Abstract
Introduction
Active case finding (ACF) can improve tuberculosis (TB) diagnosis in high-burden populations. Many programmes rely on symptom evaluation as the first screening step; however, symptom reporting may vary according to the context of evaluation.
Methods
We performed a cross-sectional analysis of persons deprived of liberty (PDL) diagnosed with TB in an ACF programme conducted in two male Brazilian prisons from February 2023 to February 2025. Every 4 months, participants underwent sputum testing by GeneXpert Ultra (Xpert) and culture. Study nurses administered symptom questionnaires during the initial evaluation. Individuals with positive results were interviewed by a physician. We compared characteristics of individuals with microbiologically confirmed TB according to their symptom reporting at initial screening and medical consultation, which was used to define four groups (e.g., combinations of symptomatic/asymptomatic at each of the two symptom assessments).
Results
Among 6919 PDL screened, 448 (6.4%) had microbiologically confirmed TB. We included 292 (65.1%) individuals in the analysis, after excluding those with missing clinical information and consults >15 days after screening. 113 (38.7%) reported symptoms at initial screening, while 192 (65.8%) reported symptoms at medical consultation (p<0.001). Compared with individuals asymptomatic at both evaluations, those who reported symptoms at the two instances were more likely to have medium or high semiquantitative Xpert results (45.0% vs 11.4%, p<0.001), report symptomatic cellmates (61.0% vs 25.2%, p<0.001), have higher LunitTB scores (95.1 vs 91.9; p=0.037) and lower body mass index (19.4 vs 21.4, p<0.001). There was a trend of more cavitary lesions (60.9% vs 50.0%, p=0.073) and bilateral disease (56.7% vs 45.0%, p=0.056) in chest X-ray of participants with symptomatic than asymptomatic TB at medical consultation.
Conclusions
More than half of individuals who reported no TB symptoms at initial screening endorsed symptoms at first medical consultation. Clinical severity and contextual factors can influence TB symptom reporting during ACF.
Keywords: Tuberculosis, Mass Screening, Population Surveillance
WHAT IS ALREADY KNOWN ON THIS TOPIC
Symptom screening is commonly used for tuberculosis (TB) active case finding and often a high proportion of asymptomatic individuals are diagnosed with active TB disease. Few studies have shown that symptom status can be inconsistently reported and how it can be related to individual characteristics.
WHAT THIS STUDY ADDS
Our study demonstrates that symptom reporting at active case finding differed substantially from the symptom status reported at medical consultation. Individuals with inconsistent symptom reporting had disease severity characteristics that fell between those who reported or denied symptoms on both occasions.
The multivariate analysis evidence that besides clinical status, perceiving a cellmate with TB symptoms was associated with symptom reporting.
HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY
Our findings suggest that symptom reporting is inconsistent and factors other than disease severity can contribute to individual responses. These results contribute to the evidence of unreliability of symptom screening for active case finding and problems that may arise in future studies that consider symptomatic disease as endpoints.
Introduction
Despite recent advances in prevention and diagnosis strategies, tuberculosis (TB) remains one of the biggest global health problems, affecting ~10.8 million people worldwide in 2024.1 With growing focus on early diagnosis, asymptomatic TB is increasingly recognised as an important driver of the global TB epidemic.2 3 To date, many active case finding (ACF) programmes rely on symptom screening.4 5 However, due to the inconsistencies in how symptoms are defined and ascertained, there is a considerable variability in asymptomatic TB burden reported in studies and prevalence surveys.6,8 Furthermore, symptom reporting, like other self-reported measures, is prone to response biases including those induced by differential recall, social pressures, mistrust of healthcare providers, or disinterest.9 10 Assessing the validity of and understanding sources of variation of symptom screening ascertainment methods is critical to accurate estimates of asymptomatic TB burden in different settings.
Additionally, whether the clinical manifestations and lung abnormalities differ between individuals with symptomatic and asymptomatic TB remains unclear. Currently, ACF strategies using X-ray are recommended by the WHO11 and can provide important data about how radiographic manifestations of TB relate to symptoms. Understanding the extent of disease manifestation among individuals with asymptomatic TB will be important for designing effective screening strategies and understanding how early detection can reduce morbidity.
