TB is one of the most deathly infections worldwide, affecting disproportionally people living with HIV (PLHIV). Furthermore, HIV co-infection is related to worse outcomes for TB patients, including lower treatment success.
MethodsUsing surveillance records of all TB cases notified in Barcelona city from 2001 to 2021, we analyzed TB treatment success according to HIV status. Additionally, we explored potential social and health related and factors associated to unsuccessful treatment in PLHIV, using multiple regression analyses.
ResultsOut of the 8406 new TB cases diagnosed during the study period, 9% were co-infected with HIV. According to our regression models, PLHIV were more frequently men, users of injected drugs (aOR=45.81; 95% CI (33.10–64.26)), had previously been treated for TB (aOR=1.77; 95% CI (1.30–2.40)) and had a lower rate of contact tracing (aOR=0.51; 95% CI (0.40–0.64)). Among PLHIV, unsuccessful treatment was related to the use of injected drugs and homelessness, but it was lower for those who had undergone contact tracing.
ConclusionPLHIV have higher odds of unsuccessful TB treatment, especially those who are homeless and use injected drugs. Contact tracing improved treatment success, calling for further efforts and resources to correctly follow-up on these patients, with the goal of increasing treatment success.
La tuberculosis (TB) es una de las infecciones más mortales en todo el mundo, afectando de manera desproporcionada a las personas que viven con VIH (PVVIH). Además, la coinfección por VIH está relacionada con peores resultados para los pacientes con TB, incluyendo una menor tasa de éxito en el tratamiento.
MétodosUtilizando registros de vigilancia de todos los casos de TB notificados en la ciudad de Barcelona desde 2001 hasta 2021, analizamos el éxito del tratamiento de TB según el estado de VIH. Adicionalmente, exploramos factores sociales y relacionados con la salud asociados al tratamiento no exitoso en PVVIH, utilizando análisis de regresión múltiple.
ResultadosDe los 8,406 nuevos casos de TB diagnosticados durante el periodo del estudio, el 9% estaban coinfectados con VIH. Según nuestros modelos de regresión, las PVVIH eran más frecuentemente hombres, usuarios de drogas inyectadas (aOR=45.81; 95% IC (33.10-64.26)), habían sido tratados previamente por TB (aOR=1.77; 95% CI (1.30-2.40)) y tenían una tasa de rastreo de contactos más baja (aOR=0.51; 95% CI (0.40-0.64)). Entre las PVVIH, el tratamiento no exitoso estaba relacionado con el uso de drogas inyectadas y la falta de hogar, pero era menor para aquellos que habían sido objeto de rastreo de contactos.
ConclusiónLas PVVIH tienen mayores probabilidades de un tratamiento no exitoso de TB, especialmente aquellos que son personas sin hogar y que usan drogas inyectadas. El rastreo de contactos mejoró el éxito del tratamiento, lo que requiere más esfuerzos y recursos para dar un seguimiento adecuado a estos pacientes, con el objetivo de aumentar el éxito en el tratamiento.
Tuberculosis (TB) was the second leading cause of death from a single agent worldwide in 2022, only after COVID-19, and caused almost twice as many deaths as HIV/AIDS.1 In the European Region an estimated 230000 people fell ill with TB in 2021, which accounts for 25 cases per 100000 people.2
People living with HIV (PLHIV) are a high-risk group for developing TB, which is one of the AIDS-defining illnesses.3 PLHIV are 16 times more likely to fall ill with TB disease than people without HIV.4 Worldwide, 6.3% of all TB cases notified in 2022 were HIV-positive, a proportion that has been steadily declining for several years, according to the WHO.1 Nevertheless, in the European Region this percentage raised to a 13% in 2021, in line with previous years.2
Despite limited treatment efficacy and availability, successful TB treatment and antiretroviral treatment in people with TB/HIV co-infection have averted an estimated of 6.4 million deaths worldwide between 2010 and 2022, and more than 200000 deaths only in the WHO European Region.1 Still, TB is the leading cause of death among people living with HIV. In 2022, about 167000 people died of HIV-associated TB worldwide, 20000 only in the European Region.2 Therefore, prevention and TB control activities are the most important measures needed to reduce morbidity and mortality among PLHIV.4
Several studies in both general and institutionalized populations have identified HIV status to be one of the main factors for TB treatment failure.5,6 However, there is still a lack of knowledge on potential associated factors of treatment failure among HIV-positive TB patients. Moreover, the profile of this specific population might have changed through time due to the several reasons, including: harm reduction strategies,7,8 better prepared health systems with advances in treatment and prevention9,10 and improved surveillance, together with notable demographic changes.11
In the city of Barcelona (Catalonia, Spain), the TB program of the Agència de Salut Pública de Barcelona (ASPB) is in charge of surveillance and control of the disease since 1987 and keeps extensive electronic records on comorbidities and treatment outcomes since 2001.
