Abstract
Objectives:
To determine the incidence of mortality and its predictors among pulmonary tuberculosis (PTB) survivors treated at a rural Ugandan tertiary hospital.
Methods:
We conducted a retrospective chart review of data between 2013 and 2023. We included all people that met the World Health Organisation’s definition of tuberculosis cure and traced them or their next of kin to determine vital status (alive/deceased). We estimated the cumulative incidence of mortality per 1000 population, crude all-cause mortality rate per 1000 person-years, and median years of potential life lost for deceased individuals. Using Cox proportional hazard models, we investigated predictors of mortality.
Results:
Of 334 PTB survivors enrolled, 38 (11.4%) had died. The cumulative incidence of all-cause mortality was 113.7 per 1000 population, and the crude all-cause mortality rate was 28.5 per 1000 person-years. The median years of potential life lost for deceased individuals was 23.8 years (IQR: 9.6-32.8). Hospitalization (adjusted hazard ratio (aHR): 4.3, 95% CI: 1.1-16.6) and unemployment (aHR: 7.04, 95% CI: 1.5-31.6) at TB treatment initiation predicted mortality.
Conclusion:
PTB survivors experience post high mortality rates after TB cure. Survivors who were hospitalized and unemployed at treatment initiation were more likely to die after cure. Social protection measures and long-term follow-up of previously hospitalized patients could improve the long-term survival of TB survivors.
Keywords: TB, Cure, Treatment completion, Survivors, Unemployment, Hospitalization, Years of potential life lost
There were an estimated 155 million tuberculosis (TB) survivors globally as of 2020 [1]. However, the risk of mortality among TB survivors is thrice that of people who have never suffered TB [2]. Early identification of risk factors for long-term mortality in people with TB, especially in rural areas where studies have found a four-fold increase in mortality, could significantly improve their long-term survival [3]. This study aimed to determine the incidence and predictors of mortality among TB survivors at a rural tertiary hospital in Uganda. The absence of a long-term follow-up policy for drug-susceptible TB survivors in Uganda leaves a critical knowledge gap regarding rural TB survivor outcomes. This study highlights the need to address this gap.
We conducted a cross-sectional study that used retrospective data from TB survivors at Masaka Regional Referral Hospital (MRRH) in Southern Uganda. First, we reviewed data for a retrospective cohort of TB survivors who were cured of TB at MRRH from 2013 to 2023. These individuals (or their next of kin) were contacted by telephone or physically traced by the district TB and Leprosy supervisors and village health team members to establish their vital status (alive or dead). The study population was all people cured of bacteriologically confirmed pulmonary TB from 2013 to 2023 at MRRH [4]. Participants were excluded if they were missing telephone contacts and could not be physically traced. Data were abstracted from the unit TB registers and treatment files. The data extracted pertained to the characteristics of survivors at the time of TB treatment initiation. Data were analyzed in SPSS (version 29.0.1). The cumulative incidence was estimated as a proportion of TB survivors who were dead to the total eligible survivors per 1000 population. The crude mortality rate was the proportion of those who died to the total person-years. For an individual TB survivor who died, the years of potential life lost (YPLL) were calculated by subtracting the age at death from 63.3 years (the life expectancy of Ugandans) [5]. This was limited to the 32 individuals who died before the life expectancy age. We performed survival analysis using Cox proportional hazard models to determine predictors of mortality. In constructing a multivariable Cox proportional hazards model, we included all factors that had P <.1 at bivariate analysis in addition to other known predictors of mortality among TB survivors (sex, time from TB diagnosis to treatment initiation, and residence type [rural vs urban]). Statistical significance has been set at P <.05.
Of 469 pulmonary TB survivors who met the World Health Organisation (WHO) definition of cure, 334 (71.2%) were included in the study. Of these, 317 (94.9%) were contacted by telephone. Among the 135 excluded survivors, 62 (45.9%) had no contact details (both telephone number and residence details), while 73 (54.1%) had invalid telephone numbers and could not be traced by the study team.