WHO screening guidelines identify people deprived of liberty (PDL) as a key population for systematic screening. Carceral settings serve as a reservoir for TB disease and transmission, which may contribute to the increase in TB incidence among the general population.11,13 In Latin America, the TB incidence rate among PDL was estimated to be nearly 27-fold higher than that of the general population. Emerging findings suggest that, even in high-prevalence settings such as prisons, asymptomatic TB still contributes substantially to Mycobacterium tuberculosis transmission.314,18
To assess variations in symptom reporting and clinical characteristics associated with symptomatic status, we evaluated symptom reporting by participants with microbiologically confirmed TB screened during an ACF study in high TB-burden prisons, comparing responses obtained during their initial TB screening and subsequent clinic-based assessment. Further, we compared demographic and clinical characteristics of individuals diagnosed with TB according to their symptom reporting in these two evaluations.
Methods
Study setting
We conducted an analysis of individuals with microbiologically confirmed TB diagnosed during a prospective cohort study of ACF conducted in two male prisons located in the state of Mato Grosso do Sul, Brazil, from February 2023 to February 2025. The prisons had an occupancy of approximately 4000 individuals incarcerated at any given time during the period of data collection.
Eligible participants included adult PDL (≥18 years old) who provided consent. The study included baseline TB screening and follow-up screening every 4 months for up to 36 for those who had (a) two negative GeneXpert Ultra (Xpert) test results (ie, pool and individual Xpert tests) at baseline screening and (b) at least 12 months remaining sentence time. For our primary analyses, we only included PDL with microbiologically confirmed TB disease who had a medical consultation within 15 days of their initial screening. We excluded individuals with missing chest radiographs or incomplete clinical data and those who were diagnosed outside of the systematic screening procedures. Prison staff were not involved in data collection and were not informed about individual decisions to participate in the study and there were no reprimands or penalties for individuals who did not want to participate.
Screening procedures
Our study was performed in the context of systematic screening offered to all PDL in the study prisons. We approached individuals through trained study team members announcing study procedures at the entrance of cell blocks during yard time to offer participation and determine eligibility. We did not collect information on individuals who refused to participate; however, in the past, participation rates were above 95% of the total population at any given time.
Study nurses obtained consent discussion for eligible individuals, followed by the administration of standardised study questionnaires to collect data on sociodemographic characteristics, medical and incarceration history, use of illicit drugs (marijuana, crack, cocaine, base paste, glue/solvents and intravenous drugs) and tobacco in the last year, and exposure to cellmates presenting with TB symptoms. Subsequently, PDL were asked about their TB symptoms (cough, fever, weight loss, night sweats, shortness of breath and chest pain), regardless of duration. Each symptom was individually queried in sequence (i.e., ‘Do you have cough?’, ‘Do you have fever?).
Following the interview, individuals who provided consent, regardless of their symptoms, were requested to provide a spot sputum sample for GeneXpert MTB/Rif Ultra (Xpert) and culture test using solid media (i.e., Lowestein-Jensen and Ogawa-Kudoh media for those with a positive Xpert result, Ogawa-Kudoh media for those with a negative Xpert result). Two sputum samples were collected at the initial screening: the first evaluated through pooling eight samples and testing by Xpert19 and the second tested individually by Xpert. After sputum sample collection, a posterior–anterior chest X-ray (CXR) image was obtained by using either a conventional Sawae unit (Prison A) or the FDR Xair portable X-ray system (Prison B). All CXR images were scored using LunitTB (V.3.1.5.1), a deep learning algorithm developed by Lunit (Seoul, South Korea). In our analyses, we included the numerical scores of each pulmonary abnormality (i.e., TB, consolidation, nodule, fibrosis, atelectasis and calcification) and a composite score called ‘abnormality score’ with the highest result among those reported. The scores provided by Lunit vary between 0 and 100, where higher scores indicate a higher likelihood of TB (for the LunitTB score) or the abnormality described.
Definitions and standard of care
We defined microbiologically confirmed TB disease by a positive Xpert or culture result. Individuals with Xpert trace results and negative culture test results were considered as having invalid Xpert results and, therefore, excluded from primary analyses. Individuals with positive Xpert or culture results were invited to visit the prison health clinic for a medical evaluation; study participants were not informed in advance about their test results prior to consult but the responsible physicians were aware of the results. Two infectious diseases physicians were responsible for patient care during the study period, performing the initial medical consultation and informing patients about the test results. The consult was performed according to Brazilian guidelines of TB care and symptom data was collected through individual report during consult, all symptoms reported were evaluated in respect to those inquired in the screening questionnaire, regardless of duration. Physicians also reviewed CXR results for abnormalities consistent with active TB, including the presence of cavitary lesions of any size, noting whether abnormal findings involved one (unilateral) or both lungs (bilateral). Height and weight (for body mass index (BMI) calculation) were also collected during the first medical consultation. Treatment was initiated at this first medical consult to individuals with microbiologically confirmed TB.