In order to explore the epidemiological trends and changes in HIV-positive TB cases characteristics in Barcelona city, we conducted a study covering all cases registered during the span of 22 years.
MethodsStudy design and populationWe performed a population based observational study using the cohort of incident TB cases registered in the city of Barcelona, Spain, from January 1st 2000 to December 31st 2021. We included all cases notified to the TB Prevention and Control Program who started TB treatment during this period. National and international guidelines of TB case definition were used.12
Data sourcesWe obtained all data from the electronic TB records of the ASPB. These records include sociodemographic and health related information for all TB cases declared in the city of Barcelona and are regularly updated with incident cases. These records are used for epidemiological surveillance and control. Information is registered for patient follow-up and contact tracing until end of treatment of the case and contacts. Cases are then considered “closed”, as well as if they either die or are lost to follow-up before finishing successful treatment. Individual-level data is retrieved from the epidemiologic survey, which is filled by a nurse of the Epidemiology Unit with information provided by the patient, healthcare professionals and clinical records. TB records were cross-referenced with the ASPB official HIV records to ensure all PLHIV that fit the inclusion criteria were considered.
Income level is indirectly obtained from an aggregate disposable household income index, according to the neighborhood of residence and categorized into quartiles (low, medium-low, medium-high and high income level). The index was yearly updated up to 2019. For the following years, neighborhoods were assigned to their 2019 quartiles.
Outcomes and relevant variablesThe main outcome of our study was unsuccessful TB treatment, guided by the opposite “Treatment success”, defined by the WHO's as “The sum of cured and treatment completed”.13 Therefore, we included death, lost to follow-up, failure or transfer as motives of unsuccessful treatment.
We considered several sociodemographic co-variables that were closely related to the epidemiology of TB and HIV, such as age (0–24; 25–64; 65 or older), sex (women-men), prison background (no-yes), homelessness (no-yes), income level and country of birth. As for clinical variables, we studied health-related habits (tobacco, alcohol use, use of injected drugs), diabetes, type of TB, chest X-ray, resistances to TB treatment, previous TB treatment, directly observed therapy, days of diagnostic delay in TB, contact tracing and final clinical result.
Statistical analysesWe estimated unweighted frequencies and percentages for all the considered variables for patients with and without HIV co-infection. Differences between these two groups were tested using χ2 and Wilcoxon test for median of days of diagnostic delay.
To identify associated factors that could explain unsuccessful TB treatment, we performed a logistic regression model for patients with HIV co-infection, to estimate adjusted odds ratios (aOR) of unsuccessful treatment for each variable considered. The level of statistical significance was set to α=0.05. Only variables that showed statistical significance in the bivariate models and were of epidemiological interest for the adjustment were considered for the multivariate analyses (age, sex, person who injects drugs, contact tracing and homelessness).
The statistical analysis was carried out using the RStudio software version 4.2.2.
Ethical considerationsTo guarantee confidentiality of the data and records, we adhered to the regulations established by the Organic Law on the Protection of Personal Data 03/18 of December 5 in Spain, the European data protection law 2018/1725 and to the ethical principles for human research defined by the Helsinki Declaration of 1964, revised and updated by the World Medical Organization (Edinburgh 2000). Data were collected for public health surveillance purposes and this study is the result of this routine epidemiological surveillance, using aggregated data, therefore the study did not require ethical approval nor is it possible to obtain it.
ResultsOut of the 8406 new TB cases diagnosed during these 22 years, our study focused on those patients who had HIV co-infection. Fig. 1 shows the evolution of HIV prevalence among new TB cases from 2000 to 2021 in Barcelona city. Though irregular, a significant decrease in HIV-positive TB incidence was observed during this time, changing from a 14.8% of co-infection in 2000 to 5.2% in 2021.
