar]
- 31.Law S, et al. Interventions to improve retention-in-care and treatment adherence among patients with drug-resistant tuberculosis: a systematic review. Eur Respir J 2019; 53(1): 1801030. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.