Study measures and definition
For our main analysis, we referred to symptom status responses collected during the medical consultation, where ‘asymptomatic TB’ was defined as TB disease in PDL who did not report any TB symptoms, irrespective of duration; otherwise PDL were categorised as ‘symptomatic TB’. To determine symptom concordance, we compared responses to TB symptom questionnaires administered during the initial screening (i.e., the same day as sputum sample collection for most participants) versus the first medical consultation. We defined concordance when PDL gave consistent answers with respect to reporting of any TB symptoms during the initial screening and the first medical consultation. Thus, discordance was defined when PDL gave different answers for the initial TB screening and the first medical consultation (i.e., reported at least one TB symptom during the initial screening but none during the first medical consultation or vice versa).
Data management and statistical analysis
Data collected during the initial TB screening and the first medical consultations were stored in an online RedCap (Vanderbilt University, Tennessee, USA) database hosted by Fiocruz, Mato Grosso do Sul, Brazil.20 We used χ² (categorical predictors) and Wilcoxon rank sum (continuous predictors) tests to determine the crude association between sociodemographic and clinical predictors with symptomatic TB and discordance of symptom reporting.
We aimed to evaluate factors related to symptomatic disease in a predictive model that was performed under the assumption that symptom reporting at medical consult would provide a better clinical description of the symptoms. Thus, we used multivariable logistic regression models to determine factors predictive of symptomatic TB, referenced as responses given in medical consultation. In the model adjustment, we included the variables that had statistically significant association to the symptom outcome in the bivariate analysis and possible confounders reported in previous studies. Xpert semiquantitative results were grouped into High/Medium and Low/Very Low/Trace for the group comparison. For the primary comparison of clinical characteristics among individuals with TB who reported symptoms at the medical visit versus those who did not, the minimum detectable OR detectable with 80% power would range from 2.00 for a covariate present in 50% of the reference group to 2.45 for a covariate present in 10% of the reference group.
In addition to evaluating any reported TB symptom, we performed secondary analyses in which we defined symptomatic TB as reporting any of the symptoms included in the WHO four symptom screening (W4SS) tool.11 All statistical analyses were performed using R (V.4.3.2).21
Patient and public involvement
There was no patient or public involvement in this study.
Results
Characteristics of study participants
During the 2-year study period, 6919 PDL provided consent to participate and were included in the study. A total of 448 (6.4%) individuals with TB disease were identified, and TB treatment was initiated. Of these, 156 were excluded from our primary analyses (23 individuals had missing CXR data or clinical data, 41 individuals were diagnosed outside of our ACF procedures, and 92 individuals had >15 days gap between their TB screening and the first medical consultation dates) (figure 1). Of the 292 individuals (65.1%) with microbiologically confirmed TB included in this analysis, the median age was 31 years (IQR 27–36), and the majority were mixed-race (80.1%) and completed 4–12 years of schooling (84.6%) (online supplemental table 1). Previous incarceration was common (87.0%). The prevalence of current tobacco smoking was 75.0%, and 79.5% of study participants reported use of illicit drugs in the past year. 24.0% of individuals reported previous TB and the prevalence of HIV co-infection was 2.4%. Nearly half (43.2%) of study participants reported having one or more cellmates with any TB symptoms.
Figure 1. Study participant selection according to the eligibility criteria and symptom status at medical consultation. TB, tuberculosis.
Overall, the median time between TB screening and the first medical consultation was 8 days (IQR 6–9), and the majority (65%) of individuals had ≤10 day interval. Most individuals had a low, very low, or trace Xpert result (72.9%). The median BMI at diagnosis was 20.5 kg/m2 (IQR 18.6–22.8), and 21.9% of our study participants were underweight (BMI<18.5 kg/m2) at the first medical consultation (online supplemental table 2).