Table 1 shows the univariate and bivariate analysis of differences in sociodemographic and clinical characteristics of patients with TB with and without HIV co-infection. Statistically significant differences between groups were found in all sociodemographic variables, except for income level. PLHIV and TB co-infection were more frequently men, aged between 25 and 64 years old, had 10 times more frequently prison background, doubled tobacco and alcohol consumption and were almost 80 times more frequently injected drug users. As for clinical variables, we found statistically significant differences in all of them. PLHIV doubled the percentage of unsuccessful TB treatment, had more frequently taken TB treatment in the past and higher rates of directly observed treatment (DOT).
Sociodemographic and clinical differences between TB patients with and without HIV co-infection in Barcelona, Spain (2000–2021). Unweighted frequencies and percentages.
| n (%) | No HIV co-infection(n=7644) | PLHIV(n=762) | p-Value* | Crude OR of having HIV | aOR of having HIV† |
|---|---|---|---|---|---|
| Sex | <0.001 | ||||
| Women | 2945 (38.5) | 194 (21.5) | Ref | Ref | |
| Men | 4699 (61.5) | 598 (78.5) | 2.29 (1.92–2.74) | 1.75 (1.37–2.24) | |
| Age | <0.001 | ||||
| <24 | 1316 (17.2) | 17 (2.2) | Ref | Ref | |
| 25–64 | 4790 (62.7) | 720 (94.5) | 11.64 (7.41–19.66) | 8.22 (4.73–15.62) | |
| 65 and over | 1536 (20.1) | 25 (3.3) | 1.26 (0.68–2.38) | 1.06 (0.52–2.28) | |
| Country of birth | <0.001 | ||||
| Foreign country | 3574 (46.8) | 287 (37.7) | Ref | Ref | |
| Spain | 4070 (53.2) | 475 (62.3) | 1.45 (1.25–1.70) | 1.47 (1.16–1.87) | |
| Income level | 0.499 | ||||
| Low | 3103 (41.6) | 275 (39.3) | Ref | Ref | |
| Medium-low | 2083 (27.9) | 196 (28.0) | 1.06 (0.88–1.28) | 1.24 (0.96–1.61) | |
| Medium-high | 1613 (21.6) | 158 (22.6) | 1.11 (0.90–1.35) | 1.86 (1.42–2.43) | |
| High | 657 (8.8) | 71 (10.1) | 1.22 (0.92–1.60) | 2.53 (1.76–3.58) | |
| Has been/is in prison | <0.001 | ||||
| No | 7540 (98.6) | 669 (87.7) | Ref | Ref | |
| Yes | 104 (1.4) | 93 (12.3) | 10.00 (7.47–13.37) | 1.13 (0.64–1.99) | |
| Smoker | <0.001 | ||||
| No | 5065 (65.3) | 288 (38.8) | Ref | Ref | |
| Yes | 2579 (33.7) | 474 (62.2) | 3.23 (2.77–3.77) | 1.08 (0.84–1.38) | |
| Problematic alcohol use | <0.001 | ||||
| No | 6460 (84.5) | 493 (64.7) | Ref | Ref | |
| Yes | 1184 (15.5) | 269 (35.3) | 2.98 (2.53–3.49) | 1.04 (0.79–1.37) | |
| Persons who inject drugs | <0.001 | ||||
| No | 7548 (98.7) | 384 (50.4) | Ref | Ref | |
| Yes | 96 (1.3) | 378 (49.6) | 77.40 (60.75–99.48) | 45.81 (33.10–64.26) | |
| Diabetes mellitus | <0.001 | ||||
| No | 7099 (92.9) | 748 (98.2) | Ref | Ref | |
| Yes | 545 (7.1) | 14 (1.8) | 0.24 (0.14–0.40) | 0.17 (0.08–0.34) | |
| Pervious TB treatment | <0.001 | ||||
| No | 7139 (93.4) | 630 (82.7) | Ref | Ref | |