Of 334 TB survivors, the median age was 32.0 (IQR: 25.0-47.0) years, 209 (62.6%) were male, 98 (29.3%) were coinfected with Human Immunodeficiency Virus (HIV), and 77 (28.4%) were unemployed at the time of TB treatment initiation. Further, 108 (34.1%) were hospitalized at the time of TB treatment initiation. Demographic and clinical characteristics of TB survivors at the time of TB treatment initiation are shown in Table 1.
Table 1.
Characteristics of TB survivors compared by crude mortality rates.
| Alive |
Deceased |
P-valuea | ||||
|---|---|---|---|---|---|---|
| Characteristic | Total 334 | N 296 | % 88.6 | N 38 | % 11.4 | |
| Age group (n = 333) | .006 | |||||
| <20 years | 32 | 30 | 10.2 | 2 | 5.3 | |
| 20-60 years | 277 | 249 | 84.4 | 28 | 73.7 | |
| 60+ years | 24 | 16 | 5.4 | 8 | 21.1 | |
| Sex | .056 | |||||
| Male | 209 | 180 | 60.8 | 29 | 76.3 | |
| Female | 125 | 116 | 39.2 | 9 | 23.7 | |
| Residence (n = 331) | .806 | |||||
| Rural | 186 | 165 | 56.1 | 21 | 56.8 | |
| Urban | 145 | 129 | 43.9 | 16 | 43.2 | |
| Education level (n = 215) | .045 | |||||
| Educated | 200 | 190 | 92.7 | 10 | 100.0 | |
| Not educated | 15 | 15 | 7.3 | 0 | 0.0 | |
| Employment status (n = 271) | .076 | |||||
| Employed | 194 | 184 | 72.7 | 10 | 55.6 | |
| Unemployed | 77 | 69 | 27.3 | 8 | 44.4 | |
| TB Resistance type | .954 | |||||
| Drug susceptible TB | 268 | 237 | 88.4 | 31 | 11.6 | |
| Drug resistance TB | 66 | 59 | 89.4 | 7 | 10.6 | |
| Hospitalization status (n = 317) | .072 | |||||
| Inpatient | 125 | 108 | 37.8 | 17 | 54.8 | |
| Outpatient | 192 | 178 | 62.2 | 14 | 45.2 | |
| TB symptoms (n = 267) | .016 | |||||
| Yes | 208 | 81.4 | 16 | 51.6 | ||
| No | 59 | 18.6 | 15 | 48.4 | ||
| Cough (n = 210) | .857 | |||||
| Yes | 199 | 94.8 | 15 | 93.8 | ||
| No | 11 | 5.2 | 1 | 6.3 | ||
| Dyspnoea (n = 208) | .915 | |||||
| Yes | 69 | 33.3 | 5 | 31.3 | ||
| No | 139 | 66.7 | 11 | 68.8 | ||
| Chest pain (n = 208) | .669 | |||||
| Yes | 104 | 51.0 | 6 | 37.5 | ||
| No | 104 | 49.0 | 10 | 62.5 | ||
| Haemoptysis (n = 206) | .241 | |||||
| Yes | 32 | 16.3 | 1 | 6.3 | ||
| No | 174 | 83.7 | 15 | 93.8 | ||
| Night sweats (n =207) | .088 | |||||
| Yes | 122 | 61.3 | 5 | 31.3 | ||
| No | 85 | 38.7 | 11 | 68.8 | ||
| Bacillary load (n = 122) | .980 | |||||
| Low or 1 + | 31 | 25.7 | 3 | 23.1 | ||
| Medium or 2+ | 43 | 34.9 | 5 | 38.5 | ||