Burden of symptomatic TB and discordance in symptom reporting
Overall, only 38.7% (113/292) of individuals with microbiologically confirmed TB disease in our study reported TB symptoms during the initial screening. However, at the first medical consultation, this number significantly increased to 65.8% (192/292, p<0.001) (table 1). Among those who reported at least one TB symptom during the initial screening, the majority (67/113; 59.3%) reported two or more TB symptoms. The top three most reported symptoms during the initial screening included cough (34.9%), chest pain (16.0%), and difficulty breathing (14.0%). Among those who reported at least one TB symptom during the first medical consultation, more than a third reported two TB symptoms (71/192, 36.9%) and the top three most frequently reported symptoms included cough (52.3%), weight loss (27.3%) and chest pain (20.8%). We observed a similar trend when we defined symptomatic TB using the W4SS definition (table 1).
Table 1. Prevalence of symptomatic and asymptomatic TB among persons deprived of liberty diagnosed with TB disease during active TB case finding efforts in Brazilian prisons (n=292).
| Variables | Screening N (%) |
Medical consultation N (%) |
|---|---|---|
| Any TB symptoms | ||
| Asymptomatic | 179 (61.3) | 100 (34.2) |
| Symptomatic | 113 (38.7) | 192 (65.8) |
| Cough | 102 (34.9) | 153 (52.3) |
| Weight loss | 32 (10.9) | 80 (27.3) |
| Chest pain | 47 (16.0) | 61 (20.8) |
| Difficulty breathing | 41 (14.0) | 46 (15.7) |
| Fever | 27 (9.2) | 33 (11.3) |
| Night sweat | 24 (8.2) | 22 (7.5) |
| Number of symptoms reported | ||
| 1 | 46 (15.7) | 61 (20.8) |
| 2 | 24 (8.2) | 71 (24.3) |
| ≥3 | 43 (14.7) | 60 (20.5) |
| W4SS | ||
| Asymptomatic | 184 (63.1) | 116 (39.7) |
| Symptomatic | 108 (36.9) | 176 (60.3) |
| Cough | 102 (34.9) | 153 (52.3) |
| Weight loss | 32 (10.9) | 80 (27.3) |
| Chest pain | 47 (16.0) | 49 (16.7) |
| Fever | 27 (9.2) | 33 (11.3) |
TB, tuberculosis; W4SS, WHO 4 symptom screening tool.
Over one-third (36.0%, 105/292) of study participants gave discordant responses to TB symptom questionnaires administered during the initial screening and the first medical consultation. Among 179 participants who were asymptomatic at initial screening, more than half (92/179) reported symptoms at the medical consultation. Among 113 participants who reported symptoms at initial screening, 11.5% (13/113) were asymptomatic when queried at the medical consultation (table 2). TB symptoms that were more often reported discordantly included cough, weight loss and chest pain (online supplemental table 2). We also assessed if there was a difference in symptom reporting disagreement due to the time between the two evaluations. There was no statistically significant difference in discordance among individuals with up to 7 days between the two evaluations and individuals with evaluations performed more than 7 days apart (41.5% vs 31.5%, χ² test, p=0.075).
Table 2. Concordance of symptoms screening during initial screening versus medical consultations to define asymptomatic TB among persons deprived of liberty in Brazilian prisons (n=292).