| Yes | 505 (6.6) | 132 (17.3) | 2.96 (2.40–3.64) | 1.77 (1.30–2.40) | |
| Sputum smear and culture | 0.029 | ||||
| Smear positive (+), cell culture (+) | 2588 (33.9) | 289 (37.9) | Ref | Ref | |
| Cell culture or PCR (+) | 3154 (41.3) | 311 (40.8) | 0.88 (0.75–1.05) | 0.92 (0.72–1.19) | |
| Negative or not done | 1902 (24.9) | 162 (21.3) | 0.76 (0.62–0.93) | 1.06 (0.55–2.13) | |
| TB location | 0.014 | ||||
| Extrapulmonary TB | 2212 (28.9) | 185 (24.3) | Ref | Ref | |
| Pulmonary | 5432 (71.1) | 577 (75.7) | 1.27 (1.07–1.51) | 1.26 (0.93–1.72) | |
| Diagnostic delay in days (median (IQR)) | 47 (21–99) | 35 (18–68) | <0.001 | ||
| Chest X-ray | <0.001 | ||||
| Cavitating | 1642 (21.5) | 93 (12.2) | Ref | Ref | |
| Abnormal, not cavitating | 4458 (58.3) | 503 (66.0) | 1.99 (1.59–2.52) | 3.07 (2.24–4.27) | |
| Not done | 135 (1.8) | 12 (1.6) | 1.57 (0.80–2.83) | 2.47 (1.03–5.40) | |
| Normal | 1409 (18.4) | 154 (20.2) | 1.93 (1.48–2.53) | 3.59 (2.31–5.63) | |
| Drug resistance | <0.001 | ||||
| Negative culture | 2146 (28.1) | 188 (24.7) | Ref | Ref | |
| No antibiogram done | 601 (7.9) | 84 (11.0) | 1.60 (1.21–2.09) | 1.47 (0.75–3.03) | |
| Drug-susceptible | 4257 (55.7) | 409 (53.7) | 1.10 (0.92–1.32) | 1.18 (0.65–2.31) | |
| Non-MDR resistance | 552 (7.2) | 62 (8.1) | 1.28 (0.94–1.72) | 1.69 (0.85–3.51) | |
| MDR | 88 (1.2) | 19 (2.5) | 2.46 (1.43–4.05) | 1.76 (0.66–4.62) | |
| Contact tracing | <0.001 | ||||
| No | 2313 (30.3) | 388 (50.9) | Ref | Ref | |
| Yes | 5331 (69.7) | 374 (49.1) | 0.42 (0.36–0.49) | 0.51 (0.40–0.64) | |
| Homeless | <0.001 | ||||
| No | 7154 (93.6) | 607 (79.7) | Ref | Ref | |
| Yes | 490 (6.4) | 155 (20.3) | 3.73 (3.05–4.54) | 0.93 (0.64–1.33) | |
| Unsuccessful TB treatment | <0.001 | ||||
| No | 6716 (87.9) | 581 (76.2) | Ref | – | |
| Yes | 928 (12.1) | 181 (23.8) | 2.25 (1.88–2.70) | – | |
| Final conclusion | <0.001 | ||||
| Cured | 6716 (87.9) | 581 (76.2) | Ref | Ref | |
| Dead | 503 (6.6) | 98 (12.9) | 2.25 (1.78–2.83) | 2.67 (1.81–3.89) | |
| Lost | 210 (2.7) | 44 (5.8) | 2.42 (1.71–3.36) | 1.79 (1.11–2.82) | |
| Emigrated/others | 215 (2.8) | 39 (5.1) | 2.10 (1.46–2.95) | 1.47 (0.84–2.46) | |
| Directly observed treatment (DOT) | <0.001 | ||||
| No | 6127 (80.2) | 397 (52.1) | Ref | Ref | |
| Yes | 1517 (19.8) | 365 (47.9) | 3.71 (3.19–4.33) | 2.20 (1.71–2.81) | |
In the multivariate model adjusted by all the considered variables, the OR of unsuccessful TB treatment in PLHIV was 2.25 times (95% CI: 1.88–2.70) the odds of that of a person without HIV infection. According to our adjusted model, PLHIV had higher odds of being using injected drugs (aOR=45.81; 95% CI (33.10–64.26)), had previously been treated for TB (aOR=1.77; 95% CI (1.30–2.40)) and lower odds of having contact tracing done (aOR=0.51; 95% CI (0.40–0.64)) and having diabetes (aOR=0.17; 95% CI (0.08–0.34)).
