RESEARCH ARTICLE Tuberculosis recurrences and predictive factors in a vulnerable population in Catalonia Sı́lvia Brugueras ID1,2,3☯*, Vinicio-Israel Molina4☯, Xavier Casas4, Yoel- Domingo González4, Nuria Forcada4, Dora Romero4, Anna Rodés5, Maria-Neus Altet4, José Maldonado4, Mario Martin-Sánchez ID6, Joan A. Caylà7, Àngels Orcau1,2,7, Cristina Rius1,2,3, Joan-Pau Millet ID1,2,4,7 1 Epidemiology Service, Agència de Salut Pública de Barcelona (ASPB), Barcelona, Spain, 2 Consorcio de Investigación Biomédica en Red de Epidemiologı́a y Salud Pública (CIBERESP), Madrid, Spain, a1111111111 3 Departamento de Pediatrı́a, Obstetricia y Ginecologı́a y Medicina Preventiva, Facultad de Medicina, a1111111111 Universidad Autónoma de Barcelona, Barcelona, Spain, 4 Serveis Clı́nics, Barcelona, Spain, 5 Agència de Salut Pública de Catalunya, Barcelona, Spain, 6 Preventive Medicine and Public Health Training Unit Parc de a1111111111 Salut Mar–Pompeu Fabra University—Public Health Agency of Barcelona (PSMar-UPF-ASPB), Barcelona, a1111111111 Spain, 7 Foundation of the Tuberculosis Research Unit of Barcelona (fuiTB), Barcelona, Spain a1111111111 ☯ These authors contributed equally to this work. * silviabruguerastorrella@gmail.com OPEN ACCESS Abstract Citation: Brugueras S, Molina V-I, Casas X, González Y-D, Forcada N, Romero D, et al. (2020) Tuberculosis recurrences and predictive factors in Background a vulnerable population in Catalonia. PLoS ONE 15 (1): e0227291. https://doi.org/10.1371/journal. Patients with a history of tuberculosis (TB) have a high probability of recurrence because pone.0227291 long-term cure is not always maintained in successfully treated patients. The aim of this Editor: Kristien Verdonck, Institute of Tropical study was to identify the probability of TB recurrence and its predictive factors in a cohort of Medicine Antwerp, BELGIUM socially vulnerable patients who completed treatment in the TB referral center in Catalonia, Received: April 30, 2019 which acts as the center for patients with social and health problems. Accepted: December 16, 2019 Methods Published: January 15, 2020 This retrospective open cohort study included all patients diagnosed with TB who were admit- Copyright: © 2020 Brugueras et al. This is an open access article distributed under the terms of the ted and successfully treated in Serveis Clı́nics between 2000 and 2016 and who remained Creative Commons Attribution License, which disease-free for a minimum of 1 year after treatment completion. We calculated the incidence permits unrestricted use, distribution, and density of TB recurrences per person-years of follow-up. We also estimated the cumulative reproduction in any medium, provided the original incidence of TB recurrence at 1, 2, 5, and 10 years of follow-up. Bivariate analysis was con- author and source are credited. ducted using Kaplan-Meier curves. Multivariate analysis was conducted using Cox regres- Data Availability Statement: Data cannot be sion. Hazard ratios (HR) were calculated with their 95% confidence intervals (95%CI). shared publicly because of ethical restrictions to the data. Furthermore, the database in itself is the property of Serveis Clı́nics and the Public Health Results Agencies of Catalonia and Barcelona and is not intended for use outside of its institutions and There were 839 patients and 24 recurrences (2.9%), representing 0.49 per 100 person- centers. However, readers interested in requesting years. The probability of a recurrence was 0.63% at 1 year of follow-up, 1.35% at 2 years, any anonymized data may contact Serveis Clı́nics and 3.69% at 5 years. The multivariate analysis showed that the predictive factors of recur- (administracio@serviciosclinicos.com; contact rence were age older than 34 years (aHR = 3.90; CI = 1.06–14.34 at age 35–45 years and person: Maria-Teresa Moliner) or the Epidemiology Service of the Public Health Agency of Barcelona aHR = 3.88; CI = 1.02–14.80 at age >45 years) and resistance to at least one anti-TB drug (mdo@aspb.cat; contact person: Carmen Serrano) (aHR = 2.91; CI = 1.11–7.65). PLOS ONE | https://doi.org/10.1371/journal.pone.0227291 January 15, 2020 1 / 18 Tuberculosis recurrences in vulnerable population to have access to the databases used to carry out Conclusions the study. Attention should be paid to socially vulnerable persons older than 34 years with a previous Funding: The health center Serveis Clinics provided episode of resistant TB. Surveillance resources should be directed toward adequately support in the form of salaries for authors VIM, XC, YDG, NF, DR, NA, JM, JPM, but did not have any treated patients who nevertheless have a high risk of recurrence. additional role in the study design, data collection and analysis, decision to publish, or preparation of the manuscript. The specific roles of these authors are articulated in the ‘author contributions’ section. Introduction VIM funded by SEPAR research grant. Project 088/ 2015. Sociedad Española de Neumologı́a y Cirugı́a In 2017, there were an estimated 10 million people with tuberculosis (TB) worldwide. Torácica (SEPAR) https://www.separ.es/. The Although considerable progress has been made in the last few years, the main obstacles to con- funders had no role in study design, data collection trolling the infection are human immunodeficiency virus (HIV), the increase in antibiotic and analysis, decision to publish, or preparation of resistance, and the problem of TB in large cities associated with socially vulnerable populations the manuscript. at risk of social exclusion, such as drug users, alcoholics, and economic immigrants [1–4]. Competing interests: The authors have declared Interest in TB recurrences has been growing because, despite efforts to control the disease that no competing interests exist. The health center globally, long-term cure is not maintained in all successfully treated patients [5,6]. Compared Serveis Clinics provided support in the form of with patients without a prior TB episode, those with a history of TB are at higher risk of con- salaries for some of the authors. This does not alter our adherence to PLOS ONE policies on sharing tracting the disease again [7]. data and materials. Tuberculosis recurrence can be defined as a new episode occurring in adequately treated patients who completed therapy [8]. Both the Centers for Disease Control and the Instituto de Salud Carlos III (ISCIII) accept that all TB episodes occurring 1 year or later after treatment completion