| Characteristics | Initial screening | P value | |||
|---|---|---|---|---|---|
| Symptomatic | Asymptomatic | ||||
| Medical consultation | Medical consultation | ||||
| Symptomatic N (%)=100 (34.2%) N (%) |
Asymptomatic N (%)=13 (4.5%) N (%) |
Symptomatic N (%)=92 (31.5%) N (%) |
Asymptomatic N (%)=87 (29.8%) N (%) |
||
| Demographic and clinical characteristics | |||||
| Median age (IQR) | 33 (27–37) | 31 (29–43) | 31 (27–36) | 30 (27–36) | 0.711 |
| Age group | |||||
| 18–44 | 92 (92.0) | 11 (84.6) | 88 (95.6) | 83 (95.4) | 0.283 |
| 45 + | 8 (8.0) | 2 (15.3) | 4 (4.4) | 4 (4.6) | |
| Race | |||||
| Black | 10 (10.0) | 1 (7.7) | 6 (6.5) | 9 (10.3) | 0.718 |
| White | 9 (9.0) | 9 (69.3) | 11 (12.0) | 9 (10.3) | |
| Mix | 81 (81.0) | 3 (23.0) | 75 (81.5) | 69 (79.3) | |
| Education level | |||||
| 4 years of schooling or less | 11 (11.0) | 2 (15.4) | 5 (5.4) | 7 (8.0) | 0.589 |
| 4–12 years of schooling | 82 (82.0) | 11 (84.6) | 82 (89.2) | 72 (82.7) | |
| >12 years of schooling | 7 (7.0) | 0 (0.0) | 5 (5.4) | 8 (9.3) | |
| Previously incarcerated | 90 (90.0) | 12 (92.3) | 76 (82.6) | 76 (87.3) | 0.490 |
| Cellmate(s) with TB symptoms | 61 (61.0) | 5 (38.4) | 38 (41.3) | 22 (25.2) | <0.001 |
| Smoking status | |||||
| Never smoker | 4 (4.0) | 3 (23.0) | 8 (8.6) | 14 (16.1) | 0.075 |
| Former smoker | 16 (16.0) | 1 (7.7) | 13 (14.1) | 14 (16.1) | |
| Current smoker | 80 (80.0) | 9 (69.3) | 71 (77.1) | 59 (67.8) | |
| Use of any illicit drugs in the past year | 81 (81.0) | 11 (84.6) | 72 (78.2) | 68 (78.1) | 0.934 |
| Median BMI (kg/m2) (IQR) | 19.4 (17.7–21.8) | 20.7 (19.7–23.1) | 20.7 (18.6–23.2) | 21.4 (19.8–23.6) | <0.001 |
| BMI classification | |||||
| Not underweight (BMI ≥18.5 kg/m2) | 67 (67.0) | 12 (92.3) | 72 (78.2) | 77 (88.7) | 0.003 |
| Underweight (BMI <18.5 kg/m2) | 33 (33.0) | 1 (7.7) | 20 (21.7) | 10 (11.4) | |
| Previous TB | 26 (26.0) | 4 (30.7) | 22 (23.9) | 18 (20.6) | 0.744 |
| Positive HIV status | 3 (3.0) | 0 (0%) | 2 (2.1) | 2 (2.2) | 1.000 |
| Clinical TB manifestation | |||||
| Semiquantitative Xpert result | |||||
| High/medium | 45 (45.0) | 3 (23.0) | 21 (22.9) | 10 (11.5) | <0.001 |
| Low/very low/trace | 55 (55.0) | 10 (76.9) | 71 (77.1) | 77 (88.5) | |
| Positive culture | 75 (75.0) | 12 (92.3) | 60 (65.2) | 59 (67.8) | 0.139 |
| Any chest X-ray abnormalities | 92 (92.0) | 13 (100.0) | 82 (89.2) | 70 (80.4) | 0.062 |
| Cavitary disease | 61 (61.0) | 7 (53.8) | 56 (60.8) | 44 (50.5) | 0.347 |
| Bilateral disease | 58 (58.0) | 7 (53.8) | 51 (55.4) | 38 (43.6) | 0.236 |
| Median Lunit Scores (IQR) | |||||
| Lunit TB score | 95.1 (84.9–98.4) | 96.1 (94.8–99.0) | 96.2 (82.1–98.7) | 91.9 (55.9–97.6) | 0.037 |
| Lunit abnormality score | 87.5 (78.4–94.8) | 88.6 (85.7–90.8) | 84.2 (67.3–92.7) | 80.9 (54.3–90.2) | 0.009 |
| Lunit consolidation score | 76.1 (21.7–91.9) | 88.2 (19.3–90.8) | 72.3 (10.9–91.1) | 32.2 (3.8–79.3) | <0.001 |
| Lunit nodule score | 32.3 (20.5–50.2) | 41.2 (28.5–42.4) | 40.2 (23.9–51.7) | 33.4 (17.9–53.4) | 0.691 |
| Lunit fibrosis score | 46.1 (18.1–80.0) | 65.6 (14.7–80.9) | 50.3 (16.0–70.9) | 46.8 (8.5–76.7) | 0.681 |
| Lunit atelectasis score | 2.2 (1.6–3.4) | 2.3 (1.8–2.6) | 1.9 (1.4–3.5) | 2.0 (1.5–3.6) | 0.424 |
| Lunit calcification score | 8.6 (4.5–17.5) | 9.2 (4.4–17.0) | 8.7 (3.9–17.5) | 8.2 (4.2–25.1) | 0.926 |
Bold indicates that the finding is significant at α=0.05.