| High or 3+ | 48 | 39.4 | 5 | 38.5 | ||
| HIV status | .333 | |||||
| Negative | 236 | 71.3 | 25 | 65.8 | ||
| Positive | 98 | 28.7 | 13 | 34.2 | ||
| Cotrimoxazole prophylaxis (n = 93)c | .461 | |||||
| Yes | 89 | 96.3 | 11 | 91.7 | ||
| No | 4 | 3.7 | 1 | 8.3 | ||
| ART (n = 95)c | .010 | |||||
| Yes | 86 | 92.7 | 10 | 76.9 | ||
| No | 9 | 7.3 | 3 | 23.1 | ||
| History of ART default (n = 60) | .626 | |||||
| Yes | 4 | 5.4 | 1 | 25.0 | ||
| No | 56 | 94.6 | 3 | 75.0 | ||
| Comorbidities | .013 | |||||
| Yes | 47 | 12.2 | 11 | 28.9 | ||
| No | 287 | 87.8 | 27 | 71.1 | ||
| Cardiometabolic diseaseb | .008 | |||||
| Yes | 30 | 7.8 | 7 | 18.4 | ||
| No | 304 | 92.2 | 31 | 81.6 | ||
| Diabetes (n = 64) | .393 | |||||
| Yes | 13 | 19.6 | 2 | 25.0 | ||
| No | 51 | 80.4 | 6 | 75.0 | ||
| Hypertension (n = 79) | .003 | |||||
| Yes | 11 | 10.0 | 4 | 44.4 | ||
| No | 68 | 90.0 | 5 | 55.6 | ||
| Renal disease (n = 73) | .007 | |||||
| Yes | 11 | 10.9 | 4 | 44.4 | ||
| No | 62 | 89.1 | 5 | 55.6 | ||
| Allergies (n = 18) | .383 | |||||
| Yes | 6 | 40.0 | 0 | 0.0 | ||
| No | 12 | 60.0 | 3 | 100.0 | ||
| Other comorbidities (n = 25) | .226 | |||||
| No | 11 | 55.6 | 1 | 14.3 | ||
| Yes | 14 | 44.4 | 6 | 85.7 | ||
Indicated P-values generated from cox regression bivariate analysis.
Cardiometabolic disease constituted clients who had a diagnosis of either Diabetes, Hypertension, or Renal Disease.
A subset of people with HIV.
TB: tuberculosis, ART: Antiretroviral therapy.
The total observation time was 1333.8 person-years, and the duration from cure to the study follow-up was a median of 46.7 (interquartile range (IQR): 27.2-63.7) months. Of 334 TB survivors enrolled, 38 (11.4%) had died. The median survival from TB cure to death was 8.8 (IQR: 1.1-34.5) months. The cumulative incidence of mortality was 113.7 (95% confidence interval (CI): 81.4-147.3) per 1000 population, and the crude all-cause mortality rate was 28.5 per 1000 person-years. The total YPLL were 708 years. The median YPLL for deceased individuals was 23.8 (IQR: 9.6-32.8) years. Hospitalization (adjusted hazard ratio (aHR): 4.3, 95% CI: 1.1-16.6, P = .034) and unemployment (aHR: 7.04, 95% CI: 1.5-31.6, P = .012) at TB treatment initiation were significantly associated with increased mortality risk at multivariable analysis (Table 2).
Table 2:
Predictors of mortality among TB survivors (N = 196).