in patients who were therefore disease-free should be considered as new TB cases and consequently as recurrences [9,10]. In areas with a low incidence of TB, most recurrences are attributed to reactivation or incomplete cure of the first TB episode, instead of a new infection due to repeat exposure. In contrast, the latter is more common in countries with a high burden of the disease, where transmission is highly frequent [5,11–13]. While the factors associated with poor treatment adherence are well known, there are few reports of recurrence rates and their associated factors in vulnerable populations in countries with a medium or low incidence [5,6,14,15]. Such knowledge would help to identify the popu- lations that are most susceptible and which have the highest risk of recurrent TB. This would allow the design of strategies and help to focus efforts and interventions on post-treatment fol- low-up and application of public health measures to minimize the risk of a new episode. In Catalonia, directly observed treatment (DOT) is not indicated in all TB patients. DOT is indicated in patients with risk factors for treatment non-adherence (alcohol or substance abuse, homelessness or immigration from countries with a high TB burden) or those with anti- biotic resistance [16]. Serveis Clı́nics is a center providing community-based DOT and also has a TB clinic that is the referral center for the whole of Catalonia and serves as the center for patients with risk factors, poor adherence, multidrug resistant TB (MDR-TB) or extensively drug-resistant TB (XDR). The aim of this study was to determine the probability of TB recurrence and its predictive factors in a cohort of especially vulnerable patients who completed treatment in Servies Clı́nics. Methods Study design and setting This retrospective, open cohort study included all patients with TB admitted to Serveis Clı́nics and who started treatment between 2000 and 2016. PLOS ONE | https://doi.org/10.1371/journal.pone.0227291 January 15, 2020 2 / 18 Tuberculosis recurrences in vulnerable population Study population The study population consisted of patients with a diagnosis of TB, who were successfully treated and disease-free a minimum of 1 year after completing treatment and who started treatment after 1st January, 2000. Patient inclusion ended with patients who were successfully treated up to 30th October, 2016. The patients were retrospectively followed-up until 30th October, 2017. Length of follow-up was calculated bearing in mind the time interval between treatment end and 1 year until recurrence, death, change of address, or end of the study. A flow diagram of included patients and recurrences is shown in S1 Fig. Definitions TB recurrence was defined as the presence of a new episode of the disease after completion of adequate treatment in persons considered to have been successfully treated and disease-free for a minimum of 1 year after the end of treatment [9,10]. A cured patient was a patient whose sputum smear or culture was positive at the beginning of the treatment but who was smear- or culture-negative in the last month of treatment and on at least one previous occasion. Patients who completed treatment but who did not have a nega- tive sputum smear or culture result in the last month of treatment and on at least one previous occasion were classified as having “completed treatment”. These two categories were consid- ered as treatment success [17]. A homeless person was defined as a person living on the street or in municipal facilities (with no fixed abode) and who could be at risk of homelessness at any moment. Prison or legal problems were defined as persons with a history of incarceration in jail. A smoker was defined as any person smoking any number of cigarettes every day during the last month, including one. Alcohol abuse was defined as alcohol consumption above 280 g per week in men and over 170 g in women. Substance abuse was defined as harmful or hazardous use of psychoactive substances. Information sources and variables Information on cases was obtained from the clinical histories at Serveis Clı́nics. Cases were fol- lowed-up by consulting the clinical histories at Serveis Clı́nics, the registry of TB cases of the TB Control Programs of Barcelona and Catalonia, the Shared Clinical History of Catalonia, the Central Registry of Insured Persons, and the Barcelona census. Epidemiological informa- tion on the disease and the sociodemographic characteristics of cases were completed by using these registries. The final follow-up date and status of TB patients who had left Catalonia or had died were also obtained from these registries. The dependent variable was the presence or absence of recurrence. The independent vari- ables were age (tertiles), sex, country of origin (Spain/outside Spain), homelessness, prison or legal problems, smoking, alcohol abuse, substance abuse, HIV infection, hepatitis C, diabetes, pulmonary comorbidities, type of TB (pulmonary, extrapulmonary), chest x-ray (presence or absence of cavitations), resistances, type of resistance (rifampicin, other drugs), treatment length, and prior TB episodes. Ethical considerations All data were fully anonymized and were kept strictly confidential following the ethical princi- ples of the Declaration of Helsinki [18], the Spanish Data Protection Law 3/2018 [19], and the General Data Protection Regulation (EU) No. 2016/679 [20]. The project was approved by the PLOS ONE | https://doi.org/10.1371/journal.pone.0227291 January 15, 2020 3 / 18 Tuberculosis recurrences in vulnerable population ethics committee for clinical research of Vall d’Hebron University Hospital in Barcelona (no. PR(AG)19/2016). Data analysis We analyzed the distribution of frequencies and calculated measures of central tendency, the median and interquartile range (IQR). Univariate analysis was conducted with quantitative and qualitative variables. Incidence density was calculated both overall and stratified by age group, country of origin, smoking, alcohol abuse, HIV infection, type of TB, prior episode of TB, and resistance to at least one drug. Incidence density was expressed in 100 person-years of follow-up. The numera- tor was the total number of TB recurrences in the study period and the denominator was the sum of all the individual follow-up periods. We calculated the cumulative incidence of TB recurrences at 1, 2, 5, and 10 years of follow- up with 95% confidence intervals (95%CI). The risk of recurrence