BMI, body mass index; TB, tuberculosis.
Individuals who reported no symptoms at initial screening but did report symptoms at medical consultation had distributions of several clinical and exposure characteristics that fell between those of individuals who were either asymptomatic or symptomatic at both evaluations. These included reporting having a cellmate with symptoms (41.3%, compared with 25.2% in those asymptomatic on both evaluations and 61% among those with symptoms on both evaluations; p<0.001 for both comparisons); underweight (21.7% vs 11.4% and 33%; p=0.027 and p=0.009); Xpert semiquantitative load medium/high (22.9% vs 11.4% and 45%; p=0.050 and p=0.001) and Lunit abnormality score (median 84.2 vs 80.9 and 87.5; p=0.032 and p=0.558) (table 2).
Clinical manifestations of symptomatic TB at medical consult
We considered symptom status at medical consult for the primary analysis. Individuals with symptomatic TB were similar to those with asymptomatic TB in regard to age distribution, race, education attainment level, previous incarceration, use of illicit drugs in the past year, previous TB history, HIV co-infection status and culture positivity (p>0.05). Compared with individuals with asymptomatic TB, individuals with symptomatic TB were more likely to report having ≥1 cellmates with any TB symptoms (51.5% vs 27.0%, p<0.001) or current smoking (78.6% vs 68.0%, p=0.013). The median BMI among individuals with symptomatic TB was lower compared with those with asymptomatic TB (19.9 vs 21.4, p=0.002) (online supplemental table 1).
Individuals with high/medium bacterial load on semi-quantitative Xpert were more likely to have symptomatic TB (34.3% vs 13.0%, p<0.001). The Lunit TB, abnormality and consolidation scores were all significantly higher in individuals with symptomatic TB compared with those with asymptomatic TB. We observed a trend of more advanced radiographic manifestations, such as cavitary lesions (60.9% vs 50.0%, p=0.073) and bilateral disease (56.7% vs 45.0%, p=0.056) in participants with symptomatic TB (online supplemental table 1).
In multivariable logistic regression analysis, the odds of symptomatic TB were higher among individuals who reported having ≥1 cellmates with any TB symptoms (adjusted OR (aOR) 2.60, 95% CI 1.51 to 4.57), current smokers (aOR 2.37, 95% CI 1.01 to 5.70), those with High/Medium Xpert semi-quantitative load (aOR 2.88, 95% CI 1.44 to 6.09) and those who had BMI <18.5 kg/m2 during the first medical consultation (aOR 2.88, 95% CI 1.44 to 6.09) (table 3).
Table 3. Crude and adjusted OR of symptomatic TB at medical visit among persons deprived of liberty in Brazilian prisons (n=292).
| Characteristics | Measure of association | |
|---|---|---|
| cOR (95% CI) | aOR (95% CI) | |
| Age group | ||
| 18–44 | Reference | Reference |
| 45+ | 1.04 (0.39 to 3.08) | 1.21 (0.40 to 3.97) |
| Ethnicity | ||
| Black | Reference | |
| Mixed | 1.25 (0.52 to 2.84) | |
| White | 1.04 (0.35 to 3.03) | |
| Education level | ||
| Less than 4 years | Reference | Reference |
| 4–12 years | 1.11 (0.45 to 2.57) | 0.96 (0.36 to 2.44) |
| >12 years or more | 0.84 (0.24 to 2.86) | 0.88 (0.23 to 3.31) |
| Previously incarcerated | ||
| No | Reference | Reference |
| Yes | 0.87 (0.40 to 1.77) | 0.89 (0.38 to 1.99) |
| Cellmate with TB symptoms | ||
| No | Reference | Reference |
| Yes | 2.87 (1.71 to 4.92) | 2.60 (1.51 to 4.57) |
| Tobacco use | ||
| Never smoked | Reference | Reference |
| Former smokers | 2.73 (1.05 to 7.36) | 2.20 (0.79 to 6.28) |
| Current smokers | 3.14 (1.43 to 7.10) | 2.37 (1.01 to 5.70) |
| Illicit drug use in the last year | ||
| No | Reference | |
| Yes | 1.04 (0.56 to 1.87) | |
| BMI | ||
| Not underweight | Reference | Reference |
| Underweight BMI (<18.5) | 3.08 (1.58 to 6.51) | 2.69 (1.31 to 5.91) |
| Previous TB | ||
| No | Reference | Reference |
| Yes | 1.18 (0.67 to 2.12) | 0.85 (0.44 to 1.64) |
| Rapid HIV test status | ||
| Negative | Reference | |
| Positive | 1.31 (0.27 to 9.26) | |
| Semi-quantitative Xpert result | ||
| Low/Very Low/Trace* | Reference | Reference |
| High/medium | 3.50 (1.87 to 7.00) | 2.88 (1.44 to 6.09) |
| Culture result | ||
| Negative | Reference | |
| Positive | 0.96 (0.56 to 1.63) | |
| Any CXR abnormalities | ||
| No | Reference | |
| Yes | 1.97 (0.96 to 4.05) | |
| Cavitary lesions | ||
| None | Reference | Reference |
| At least 1 | 1.56 (0.95 to 2.54) | 0.97 (0.93 to 1.71) |
| Lung involvement | ||
| Unilateral | Reference | |
| Bilateral disease | 1.60 (0.98 to 2.61) | |
Bold indicates that the finding is significant at α<0.05.