| Characteristic | cHR | 95% CI | P-value | aHR | 95% CI | P-value |
|---|---|---|---|---|---|---|
| Age (every additional year) | 1.038 | (1.018, 1.059) | <.001 | 1.029 | (0.989, 1.07) | .158 |
| Sex | ||||||
| Female | Ref | Ref | ||||
| Male | 2.083 | (0.918, 4.424) | .056 | 3.512 | (0.715, 17.249) | .122 |
| Residence | ||||||
| Urban | Ref | Ref | ||||
| Rural | 1.085 | (0.565, 2.085) | .806 | 0.419 | (0.111, 1.583) | .199 |
| Employment status | ||||||
| Employed | Ref | Ref | ||||
| Unemployed | 2.347 | (0.916, 6.018) | .076 | 6.957 | (1.531, 31.612) | .012 |
| Hospitalization status | ||||||
| Outpatient | Ref | Ref | ||||
| Inpatient | 1.944 | (0.942, 4.014) | .072 | 4.314 | (1.118, 16.638) | .034 |
| Night sweats | ||||||
| No | Ref | Ref | ||||
| Yes | 0.396 | (0.137, 1.147) | .088 | 0.433 | (0.124, 1.511) | .189 |
| HIV status | ||||||
| Negative | Ref | Ref | ||||
| Positive | 0.716 | (0.365, 1.407) | .333 | 2.985 | (0.713, 12.493) | .134 |
| Cardiometabolic disease | ||||||
| No | Ref | Ref | ||||
| Yes | 3.110 | (1.347, 7.181) | .008 | 4.956 | (0.467, 52.588) | .184 |
| Time from diagnosis to treatment (every additional day) | 0.999 | (0.992, 1.007) | .867 | 1.02 | (0.852, 1.222) | .83 |
aHR: Adjusted Hazard ratio; cHR: Crude Hazard ratio; CI: confidence interval.
The mortality rate in our study (25 per 1000 person-years) is comparable to that reported in rural Ethiopia among TB survivors [3]. This rate is 5-9.5 times higher than the mortality rate reported in the general rural population in Uganda [6,7]. The high incidence of mortality is of concern since the majority of those who died were in their productive years (20-60 years of age). This translates to significant losses to families and the national economy, suggesting current TB mortality estimates likely underestimate overall TB-related mortality [8]. The findings advocate for long-term follow-up of previously hospitalized patients to address lingering health issues and potential complications. Additionally, the median survival of deceased individuals suggests a follow-up period of at least 1 year might be necessary. Unemployment’s association with mortality aligns with observations in India and underscores the detrimental role of socio-economic factors [9]. Social protection measures like cash transfers, education, and unemployment insurance could improve TB survivors’ prospects [10]. In our study setting, we have previously demonstrated that a 1-dollar incentive improved TB treatment success and reduced rates of lost-to-follow-up [11]. Unfortunately, participation in existing social protection programs is low in Uganda, necessitating the design of specific and accessible programs tailored to TB survivors [12].
Study limitations include missing data on known mortality predictors like alcohol use, smoking, and nutrition, highlighting the need for improved data collection in TB registers. Additionally, almost 29% of potential participants were excluded because they had no contacts and could not be traced physically. This implies that we might have underestimated the mortality rate among TB survivors in this setting. While the individuals we failed to trace could have been matched to mortality registries, such registries are not readily available in Uganda. Addressing these challenges through robust registries, long-term studies, and support groups for TB survivors is crucial to gain a deeper understanding of their experiences and improve their outcomes. These also would be important in surveilling for TB reinfection and relapse rates, which were data points missed in our study.
Funding
Research reported in this publication was supported by the Fogarty International Centre of the National Institutes of Health, US Department of State’s Office of the US Global AIDS Coordinator and Health Diplomacy (S/GAC), and President’s Emergency Plan for AIDS Relief (PEPFAR) under award number 1R25TW011213. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.
Footnotes
Use of generative AI
During the preparation of this work the authors used Google Bard AI in order to improve the grammar and sentence construction. After using this tool, the authors reviewed and edited the content as needed and take full responsibility for the content of the publication.
Declaration of competing interest
The authors declare no conflict of interests.
Ethical approval
The study was approved by the Mildmay Uganda Research Ethics Committee (Protocol number MUREC-2023-187) and the Uganda National Council of Science and Technology (HS2947ES). TB survivors or their next of kin provided verbal consent for using their retrospective data. Waiver of written consent for adults and assent for minors was provided by the Mildmay Uganda Research Ethics Committee.