at 5 years was calculated according to the various independent variables. The cumulative incidence of TB recurrence was estimated by Kaplan-Meier curves both overall and stratified by country of origin, smok- ing, age group, and prior TB episodes. Groups were compared using the log rank test, taking a P-value less than 0.05 as statistically significant. Based on Cox proportional hazards models, crude and adjusted hazard ratios (HR) for the risk of recurrence and their 95% CI were calculated by multivariate analysis. A model was cre- ated with all the variables and those with a P-value higher than 0.2 in the bivariate analysis were progressively excluded. For each variable included in the model, we confirmed that the risks were proportional through Shoenfeld’s partial residuals. To compare the Cox propor- tional hazards models, we used Akaike’s information criterion and Bayesian information criterion. The statistical analysis was performed using the SPSS25 and STATA (Versión 13.0; Stata Corp, College Station, TX) programs. Results Between 2000 and 2016, a total of 1064 patients with TB were admitted to Serveis Clı́nics. Of these, we excluded 225 patients from the cohort (55 lost to follow-up during treatment, 20 moved home, 77 had no health card number, and 73 did not have a minimum of 1 year of fol- low-up after cure). In all, 839 successfully treated patients who were disease-free at a minimum of 1 year were followed-up (S1 Fig). Among the 839 patients, the median age was 40 years (IQR, 31–50), 727 (86.7%) were men and 467 (55.7%) were immigrants. A total of 158 (18.8%) were homeless, 74 (8.8%) had been in prison or had legal problems, 583 (69.5%) smoked, 379 (45.2%) showed risky alcohol con- sumption, and 171 (20.4%) were drug users. In all, 185 (22.1%) had HIV infection and 141 (16.8%) had hepatitis C infection. Most (88.0%) had pulmonary TB, 40.9% had cavities on chest x-ray, and 13.6% were resistant to at least one drug. There were 24 recurrences (2.9%) during the follow-up period (2000–2017) (Table 1). The mean length of follow-up in the 839 patients was 5.81 years and the median was 4.74 years (IQR 2.20–9.01). Among the 839 TB cases in the cohort, the Incidence density of recur- rence was 0.49 per 100 person-years (492.65 per 100,000 PY). There were no significant differ- ences between the groups in the variables analyzed. Even so, comparison of Incidence density by age groups showed that it was higher among patients aged 35–45 years, followed by the group aged 45 years, and lastly by that aged 15–34 years. Incidence density was higher in patients born in Spain than in those born outside Spain, among smokers, patients with risky PLOS ONE | https://doi.org/10.1371/journal.pone.0227291 January 15, 2020 4 / 18 Tuberculosis recurrences in vulnerable population Table 1. Probability of recurrence by various sociodemographic, epidemiological and clinical factors in the 2000–2016 tuberculosis cohort of Serveis Clı́nics, Catalonia. Variables Categories Cases (N = 839) N Recurrences (N = 24) No reccurrences cHR (95% P- aHR (95% P- (%) N (%) (N = 815) N (%) CI) value CI)�� value Age Median 40 (31–50) 43.5 (37–51) 40 (31–49) Age groups 15–34 296 (35.3) 3 (12.5) 293 (36.0) 1 1 35–45 263 (31.3) 11 (45.8) 252 (30.9) 4.41 (1.23– 0.023 3.90 (1.06– 0.04 15.80) 14.34) >45 280 (33.4) 10 (41.7) 270 (33.1) 3.81 (1.05– 0.042 3.88 (1.02– 0.047 13.86) 14.80) Sex Men 727 (86.7) 23 (95.8) 704 (86.4) 3.75 (0.51– 0.195 27.80) Women 112 (13.3) 1 (4.2) 111 (13.6) 1 Country of origin Spain 372 (44.3) 15 (62.5) 357 (43.8) 2.06 (0.90– 0.086 4.71) Outside Spain 467 (55.7) 9 (37.5) 458 (56.2) 1 Homelessness No 681 (81.2) 21 (87.5) 660 (81.0) 1 Yes 158 (18.8) 3 (12.5) 155 (19.0) 0.69 (0.21– 0.553 2.33) Prison or legal No 765 (91.2) 22 (91.7) 743 (91.2) 1 problems Yes 74 (8.8) 2 (8.3) 72 (8.8) 0.97 (0.23– 0.971 4.14) Smoking No 256 (30.5) 2 (8.3) 254 (31.2) 1 1 Yes 583 (69.5) 22 (91.7) 561 (68.8) 4.54 (1.07– 0.04 3.51 (0.80– 0.095 19.32) 15.37) Alcohol abuse No 460 (54.8) 9 (37.5) 451 (55.3) 1 Yes 379 (45.2) 15 (62.5) 364 (44.7) 1.93 (0.85– 0.118 4.42) Substance abuse No 668 (79.6) 18 (75.0) 650 (79.8) 1 Yes 171 (20.4) 6 (25.0) 165 (20.2) 1.29 (0.51– 0.588 3.25) HIV infection No 654 (77.9) 17 (70.8) 637 (78.2) 1 Yes 185 (22.1) 7 (29.2) 178 (21.8) 1.52 (0.63– 0.348 3.68) Hepatitis C No 698 (83.2) 18 (75.0) 680 (83.4) 1 Yes 141 (16.8) 6 (25.0) 135 (16.6) 1.76 (0.70– 0.231 4.43) Diabetes No 788 (93.9) 21 (87.5) 767 (94.1) 1 Yes 51 (6.1) 3 (12.5) 48 (5.9) 2.28 (0.68– 0.182 7.64) Pulmonary No 740 (88.2) 21 (87.5) 719 (88.2) 1 comorbidities Yes 99 (11.8) 3 (12.5) 96 (11.8) 1.16 (0.35– 0.809 3.89) Type of TB Extrapulmonary 101 (12.0) 5 (20.8) 96 (11.8) 1.90 (0.71– 0.201 2.57 (0.94– 0.065 5.10) 6.99) Pulmonary 738 (88.0) 19 (79.2) 719 (88.2) 1 1 Chest x-ray Absence of 496 (59.1) 16 (66.7) 480 (58.9) 1 cavitations Presence of 343 (40.9) 8 (33.3) 335 (41.1) 0.71 (0.30– 0.419 cavitations 1.65) Resistance No 725 (86.4) 18 (75.0) 707 (86.7) 1 1 Yes 114 (13.6) 6 (25.0) 108 (13.3) 2.26 (0.90– 0.083 2.91 (1.11– 0.031 5.71) 7.65) (Continued ) PLOS ONE | https://doi.org/10.1371/journal.pone.0227291 January 15, 2020 5 / 18 Tuberculosis recurrences in vulnerable population Table 1. (Continued) Variables Categories Cases (N = 839) N Recurrences (N = 24) No reccurrences cHR (95% P- aHR (95% P- (%) N (%) (N = 815) N (%) CI) value CI)�� value Type of resistance No 725 (86.4) 18 (2.5) 707 (97.5) 1 Rifampicin 60 (7.2) 3 (5.0) 57 (95.0) 2.09 (0.61– 0.239 7.08) Other resistances 54 (6.4) 3 (5.6) 51 (94.4) 2.48 (0.73– 0.146 8.41) Treatment length 6 months 397 (47.3) 7 (29.2) 390 (47.9) 1 9 months 238 (28.4) 11 (45.8) 227 (27.9) 2.76 (1.07– 0.036 7.11) 12 months 70 (8.3) 1 (4.2) 69 (8.5) 0.83 (0.10– 0.863 6.76) 18–24 months 134 (16.0) 5 (20.8) 129 (15.8) 2.28 (0.72– 0.16 7.18) Prior TB episodes No 673 (80.2) 15 (62.5) 658 (80.7) 1 1 Yes 166 (19.8) 9 (37.5) 157 (19.3) 2.49 (1.09– 0.031 2.12 (0.91– 0.081 5.69) 4.94) �� Schoenfeld Residuals Test chi-square = 6.13 p-value = 0.09 cHR: crude hazard ratio; aHR: adjusted hazard ratio; 95%CI: 95% confidence interval; HIV: human immunodeficiency virus; TB: tuberculosis. https://doi.org/10.1371/journal.pone.0227291.t001 alcohol intake, and HIV-infected patients, as well as among patients with pulmonary TB, patients with resistance to at least one drug, and those with a prior episode of TB (Table 2). The probability of recurrence in patients with TB admitted to Serveis Clı́nic and success- fully treated was 0.63% (95% CI: 