Included 131 individuals with low, 81 individuals with very low and 1 individual with trace Xpert result.
aOR, adjusted OR; BMI, body mass index; cOR, crude OR; CXR, chest X-ray; TB, tuberculosis.
Discussion
In this systematic screening study conducted in two high TB burden prisons, we found that half of individuals with microbiologically confirmed TB who reported no symptoms at initial screening endorsed symptoms at a first clinic visit with a physician. Compared with participants without symptoms, those who were symptomatic at medical consult had higher sputum bacillary load, lower BMI and more extensive radiographic abnormalities. Symptomatic patients were also substantially more likely to report other individuals in their cells with TB symptoms. Individuals who reported no symptoms initially but endorsed symptoms at the clinic visits generally had clinical characteristics that fell between those who were asymptomatic or symptomatic at both visits, suggesting that symptom reporting might be more variable among individuals with less advanced disease.
Overall, one in three individuals with TB disease detected through our ACF screening study did not report any TB symptoms during the first medical consultation despite all individuals included in the analysis being referred to treatment. This finding is consistent with several studies, including those conducted in prison settings, which reported a range of asymptomatic TB prevalence between 34.4% and 70.5%, despite differences in definitions of asymptomatic TB.16 17 22 23 Our study design likely allowed us to detect individuals with TB early on, when disease manifestations and pathology are mild.
Over one-third of our study participants provided different responses to symptom screening questionnaires between the initial screening and the first medical consultation. Although it is plausible that the observed differences in symptom reporting were due to natural disease progression,24 25 it may also reflect the challenge of capturing TB symptoms accurately in ACF settings. This finding is similar to those reported from a community ACF study in Vietnam, where one third of the study population was symptomatic before the diagnosis and when questioned again after TB diagnosis, 47% reported symptoms.26
Also, studies conducted in carceral settings may be limited by space and security constraints. All our study procedures were performed outside of the prison cell blocks during PDLs’ yard time; it is plausible that our study participants may feel rushed in answering symptom questionnaires because they wanted to resume their activities during this limited time. Since medical consults were only performed in individuals who had tested positive for TB, individuals who are called to the prison clinic might perceive that there must be a reason for the visit (i.e., a positive TB test) and may be more likely to report any symptoms compared with the initial screening.
Another factor that may contribute to the high rate of discordance on symptom reporting is stigma around TB, which could lead to exclusion among cellmates. PDL may deny the presence of any TB symptoms during screening, as responses may be overheard by others. Additionally, individuals frequently associate the two most discordant symptoms (cough and weight loss) with the incarceration process. During closed cell time, burning plastics and paper for light or heat is customary, and more than half of the total PDL population reports current tobacco use, rendering cough a common occurrence. Weight loss is also perceived as a habitual consequence of an insufficient or poor-quality diet inside Brazilian prisons.27 28
In our study, symptomatic TB was associated with higher bacterial burden and higher Lunit scores. This finding is consistent with other studies indicating that higher bacterial burden and more CXR abnormalities are correlated with symptomatic disease.8 29 30 We hypothesise that these findings could be indicators of disease severity31 and thus these patients would more likely be symptomatic.