Table 2 shows the optimized multivariate model for the subsample of PLHIV. Among them, those who used injected drugs had higher rates of unsuccessful treatment, OR=1.56 (95% CI: 1.11–2.19), as well as those who were homeless OR=1.61 (95% CI: 1.07–2.42). On the other hand, cases for which contact tracing was conducted, had lower rates of unsuccessful treatment (OR: 0.53; 95% CI: 0.37–0.75).
Differences in treatment success among PLHVI according to base characteristics in Barcelona, Spain (2000–2021).
| PLVIH successful treatment(n (%)) | PLVIH unsuccessful treatment(n (%)) | p-Value* | Crude OR unsuccessful treatment PLHIV | aOR¥ unsuccessful treatment PLHIV | |
|---|---|---|---|---|---|
| n | 581 | 181 | |||
| Sex | 0.611 | ||||
| Women | 128 (22.0) | 36 (19.9) | Ref | Ref | |
| Men | 453 (78.0) | 145 (80.1) | 1.14 (0.76–1.74) | 0.99 (0.65–1.53) | |
| Age | 0.010 | ||||
| <24 | 9 (1.5) | 8 (4.4) | Ref | Ref | |
| 25–64 | 557 (95.9) | 163 (90.1) | 0.33 (0.12–0.89) | 0.37 (0.14–1.03) | |
| 65 and over | 15 (2.6) | 10 (5.5) | 0.75 (0.21–2.62) | 1.17 (0.32–4.28) | |
| Country of birth | 0.683 | ||||
| Spain | 365 (62.9) | 110 (60.8) | Ref | – | |
| Foreigner | 216 (37.1) | 71 (39.2) | 0.92 (0.65–1.29) | – | |
| Income level | 0.841 | ||||
| Low | 212 (39.2) | 63 (39.9) | Ref | – | |
| Medium-low | 151 (27.9) | 45 (28.5) | 1.00 (0.65–1.55) | – | |
| Medium-high | 121 (22.3) | 37 (23.4) | 1.03 (0.64–1.63) | – | |
| High | 58 (10.7) | 13 (8.2) | 0.75 (0.37–1.43) | – | |
| Has been/is in prison | 0.641 | ||||
| No | 501 (87.3) | 161 (89.0) | Ref | – | |
| Yes | 73 (12.7) | 20 (11.0) | 0.85 (0.49–1.42) | – | |
| Smoker | 1.000 | ||||
| No | 220 (37.9) | 68 (37.6) | Ref | – | |
| Yes | 361 (62.1) | 113 (62.4) | 1.01 (0.72–1.43) | – | |
| Problematic alcohol use | 0.521 | ||||
| No | 380 (65.4) | 113 (62.4) | Ref | – | |
| Yes | 201 (34.6) | 68 (37.6) | 1.14 (0.80–1.60) | – | |
| Person who inject drugs | 0.012 | ||||
| No | 308 (53.0) | 76 (42.0) | Ref | Ref | |
| Yes | 273 (47.0) | 105 (58.0) | 1.56 (1.11–2.19) | 1.54 (1.07–2.21) | |
| DM | 1.000 | ||||
| No | 570 (98.1) | 178 (98.3) | Ref | – | |
| Yes | 11 (1.9) | 3 (1.7) | 0.87 (0.20–2.83) | – | |
| Previous TB treatment | 0.848 | ||||
| No | 479 (82.4) | 102 (83.4) | Ref | – | |
| Yes | 102 (17.6) | 30 (16.6) | 0.93 (0.59–1.44) | – | |
| TB location | 0.094 | ||||
| Extrapulmonary | 150 (25.8) | 35 (19.3) | Ref | – | |
| Pulmonary | 431 (74.2) | 146 (80.7) | 1.45 (0.97–2.22) | – | |
| Sputum smear and culture | 0.931 | ||||
| Smear+ & culture + | 220 (37.9) | 69 (38.1) | Ref | – | |
| Culture or pcr+ | 239 (41.1) | 72 (39.8) | 0.96 (0.66–1.40) | – | |
| Negative or not conducted | 122 (21.0) | 40 (22.1) | 1.05 (0.66–1.63) | – | |
| Diagnostic delay in days (median (IQR))*Wilcoxon test | 36.0 (19.5–67.0) | 33.5 (16.0–69.7) | 0.4004 | – | |
| Chest X-ray | 0.654 | ||||
| Cavitated | 72 (12.4) | 21 (11.6) | Ref | – | |
| Abnormal no cavitated | 377 (64.9) | 126 (69.6) | 1.15 (0.69–1.98) | – | |
| Not performed | 10 (1.7) | 2 (1.1) | 0.69 (0.10–2.87) | – | |
| Normal | 122 (21.0) | 32 (17.7) | 0.90 (0.48–1.69) | – | |
| Drug resistance | 0.850 | ||||
| Negative culture | 142 (24.4) | 46 (25.4) | Ref | – | |
| Not performed | 62 (10.7) | 22 (12.2) | 1.10 (0.60–1.96) | – | |
| Multisensible | 318 (54.7) | 91 (50.3) | 0.88 (0.59–1.33) | – | |
| Resistance not MDR | 45 (7.7) | 17 (9.4) | 1.17 (0.60–2.21) | – | |
| MDR | 14 (2.4) | 5 (2.8) | 1.10 (0.34–3.06) | – | |
| Contact tracing | <0.001 | ||||