0.26–1.51) at the end of 1 year of follow-up, and was 1.35% at 2 years (95% CI:0.73–2.49), 3.69% at 5 years (95% CI: 2.45–5.54) and 3.98% at 10 years (95% Table 2. Incidence density of recurrence among tuberculosis patients in the 2000–2016 cohort of Serveis Clı́nics, Catalonia. Variables Categories N ID per 100 PY 95%CI Global 839 0.49 0.33–0.74 Age groups 15–34 296 0.16 0.05–0.51 35–45 263 0.73 0.40–1.32 >45 280 0.66 0.35–1.22 Country of origin Spain 372 0.66 0.40–1.10 Outside Spain 467 0.35 0.18–0.66 Smoking No 256 0.15 0.04–0.59 Yes 583 0.63 0.41–0.95 Alcohol abuse No 460 0.35 0.18–0.67 Yes 379 0.65 0.39–1.08 HIV infection No 654 0.45 0.28–0.72 Yes 185 0.66 0.32–1.39 TB localization Extrapulmonary 101 0.95 0.39–2.28 Pulmonary 738 0.44 0.28–0.69 Resistances No 725 0.42 0.27–0.67 Yes 114 0.97 0.44–2.16 Prior TB episodes No 673 0.39 0.23–0.64 Yes 166 0.92 0.48–1.77 ID: incidence density; PY: person-years of follow-up; 95%CI: 95% confidence intervals; HIV: human immunodeficiency virus; TB: tuberculosis https://doi.org/10.1371/journal.pone.0227291.t002 PLOS ONE | https://doi.org/10.1371/journal.pone.0227291 January 15, 2020 6 / 18 Tuberculosis recurrences in vulnerable population Table 3. Cumulative incidence of tuberculosis recurrences at 1, 2, 5, and 10 years of follow-up. Time (years of follow-up) Risk (%) 95%CI 1 year 0.63 0.26–1.51 2 years 1.35 0.73–2.49 5 years 3.69 2.45–5.54 10 years 3.98 2.65–5.95 95%CI: 95% confidence intervals. https://doi.org/10.1371/journal.pone.0227291.t003 CI: 2.65–5.95) (Table 3). No statistically significant differences were found in recurrence risk at 5 years among the variables analyzed (Table 4). The recurrence risk was higher among smokers (p = 0.024), patients older than 34 years (p = 0.043), and those with a prior TB episode (p = 0.025) (Figs 1–6). The Schoenfeld test showed that the risks were proportional over time for all variables. On univariate analysis, the factors predictive of a subsequentTB episode were age between 35 and 45 years (cHR: 4.406 CI: 1.229–15.795) and age older than 45 years (cHR: 3.814 CI: 1.049–13.859), smoking (cHR: 4.542 CI: 1.068–19.317), having received treatment for 9 months (cHR: 2.755 CI: 1.068–7.108), and having had a prior TB episode (cHR: 2.49 CI: 1.089–5.689) (Table 1). On multivariate analysis, the factors predictive of recurrence were age older than 34 years (aHR = 3.90; CI = 1.06–14.34 in the group aged 35–45 years and aHR = 3.88; CI = 1.02–14.80 in that aged >45 years) and resistance to at least one anti-TB drug (aHR = 2.91; CI = 1.11– 7.65) (Table 1). Discussion Among the 839 TB cases in the cohort, there were 24 recurrences (2.9%), representing a rate of 0.49 per 100 person-years (95%CI: 0.33–0.74), and indicating that, for every 200 TB patients successfully treated and disease-free at 1 year, there was 1 recurrence. The factors predictive of recurrence among patients admitted to Serveis Clı́nics and successfully treated between 2000 and 2016 were age and drug resistance in the first TB episode. Considering that this study was conducted in vulnerable individuals with risk factors, we believe that the incidence density of recurrences was low in comparison with incidence rates of TB recurrence reported by other studies conducted in the city of Barcelona (0.53 per 100 person-years in the period 1995–1997 and 0.34 per 100 person-years between 2003–2006 since one year after treatment completion) [14,21], in other European regions such as England and Wales (0.41 per 100 person-years after 12 months from the initial notification from 1998– 2005) [22], New South Wales, Australia (71 per 100,000 person-years since treatment comple- tion from 1994–2006) [23] and in countries with a low TB burden (1.8 per 100 person-years at 12 months of follow-up) [6]. This low recurrence rate, which is similar to recurrence rates reported in the general popu- lation, could indicate that recurrence is not related to the sociodemographic and clinical risk factors present in the population admitted to the TB clinic of Serveis Clı́nics. That is, that patients with risk factors for poor adherence have the same recurrence risk as the general population. The incidence of TB in Catalonia during the study period oscillated between 27.6/100,000 inhabitants in the year 2000 and 13.3/100,000 inhabitants in 2016 [24]. The incidence rate for recurrence in the clinic was between 17 and 36 times higher than that for TB in the general population. As already indicated by other studies, the recurrence rate is higher than the rate of PLOS ONE | https://doi.org/10.1371/journal.pone.0227291 January 15, 2020 7 / 18 Tuberculosis recurrences in vulnerable population Table 4. Percentage of recurrence risk according to various clinical-epidemiological factors at 5 years of follow-up among tuberculosis patients in the 2000–2016 cohort, Serveis Clı́nics, Catalonia. Variables Categories Recurrences N = 24 (%) Recurrence risk at 5 years (%) 95%CI Age groups (years) 15–34 3 (12.5) 1.55 0.49–4.83 35–45 11 (45.8) 5.61 3.11–9.99 >45 10 (41.7) 4.21 2.18–8.03 Sex Men 23 (95.8) 4.1 2.69–6.21 Women 1 (4.2) 1.12 0.16–7.71 Country of origin Spain 15 (62.5) 4.76 2.82–7.97 Outside Spain 9 (37.5) 2.83 1.47–5.43 Homelessness No 21 (87.5) 3.77 2.43–5.81 Yes 3 (12.5) 3.49 1.11–10.7 Prison or legal problems No 22 (91.7) 3.73 2.43–5.72 Yes 2 (8.3) 3.08 0.77–11.82 Smoking No 2 (8.3) 1.44 0.35–5.8 Yes 22 (91.7) 4.62 3.02–7.04 Alcohol abuse No 9 (37.5) 2.57 1.28–5.12 Yes 15 (62.5) 5 3.02–8.24 Substance abuse No 18 (75.0) 3.4 2.11–5.46 Yes 6 (25.0) 4.85 2.16–10.71 HIV infection No 17 (70.8) 3.22 1.97–5.25 Yes 7 (29.2) 5.47 2.59–11.37 Hepatitis C No 18 (75.0) 3.24 2.01–5.19 Yes 6 (25.0) 6.2 2.74–13.74 Diabetes No 21 (87.5) 3.4 2.19–5.26 Yes 3 (12.5) 7.87 2.56–22.86 Pulmonary comorbidities No 21 (87.5) 3.62 2.33–5.61 Yes 3 (12.5) 4.09 1.32–12.28 TB localization Extrapulmonary 5 (20.8) 7.77 3.23–18.07 Pulmonary 19 (79.2) 3.13 1.97–4.96 Chest x-ray Absence of 16 (66.7) 4.6 2.81–7.48 cavitations Presence of 8 (33.3) 2.45 1.17–5.1 cavitations Resistance No 18 (75.0) 3.17 1.96–5.11 Yes 6 (25.0) 7.32 3.35–15.63 Treatment length 6 months 7 (29.2) 1.73 0.78–3.84 9 months 11 (45.8) 7.04 3.91–12.53 12 months 1 (4.2) 1.79 0.25–12.01 18–24 months 5 (20.8) 4.86 2.03–11.39 Prior TB episodes No 15 (62.5) 2.87 1.69–4.85 Yes 9 (37.5) 7.01 3.67–13.18 95%CI: 95% confidence intervals; HIV: human immunodeficiency virus; TB: tuberculosis. https://doi.org/10.1371/journal.pone.0227291.t004 new TB cases [7]. A review by