Availability of data
Datasets used in this analysis are available from the corresponding author upon reasonable request.
References
- [1].Dodd PJ, Yuen CM, Jayasooriya SM, van der Zalm MM, Seddon JA. Quantifying the global number of tuberculosis survivors: a modelling study. Lancet Infect Dis 2021;21:984–92. doi: 10.1016/S1473-3099(20)30919-1. [DOI] [PubMed] [Google Scholar]
- [2].Romanowski K, Baumann B, Basham CA, Khan FAhmad, Fox GJ, Johnston JC. Long-term all-cause mortality in people treated for tuberculosis: a systematic review and meta-analysis. Lancet Infect Dis 2019;19:1129–37. doi: 10.1016/S1473-3099(19)30309-3. [DOI] [PubMed] [Google Scholar]
- [3].Datiko DG, Lindtjørn B. Mortality in successfully treated tuberculosis patients in southern Ethiopia: retrospective follow-up study. Int J Tuberc Lung Dis 2010;14:866–71. [PubMed] [Google Scholar]
- [4].Linh NN, Viney K, Gegia M, Falzon D, Glaziou P, Floyd K, et al. World Health Organization treatment outcome definitions for tuberculosis: 2021 update. Eur Respir J 2021;58:1–7. doi: 10.1183/13993003.00804-2021. [DOI] [PubMed] [Google Scholar]
- [5].Uganda Bureau of Statistics, Uganda Bureau of Statistics 2021 Statistical Abstract, Kampala, Uganda, 2021. https://www.ubos.org/wp-content/uploads/publications/01_20222021_Statistical_Abstract.pdf. [Accessed 19 April 2024]. [Google Scholar]
- [6].Kalyesubula R, Sekitoleko I, Tomlin K, Hansen CH, Ssebunya B, Makanga R, et al. Association of impaired kidney function with mortality in rural Uganda: results of a general population cohort study. BMJ Open 2022;12:e051267. doi: 10.1136/bmjopen-2021-051267. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [7].Nabukalu D, Reniers G, Risher KA, Blom S, Slaymaker E, Kabudula C, et al. Population-level adult mortality following the expansion of antiretroviral therapy in Rakai, Uganda. Popul Stud (Camb) 2019;74:93–102. doi: 10.1080/00324728.2019.1595099. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [8].Chakaya J, Kirenga B, Getahun H. Long term complications after completion of pulmonary tuberculosis treatment: a quest for a public health approach. J Clin Tuberc Other Mycobact Dis 2016;3:10–12. [Google Scholar]
- [9].Selvaraju S, Thiruvengadam K, Watson B, Thirumalai N, Malaisamy M, Vedachalam C, et al. Long-term survival of treated tuberculosis patients in comparison to a general population in South India: a matched cohort study. Int J Infect Dis 2021;110:385–93. doi: 10.1016/j.ijid.2021.07.067. [DOI] [PubMed] [Google Scholar]
- [10].Nightingale R, Carlin F, Meghji J, McMullen K, Evans D, van der Zalm MM, et al. Post-TB health and wellbeing. Int J Tuberc Lung Dis 2023;27:248–83. doi: 10.5588/ijtld.22.0514. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [11].Baluku JB, Nakazibwe B, Twinomugisha B, Najjuuko R, Isabella N, Nassozi S, et al. One dollar incentive improves tuberculosis treatment outcomes in programmatic settings in rural Uganda. Sci Rep 2021;11:19346. doi: 10.1038/s41598-021-98770-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [12].Nanyunja G, Kadota JL, Namale C, Hudson M, Nalugwa T, Turyahabwe S, et al. Feasibility of a social protection linkage program for individuals at-risk for tuberculosis in Uganda. PLOS Glob Public Health 2023;3:e0002122. doi: 10.1371/journal.pgph.0002122. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Data Availability Statement
Datasets used in this analysis are available from the corresponding author upon reasonable request.