| No | 272 (46.8) | 116 (64.1) | Ref | Ref | |
| Yes | 309 (53.2) | 65 (35.9) | 0.49 (0.35–0.69) | 0.53 (0.37–0.75) | |
| Homeless | 0.001 | ||||
| No | 479 (82.4) | 102 (70.7) | Ref | Ref | |
| Yes | 102 (17.6) | 53 (29.3) | 1.94 (1.32–2.85) | 1.61 (1.07–2.42) | |
| Final conclusion | <0.001 | ||||
| Successful treatment | 581 (100.0) | 0 (0.0) | – | – | |
| Death | 0 (0.0) | 98 (54.1) | – | – | |
| Lost | 0 (0.0) | 44 (24.3) | – | – | |
| Emigrated/others | 0 (0.0) | 39 (21.5) | – | – | |
| Directly observed treatment (DOT) | 1.000 | ||||
| No | 303 (52.2) | 94 (51.9) | Ref | – | |
| Yes | 278 (47.8) | 87 (48.1) | 1.01 (0.72–1.41) | – | |
p-Values for differences between these two groups were tested using χ2 and Wilcoxon test for median of days of diagnostic delay.
Adjusted odds ratios (aOR) of unsuccessful treatment were obtained with logistic regressions. The level of statistical significance was set to α=0.05. Only variables that showed statistical significance in the bivariate models and were of epidemiological interest for the adjustment were considered for the multivariate analyses (age, sex, person who injects drugs, contact tracing and homelessness).
Overall, the characteristics of PLHIV and TB co-infection population in Barcelona are very similar to the ones reported by WHO.1 Men are usually more affected by TB mainly because of a higher exposure to settings that favor transmission as well as their engagement in hazardous careers.14
In our population, HIV prevalence in newly diagnosed TB cases decreased almost 10 percentage points during the study period. This is consistent with available evidence, which has shown a global decrease on HIV and TB co-infection, mainly due to changes in the clinical management and treatment of PLHIV such as immediate ART treatment.15 Global strategies addressing HIV and TB simultaneously have also played a role in co-infection reduction.16
As opposed to the trend registered for the European Region, where HIV-positive TB diagnoses have increased since the year 2000,2 HIV-positive TB incident cases in Barcelona in 2021 followed a decreasing trend and were slightly lower to the incidence found globally (6.7%).1
Out of all TB cases diagnosed, we identified several factors related to a higher association with having HIV co-infection. PLHIV were more often men, between the age of 25–64 and Spanish-born. Similar results regarding gender where recently found in Brazil,17 but they found no significant differences among age groups, which may indicate different epidemiological patterns of TB and HIV between both populations. PLHIV were also more frequently injected drug users. Moreover, injected drug users have previously been identified as a risk group for HIV and TB co-infection in other European settings such as Bucharest, Romania, where the epidemiological situation is similar than observed in Barcelona in the nineties.18 On the other hand, PLHIV were less likely to be diabetic. Although HIV is known to increase the risk of developing diabetes,9 it is mostly due to the iatrogenic effects of HAART therapy.19 PLHIV in our population may be underdiagnosed in comparison with the rest of tuberculosis patients. We hypothesized that PLHIV in our population do not access the healthcare system due to their social conditions, hindering the diagnosis of chronic metabolic diseases like diabetes. Diabetes could also be considered as a side effect of the treatment, hindering its correct characterization.