Rosser et al. also indicates that patients cured of a TB episode have a higher risk of TB than those without prior infection. These authors estimated that the TB rate was 31.5 times higher in recurrences than in new TB cases [5]. Recurrences include both endogenous reactivations and exogenous reinfections. Other studies performed in areas with a low burden of the disease have concluded that relapses of a PLOS ONE | https://doi.org/10.1371/journal.pone.0227291 January 15, 2020 8 / 18 Tuberculosis recurrences in vulnerable population Fig 1. Kaplan-Meier curves of tuberculosis recurrence risk among all tuberculosis patients in the 2000–2016 cohort of Serveis Clı́nics, Catalonia. https://doi.org/10.1371/journal.pone.0227291.g001 prior infection continue to be the most common cause of TB recurrences. Infections with new Mycobacterium tuberculosis strains are possible but infrequent in low endemic areas [5,12,23,25,26]. Nevertheless, two studies conducted in Spain highlight the importance of TB infections in contexts with a low or moderate incidence [27,28]. El Sahly et al. also concluded that reinfection with a new M. tuberculosis strain caused a significant proportion of TB recur- rences in an area with a low incidence [29]. The possibility of exogenous reinfection increases when the prevalence of the disease is higher [8,30,31]. In this study, factors predictive of recurrence were age between 35 and 45 years, age older than 45 years, and resistance to at least one anti-TB drug in the prior TB episode. Other studies have reported a relationship between age and TB recurrences, although the results were not homogeneous. A study in England and Wales reported that children aged 0–14 years had a lower recurrence risk than patients aged 15–44 years [22]. In Brazil, the risk was lower in HIV-infected patients aged 40–49 years than in those aged <30 years [32]. In a 13-year cohort of TB cases in the United States, recurrence risk was higher in patients � 65 years than in those aged <45 years [33]. A study performed in South Carolina also identified PLOS ONE | https://doi.org/10.1371/journal.pone.0227291 January 15, 2020 9 / 18 Tuberculosis recurrences in vulnerable population Fig 2. Kaplan-Meier curves of tuberculosis recurrence risk among all tuberculosis patients in the 2000–2016 cohort of Serveis Clı́nics, Catalonia, according to age group (p value Log rank test = 0.043). https://doi.org/10.1371/journal.pone.0227291.g002 age > 46 years as a risk factor for recurrence [34]. In 1980 in the United States, TB recurrences were more common among patients aged > 30 years [35]. In South Africa, an association was found between advanced age at the first diagnosis of TB and recurrence in HIV-negative patients [36]. In addition to age, the other predictive factor for recurrence among TB patients treated in Serveis Clı́nics was resistance to at least one anti-TB drug. In China and Vietnam, an associa- tion was found between MDR and TB recurrence [37,38], while TB recurrences among South African miners were associated with anti-TB drug resistance [39]. In Uzbekistan, resistance to treatment was also found to substantially contribute to TB recurrences [40]. In southern India, isoniacide and/or rifamicin resistance was a predictive factor for recurrence among pulmonary TB patients under DOT [41]. In areas of low incidence, resistant TB was also identified as a risk factor for recurrence [5]. In California, pyrazinamide monoresistance was associated with TB recurrences [42]. Two studies performed in Italy showed a higher risk of recurrence versus reinfection in patients infected with MDR TB [25,43]. A meta-analysis concluded that one of the globally recognized PLOS ONE | https://doi.org/10.1371/journal.pone.0227291 January 15, 2020 10 / 18 Tuberculosis recurrences in vulnerable population Fig 3. Kaplan-Meier curves of tuberculosis recurrence risk among all tuberculosis patients in the 2000–2016 cohort of Serveis Clı́nics, Catalonia, according to country of origin (p value Log rank test = 0.08). https://doi.org/10.1371/journal.pone.0227291.g003 risk factors for TB recurrence was multidrug resistance [44]. Moreover, among patient with resistant TB, recurrences were associated with a higher number of drugs and with resistance to all injectable drugs [45]. Among lifestyle factors, although smoking was not a predictive factor for recurrence in this study, smokers had a higher recurrence risk, with a lower CI limit close to 1, which was not sig- nificant, possibly due to a lack of statistical power. In this regard, a study conducted in Taipei concluded that the risk of TB recurrence among persons smoking >10 cigarettes/day was twice that of nonsmokers and exsmokers [46]. Another study performed in the United Kingdom found that smoking was a predictive factor for TB recurrence [47]. In Yemen, patients smoking > 20 cigarrettes/day had a higher risk of TB recurrence than those smoking <20 cigarrettes/day [48]. In South India, smoking was a predictive factor for recurrence [41]. Few studies have reported an association between treatment length and recurrences. While a study conducted in Hong Kong showed that prolonging both the intensive phase of treat- ment and general treatment by 50% or more protected against recurrence [49], another sutdy PLOS ONE | https://doi.org/10.1371/journal.pone.0227291 January 15, 2020 11 / 18 Tuberculosis recurrences in vulnerable population Fig 4. Kaplan-Meier curves of tuberculosis recurrence risk among all tuberculosis patients in the 2000–2016 cohort of Serveis Clı́nics, Catalonia, according to smoking (p value Log rank test = 0.024). https://doi.org/10.1371/journal.pone.0227291.g004 conducted in South Carolina found an interraction between treatment length longer than 24 months and poor adherence, which was a predictive factor for recurrence [34]. In this line, Thomas et al. concluded that patients taking their medication irregularly had twice the proba- bility of experiencing a relapse (aOR = 2.5; 95%CI 1.4–4.6) [41]. This could be a cause of the higher number of recurrences found in patients receiving treatment lasting for 9 months. Few studies have analyzed the effect of a prior TB episode. A case-control study conducted in Vietnam concluded that one of the factors associated with recurrence was having reported a prior history of TB [38]. A previous study conducted in Barcelona reported that a history of TB treatment before study inclusion was a risk factor for recurrence [14]. Using Kaplan-Meier curves, this study also revealed a tendency by country of origin, with the percentage of recurrences being higher among Spanish-born patients. In line with these results, a study conducted in California found that US-born patients had a higher recurrence risk than those born outside the US [42]. In contrast, other studies reported an association between foreign-born patients or immigrants and TB recurrences [21,47,50]. It has been reported that immigrants usually have a higher association with TB recurrence caused by PLOS ONE | https://doi.org/10.1371/journal.pone.0227291 January 15, 2020 12 / 18 Tuberculosis recurrences in vulnerable population Fig 5. Kaplan-Meier curves of tuberculosis recurrence risk among all tuberculosis patients in the 2000–2016 cohort of Serveis Clı́nics, Catalonia, according to prior TB episode (p value Log rank test = 0.025). https://doi.org/10.1371/journal.pone.0227291.g005 reinfection [25,43]. In the present study, some recurrences may not have been detected in patients born outside Spain if they left Catalonia or returned to their country of origin. Finally, patients with a disease or receiving treatment altering immune system functioning are more susceptible to TB recurrence. However, in this study, no significant association was found between diseases or treatments that could compromise the immune system and TB recurrence. This study could show a selection bias because we could not follow-up persons without a health card number and they were excluded from the study (S1 Table). These persons could have had different characteristics from those included in the study and more risk factors for recurrence. Comparison of sociodemographic, epidemiological and clinical factors between included and excluded persons revealed differences in the following variables: country of birth (the percentage of individuals born outside Spain was higher among excluded patients), home- lessness, having been in prison or having problems with the law, treatment length, and prior TB treatment. Therefore, we may have underestimated some associations between these risk factors and TB recurrences. PLOS ONE | https://doi.org/10.1371/journal.pone.0227291 January 15, 2020 13 / 18 Tuberculosis recurrences in vulnerable population Fig 6. Kaplan-Meier curves of tuberculosis recurrence risk among all tuberculosis patients in the 2000–2016 cohort of Serveis Clı́nics, Catalonia, according to resistance (p value Log rank test = 0.075). https://doi.org/10.1371/journal.pone.0227291.g006 We may also have missed recurrences that could have occurred outside Catalonia in those patients who moved during treatment, at the end of treatment, or during follow-up. For these patients we only have information on the final follow-up date and their status just before they left Catalonia. Nevertheless, a strength of the study is that we were able to detect all recurrences that occurred not only in Serveis Clı́nics but also throughout Catalonia. Another limitation is the retrospective cohort design that did not consider variability over time among the independent variables. The study was based on the baseline variables of each patient gathered during the first TB episode. Finally, we could not ascertain whether recur- rences were due to endogenous reactivation or exogenous reinfection. The study cohort included all patients admitted to Serveis Clı́nics over a 16-year period and with a long follow-up. The data sources are both reliable and diverse, allowing us to compare and contrast the distinct sources and complete the database. To do this, we performed broad, in-depth field work. We assessed the data quality, including the search and identification of missing data to complete the database. PLOS ONE | https://doi.org/10.1371/journal.pone.0227291 January 15, 2020 14 / 18 Tuberculosis recurrences in vulnerable population Another strength of the study is its setting, especially the good functioning of the TB Pro- gram of Barcelona and Catalonia and Serveis Clı́nics, which is specialized in TB, has more than 25 years’ experience, and closely follows patients up until cure. DOT is carried out daily in all patients. Conclusions This study shows that patients successfully treated in a TB referral clinic specialized in vulnera- ble individuals have a not neglectablerecurrence risk, which is higher than that of TB disease in the general population. More specifically, patients with older age and with resistance seem to be at particularly high risk. These findings support the importance of close follow-up of per- sons with a prior TB episode and of devoting surveillance resources to these individuals because, despite successful treatment, they have a high risk of recurrent disease. Supporting information S1 Fig. Flowchart of tuberculosis patients in Serveis Clı́nics, Catalonia, 2000–2016. (DOCX) S1 Table. Sociodemographic and clinical characteristics of excluded and included tubercu- losis patients in the 2000–2016 cohort, Serveis Clı́nics, Catalonia. (DOCX) Acknowledgments The authors would like to thank all the health care and administrative staff of Serveis Clı́nics, and Directly Observed Outpatient Treatment Team (ETODA), of the Epidemiology Service, Barcelona Public Health Agency (PHAB) and the Catalonian Public Health Agency and all physicians and TB nurses from the Catalonia hospitals, epidemiological surveillance units and primary healthcare centers. Without them, this work would not have been possible. Author Contributions Conceptualization: Vinicio-Israel Molina, Joan-Pau Millet. Data curation: Sı́lvia Brugueras, Vinicio-Israel Molina, Xavier Casas, Yoel-Domingo Gonzá- lez, Nuria Forcada, Dora Romero, Anna Rodés, Maria-Neus Altet, José Maldonado, Joan A. Caylà, Àngels Orcau, Cristina Rius, Joan-Pau Millet. Formal analysis: Sı́lvia Brugueras, Mario Martin-Sánchez. Funding acquisition: Vinicio-Israel Molina, Joan-Pau Millet. Investigation: Sı́lvia Brugueras, Vinicio-Israel Molina, Xavier Casas, Yoel-Domingo González, Nuria Forcada, Dora Romero, Anna Rodés, Maria-Neus Altet, José Maldonado, Mario Martin-Sánchez, Joan A. Caylà, Àngels Orcau, Cristina Rius, Joan-Pau Millet. Methodology: Sı́lvia Brugueras, Mario Martin-Sánchez, Joan A. Caylà, Àngels Orcau, Cristina Rius, Joan-Pau Millet. Project administration: Sı́lvia Brugueras, Joan-Pau Millet. Resources: Sı́lvia Brugueras, Vinicio-Israel Molina, Xavier Casas, Yoel-Domingo González, Nuria Forcada, Dora Romero, Anna Rodés, Maria-Neus Altet, José Maldonado, Àngels Orcau, Cristina Rius, Joan-Pau Millet. PLOS ONE | https://doi.org/10.1371/journal.pone.0227291 January 15, 2020 15 / 18 Tuberculosis recurrences in vulnerable population Supervision: Sı́lvia Brugueras, Xavier Casas, Maria-Neus Altet, Joan A. Caylà, Àngels Orcau, Cristina Rius, Joan-Pau Millet. Validation: Sı́lvia Brugueras, Xavier Casas, Maria-Neus Altet, Joan A. Caylà, Àngels Orcau, Cristina Rius, Joan-Pau Millet. Visualization: Sı́lvia Brugueras, Vinicio-Israel Molina, Joan-Pau Millet. Writing – original draft: Sı́lvia Brugueras, Vinicio-Israel Molina, Xavier Casas, Joan-Pau Millet. Writing – review & editing: Sı́lvia Brugueras, Vinicio-Israel Molina, Xavier Casas, Yoel- Domingo González, Nuria Forcada, Dora Romero, Anna Rodés, Maria-Neus Altet, José Maldonado, Mario Martin-Sánchez, Joan A. Caylà, Àngels Orcau, Cristina Rius, Joan-Pau Millet. References 1. WHO. Global tuberculosis report 2018. Geneva; 2018. 