When comparing TB management, PLHIV had more often been previously treated for TB, and had more often treatment under DOT. Contact tracing, was significantly lower for these patients, as well as treatment success. All these characteristics show how HIV co-infection patients tend to have more complex clinical management and require special attention during treatment follow-up and epidemiologic surveillance, hence the higher likeness of DOT.
Treatment success rates among TB patients, both HIV+ and HIV− were found to be similar to those reported globally, and even slightly higher among HIV− individuals in Barcelona.6 Still, overall treatment success was under the 90% WHO goal for 2025.20
The higher TB unsuccessful treatment among PLHIV could be explained by the fact that the mortality rate is higher among HIV-positive patients,6,21–26 probably due to AIDS as TB in these patients tends to be more aggressive,27 and lost and emigrated cases almost double the rates of those without HIV, hindering treatment completion. As for TB treatment drug resistance, we did not find any statistically significant differences for treatment failure according to resistances in PLHIV, contrary to what previous studies have shown.28
In our population of PLHIV, cases with contact tracing performed were significantly less likely to have unsuccessful treatment. Our hypothesis is that the collaborative effort to provide contacts’ personal data could be extensive to a higher willingness to follow treatment instructions. Similar results to ours were found recently in Uganda.29 On the other hand, higher willingness and resources to perform contact tracing by healthcare professionals could indirectly imply more exhaustive treatment follow-up. Contrarily, homelessness and use of injected drugs were directly associated with unsuccessful treatment among PLHIV, highlighting how socioeconomic and housing instability can hinder treatment success. A recent review linked lower treatment adherence to barriers for access to healthcare, including stigma and dehumanization of homeless people.30 Directly observed treatment programs and collaboration with care centers for substance abuse could be helpful in reducing these barriers and increase adherence.
Our study had some limitations. We could not obtain data on HIV transmission route. We are aware that the transmission route and other HIV-related variables, like CD4 count at diagnosis and initiation of HAART therapy, could shed new light towards factors related to TB treatment success or not. The observational transversal study design did not allow exploring causality or changes over time. Nevertheless, our exhaustive data records and sample size allowed us to assess factors associated to HIV and treatment success among all TB cases in the city for over 20 years, giving our results high robustness that may be transferable to similar settings. Moreover, from an epidemiological surveillance and control framework, our research aims to identify easily collectable sociodemographic characteristics that can inform us on the probability of unsuccessful treatment, making the best use of the available resources (DOT, inpatient treatment).
Finally, it should be taken into consideration that the last 2 years of our analyses (2020–2021) were affected by the COVID-19 pandemic. Although this had a great impact in late notification of cases, by the end of 2020, TB cases reported matched those expected according to historic tendencies in the city, and although it is possible that treatment success rates changed during this period, cases from these years only represent 5.7% of our population, so it could hardly affect our general results.
ConclusionsDespite the decrease of proportion of TB/HIV co-infection in our setting, TB treatment success is still lower for PLHIV. This could be related to higher mortality and loss of follow-up among these patients. PLHIV were more likely to successfully complete TB treatment when they had contact tracing. On the other hand, injected drugs use and being homeless were associated to lower treatment success among this population. Further efforts and resources are needed to improve treatment success in PLHIV with TB, including an interdisciplinary approach beyond healthcare systems and intensive follow-up.
CRediT authorship contribution statementAll authors made significant contributions to the study and reviewed and approved the final version of the paper. No other person made a substantial contribution to the paper. JPM, GB, IM and AO contributed to study concept and design. GB and IM composed the statistical dataset and performed the analyses. GB, IM, JPM, AO and CP contributed to the interpretation of the data and results.
FundingThis research did not receive any specific grant from funding agencies in the public commercial, or not-for-profit sectors.
Conflict of interestThe authors declare no conflicts of interest.