2. Santı́n Cerezales M, Navas Elorza E. Tuberculosis in special populations. Enferm Infecc Microbiol Clin. 2011; 29(Supl 1):20–5. 3. Caylà J a, Orcau A. Control of tuberculosis in large cities in developed countries: an organizational prob- lem. BMC Med [Internet]. 2011 Jan [cited 2014 Dec 18]; 9(1):127. Available from: http://www. pubmedcentral.nih.gov/articlerender.fcgi?artid=3283473&tool=pmcentrez&rendertype=abstract 4. Lönnroth K, Migliori GB, Abubakar I, Ambrosio LD, Vries G De, Diel R, et al. action framework for low- incidence countries.:928–52. 5. Rosser A, Marx FM, Pareek M. Recurrent tuberculosis in the pre-elimination era. Int J Tuberc Lung Dis. 2018; 22(2):139–50. https://doi.org/10.5588/ijtld.17.0590 PMID: 29506610 6. Panjabi R, Comstock GW, Golub JE. Recurrent tuberculosis and its risk factors: adequately treated patients are still at high risk. 2007; 11(October 2006):828–37. 7. Verver S, Warren RM, Beyers N, Richardson M, Spuy GD Van Der, Borgdorff MW, et al. Rate of Rein- fection Tuberculosis after Successful Treatment Is Higher than Rate of New Tuberculosis. 2005; 8. Lambert M, Hasker E, Deun A Van, Roberfroid D, Boelaert M, Stuyft P Van Der. Recurrence in tubercu- losis: relapse or reinfection? 2003; 3(May):282–7. 9. Red Nacional de Vigilancia Epidemiológica. Ampliación de la definición de caso de tuberculosis en la red nacional de vigilancia epidemiológica. 2003. 10. Centers for Disease Control and Prevention CDC. Tuberculosis 2009 Case Definition. 2009. 11. Parvaresh L, Crighton T, Martinez E, Bustamante A, Chen S. Recurrence of tuberculosis in a low- inci- dence setting: a retrospective cross- sectional study augmented by whole genome sequencing. 2018;4–9. 12. Jasmer RM, Bozeman L, Schwartzman K, Cave MD, Saukkonen JJ, Metchock B, et al. Recurrent Tuberculosis in the United States and Canada Relapse or Reinfection? 2004; 170:1360–6. https://doi. org/10.1164/rccm.200408-1081OC PMID: 15477492 13. Pettit AC, Kaltenbach LA, Maruri F, Cummins J, Smith TR, Warkentin JV, Griffin MR ST. Chronic lung disease and HIV infection are risk factors for recurrent tuberculosis in a low-incidence setting. 2011; 15 (July 2011):906–11. 14. Millet J-P, Shaw E, Orcau A, Casals M, Miró JM, Caylà J a. Tuberculosis recurrence after completion treatment in a European city: reinfection or relapse? PLoS One [Internet]. 2013 Jan [cited 2014 Dec 8]; 8(6):e64898. Available from: http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid= 3679149&tool=pmcentrez&rendertype=abstract https://doi.org/10.1371/journal.pone.0064898 PMID: 23776440 15. Caylà J a, Rodrigo T, Ruiz-Manzano J, Caminero J a, Vidal R, Garcı́a JM, et al. Tuberculosis treatment adherence and fatality in Spain. Respir Res [Internet]. 2009 Jan [cited 2014 Dec 18]; 10:121. Available from: http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=2794858&tool= pmcentrez&rendertype=abstract https://doi.org/10.1186/1465-9921-10-121 PMID: 19951437 16. Alsedà Graells M, Jansà López del Vallado J. Protocol per a la prevenció i el control de la tuberculosi. Documents de vigilància epidemiológica; Tuberculosi—Prevenció. [Internet]. 2008 [cited 2019 Apr 18]. Available from: www.gencat.cat/salut PLOS ONE | https://doi.org/10.1371/journal.pone.0227291 January 15, 2020 16 / 18 Tuberculosis recurrences in vulnerable population 17. World Health Organization (WHO). Treatment of tuberculosis: Guidelines 4th edition. [Internet]. Geneva: World Health Organization. 2003 [cited 2019 Apr 18]. Available from: https://apps.who.int/iris/ bitstream/handle/10665/44165/9789241547833_eng.pdf;jsessionid=E9979DA618E9A71FF6C0E 42AE5628257?sequence=1 18. Declaración de Helsinki de la AMM. Principios éticos para las investigaciones médicas en seres huma- nos. Asoc Médica Mund [Internet]. 2013; Available from: http://www.isciii.es/ISCIII/es/contenidos/fd- investigacion/fd-evaluacion/fd-evaluacion-etica-investigacion/Declaracion-Helsinki-2013-Esp.pdf 19. Ley Orgánica 3/2018, de 5 de diciembre, de Protección de Datos Personales y garantı́a de los derechos digitales [Internet]. Madrid: Noticias jurı́dicas BOE núm. 294. 2018 [cited 2019 Apr 18]. Available from: http://noticias.juridicas.com/base_datos/Laboral/632849-lo-3-2018-de-5-dic-proteccion-de-datos- personales-y-garantia-de-los-derechos.html 20. Regulation (EU) 2016/679 of the European Parliament and of the Council of 27 April 2016 on the protec- tion of natural persons with regard to the processing of personal data and on the free movement of such data, and repealing Directive 95/46/EC (General Da [Internet]. Official Journal of the European Union Official Journal of the European Union, Vol. L119. 2016 [cited 2019 Apr 18]. p. 1–88. Available from: https://publications.europa.eu/en/publication-detail/-/publication/3e485e15-11bd-11e6-ba9a- 01aa75ed71a1/language-en 21. Millet J-P, Orcau À, Garcı́a de Olalla P, Casals M, Rius C, Cayla JA. Tuberculosis recurrence and its associated risk factors among successfully treated patients. 2009; 63(10):799–804. 22. Crofts JP, Andrews NJ, Barker RD, Delpech V, Abubakar I. Risk factors for recurrent tuberculosis in England and Wales, 1998–2005. Thorax. 2010; 65(4):310–4. https://doi.org/10.1136/thx.2009.124677 PMID: 20388755 23. Dobler C, Crawford A, Jelfs P, Gilbert G, Marks G. Recurrence of tuberculosis in a low-incidence set- ting. Eur respir J. 2009; 33(1):160–7. https://doi.org/10.1183/09031936.00104108 PMID: 18829676 24. Rodés Monegal A, López Espinilla M, Garcı́a Lebrón M. Informe anual 2016. Situació epidemiològica i tendència de l’endèmia tuberculosa a Catalunya [Internet]. 2018 [cited 2019 Apr 18]. Available from: http://salutpublica.gencat.cat/web/.content/minisite/aspcat/vigilancia_salut_publica/20180226_ informe_anual_tuberculosi_2016.pdf 25. Schiroli C, Carugati M, Zanini F, Bandera A, Di S, Stuppino N, et al. Exogenous reinfection of tuberculo- sis in a low - burden area. Infection. 2015; 43:647–53. https://doi.org/10.1007/s15010-015-0759-9 PMID: 25754899 26. De Boer AS, Borgdorff MW, Vynnycky E, Sebek MM, Van Soolingen D. Exogenous re-infection as a cause of recurrent tuberculosis in a low-incidence area. Int J Tuberc Lung Dis. 2003; 7(2):145–52. PMID: 12588015 27. Caminero J, Pena M, Campos-Herrero M, Rodrı́guez J, Afonso O, Martin C, et al. Exogenous Reinfec- tion with Tuberculosis on a European island with a moderate incidence of disease. 2001; 163(3):717– 20. 28. Garcı́a de Viedma D, Marı́n M, Hernangómez S, Dı́az M, Ruiz Serrano M, Alcalá L, et al. Tuberculosis Recurrences. Reinfection plays a role in a population whose clinical/epidemiological characteristics do not favor reinfection. 2002; 162(16):1873–9. 29. El Sahly HM, Wright JA, Soini H, Bui TT, Escalante P. Recurrent tuberculosis in Houston, Texas: a pop- ulation-based study. Int J Tuberc Lung Dis. 2004; 8(3):333–40. PMID: 15139472 30. Chiang CY, Riley L. Exogenous reinfection in tuberculosis. Lancet Infect Dis. 2005; 5:629–36. https:// doi.org/10.1016/S1473-3099(05)70240-1 PMID: 16183517 31. Wang J, Lee L, Lai H, Hsu H, Liaw Y, Hsueh P, et al. Prediction of the Tuberculosis Reinfection Propor- tion from the Local Incidence. J Infect Dis [Internet]. 2007; 196(2):281–8. Available from: https:// academic.oup.com/jid/article-lookup/doi/10.1086/518898 PMID: 17570116 32. Golub JE, Durovni B, King BS, Cavalacante SC, Pacheco AG, Moulton LH, et al. Recurrent tuberculosis in HIV-infected patients in Rio de Janeiro, Brazil. Aids. 2008; 22(18):2527–33. https://doi.org/10.1097/ QAD.0b013e328311ac4e PMID: 19005276 33. Kim L, Moonan PK, Heilig CM, Woodruff RSY, Steve J, Haddad MB, et al. Factors associated with recurrent tuberculosis more than 12 months after treatment completion. Int J Tuberc Lung Dis. 2016; 20 (1):49–56. https://doi.org/10.5588/ijtld.15.0442 PMID: 26688528 34. Selassie AW, Pozsik C, Wilson D, Ferguson PL. Why pulmonary tuberculosis recurs: A population- based epidemiological study. Ann Epidemiol. 2005; 15(7):519–25. https://doi.org/10.1016/j.annepidem. 2005.03.002 PMID: 15921928 35. Centers for Disease Control and Prevention. Patients with Recurrent Tuberculosis. Morb Mortal Wkly Rep. 1982; 30(52):645–7. PLOS ONE | https://doi.org/10.1371/journal.pone.0227291 January 15, 2020 17 / 18 Tuberculosis recurrences in vulnerable population 36. van der Heijden YF, Karim F, Chinappa T, Mufamadi G, Zako L, Shepherd BE, et al. Older age at first tuberculosis diagnosis is associated with tuberculosis recurrence in HIV-negative persons. Int J Tuberc Lung Dis. 2018; 22(8):871–7. https://doi.org/10.5588/ijtld.17.0766 PMID: 29991395 37. Sun Y, Harley D, Vally H, Sleigh A. Impact of multidrug resistance on tuberculosis recurrence and long- term outcome in China. PLoS One. 2017; 12(1):1–11. 38. Bestrashniy JRBM Nguyen VN, Nguyen TL Pham TL, Nguyen TA Pham DC, et al. Recurrence of tuber- culosis among patients following treatment completion in eight provinces of Vietnam: A nested case- control study. Int J Infect Dis [Internet]. 2018; 74:31–7. Available from: https://doi.org/10.1016/j.ijid. 2018.06.013 PMID: 29944930 39. Mallory KF, Churchyard GJ, Kleinschmidt I, De Cock KM, Corbett EL. The impact of HIV infection on recurrence of tuberculosis in South African gold miners. Int J Tuberc Lung Dis. 2000; 4(5):455–62. PMID: 10815740 40. Cox H, Kebede Y, Allamuratova S, Ismailov G, Davletmuratova Z, Byrnes G, et al. Tuberculosis recur- rence and mortality after successful treatment: Impact of drug resistance. PLoS Med. 2006; 3 (10):1836–43. 41. Thomas A, Gopi P, Santha T, Chandrasekaran V, R S, N S, et al. Predictors of relapse among pulmo- nary tuberculosis patients treated in a DOTS programme in South India. Int J Tuberc Lung Dis [Internet]. 2005; 9(5):556–61. Available from: http://openurl.ingenta.com/content?genre=article&genre= article&genre=article&date=2005-05&issn=1027-3719&eissn=1815-7920&atitle=Predictors+of +relapse+among+pulmonary+tuberculosis+patients+treated+in+a+DOTS+programme+in+South +India.&volume=9&issue=5&spage= PMID: 15875929 42. Pascopella L, DeRiemer K, Watt JP, Flood JM. When tuberculosis comes back: Who develops recur- rent tuberculosis in California? PLoS One. 2011; 6(11). 43. Bandera A, Gori A, Catozzi L, Degli Esposti A, Marchetti G, Molteni C, et al. Molecular epidemiology study of exogenous reinfection in an area with a low incidence of tuberculosis. J Clin Microbiol. 2001; 39 (6):2213–8. https://doi.org/10.1128/JCM.39.6.2213-2218.2001 PMID: 11376059 44. Schiroli C, Franzetti F. La tubercolosi ricorrente: Recidiva o reinfezione esogena? Infez Med. 2013; 21 (4):251–60. PMID: 24335455 45. Blöndal K, Viiklepp P, Guomundsson LJ, Altraj A. Predictors of recurrence of multidrug-resistant and extensively drug-resistant tuberculosis. Int J Tuberc Lung Dis. 2012; 16(9):1228–33. https://doi.org/10. 5588/ijtld.12.0037 PMID: 22748131 46. Yen YF, Yen MY, Lin YS, Lin YP, Shih HC, Li LH, et al. Smoking increases risk of recurrence after suc- cessful anti-tuberculosis treatment: A population-based study. Int J Tuberc Lung Dis. 2014; 18(4):492– 8. https://doi.org/10.5588/ijtld.13.0694 PMID: 24670708 47. Rosser A, Richardson M, Wiselka MJ, Free RC, Woltmann G, Mukamolova G V., et al. A nested case- control study of predictors for tuberculosis recurrence in a large UK Centre. BMC Infect Dis. 2018; 18 (1):1–9. https://doi.org/10.1186/s12879-017-2892-9 48. Anaam MS, Ibrahim MIM, Al Serouri AW, Bassili A, Aldobhani A. A nested case-control study on relapse predictors among tuberculosis patients treated in Yemen’s NTCP. Public Heal Action. 2012; 2(4):168– 73. 49. Chang KC, Leung CC, Yew WW, Ho SC, Tam CM. A nested case-control study on treatment-related risk factors for early relapse of tuberculosis. Am J Respir Crit Care Med. 2004; 170(10):1124–30. https://doi.org/10.1164/rccm.200407-905OC PMID: 15374844 50. Interrante JD, Haddad MB, Kim L, Gandhi NR. Exogenous reinfection as a cause of late recurrent tuber- culosis in the United States. Ann Am Thorac Soc. 2015; 12(11):1619–26. https://doi.org/10.1513/ AnnalsATS.201507-429OC PMID: 26325356 PLOS ONE | https://doi.org/10.1371/journal.pone.0227291 January 15, 2020 18 / 18