We also found that individuals who reported having a cellmate with any TB symptoms were more likely to report TB symptoms during the initial screening, the first medical consultation or both. It is plausible that this subgroup represents those who had better awareness of TB symptoms because they had personal experiences with TB. Of note, our study population has been exposed to multiple active TB case finding efforts in the last decade; thus, it is not surprising that some PDL may gain TB knowledge from their prior and current participation in the TB ACF studies. This finding demonstrates that there are factors outside of disease progression and individual burden of disease that could define symptom status. Understanding better these factors is paramount, and future clinical trials for new TB interventions that plan to include asymptomatic TB endpoints32 need to consider the potential influence of these factors.
Our study is subject to several limitations. First, our study was performed in a single prison health centre, serving two prisons, with two physicians evaluating all individuals with TB disease. Thus, our findings may not be generalisable to other carceral settings with different TB burdens or the general population. Also, all the individuals included were male; previous studies have shown that there is variability in symptom reporting between male and female individuals with active TB33 that would not be detected in our study. Second, due to the nature of our systematic screening and prison activities, including transfers, security searches, and riots, the first medical consultation is often delayed. We decided to use the 15-day cut-off to define our inclusion, as TB symptoms typically progress slowly during the earlier phase of the disease, and more distinct/severe symptoms tend to emerge later.24 25 Third, our study may be affected by response bias introduced during the consent process for systematic screening. Although individuals were not informed about their TB diagnosis before the first medical consultation, they may have retained TB-related knowledge from the consent process and/or their past participation in other TB ACF efforts. Clinicians evaluating the individuals with positive TB results were also aware of the diagnosis at the moment of medical consultation. This knowledge may have influenced their perception of TB symptoms. Fourth, we did not compare Lunit scores with physicians’ interpretation of the CXR images, and we did not have a second reader to interpret CXR images to evaluate consistent and objective readings. Lastly, we did not perform a more comprehensive evaluation of mental health aspects, such as mental disorders diagnosis or patterns of substance abuse that could influence symptom reporting.
Conclusions
In conclusion, we found that over one-third of individuals with microbiologically confirmed pulmonary TB provided different answers to symptom inquiry performed at initial screening compared with the first medical consultation. More often, they presented as asymptomatic at first contact and then reported symptoms during consults. Symptom reporting at medical consult was associated with higher bacterial load, lower BMI, current smoking, and reporting cellmates with TB symptoms. Those who reported no symptoms at initial screening but reported symptoms at medical consultation had disease severity markers that fell between those who reported symptoms or no symptoms at both visits. Our findings highlight that CXR findings, Xpert semiquantitative load, tobacco use, weight and perceptions of other symptomatic cellmates are associated with symptom reporting among individuals identified with TB during ACF.
Supplementary material
Acknowledgements
The authors thank the State Agency of Administration Prisons (AGEPEN) for their full support during the study period, the study participants for their kind cooperation during the data collection process and the Central Laboratory (LACEN) of the state of Mato Grosso do Sul for the support in the accomplishment of the laboratory tests.
The funding institute was not involved in the study design, data collection, analysis, interpretation, report writing and the decision to submit for publication.
Footnotes
Funding: This work was supported in part by grants from the National Institutes of Health (NIH) including the National Institute of Allergy and Infectious Diseases (NIAID) (R01AI130058 to JRA and JC).
Provenance and peer review: Not commissioned; externally peer reviewed.
Patient consent for publication: Not applicable.
Ethics approval: This study was submitted, reviewed and approved by the Research Ethics Committee of the Federal University of Mato Grosso do Sul (#5.730.361), the National Ethics Committee of Brazil (CONEP) (#5.899.470) and the Institutional Review Boards (IRB) at Stanford University (IRB#67287). Study participants provided written informed consent prior to study participation. All study procedures were performed in accordance with relevant local guidelines and regulations.
Patient and public involvement: Patients and/or the public were not involved in the design, or conduct, or reporting, or dissemination plans of this research.
Data availability statement
Data are available on reasonable request.
References
- 1.Geneva: World Health Organization; 2025. Global tuberculosis report 2025, 1st ed. [Google Scholar]
- 2.Geneva, Switzerland: 2024. [18-May-2025]. Report of the WHO consultation on asymptomatic tuberculosis.https://www.who.int/publications/i/item/9789240106932 Available. accessed. [Google Scholar]
- 3.Emery JC, Dodd PJ, Banu S, et al. Estimating the contribution of subclinical tuberculosis disease to transmission: An individual patient data analysis from prevalence surveys. Elife. 2023;12:e82469. doi: 10.7554/eLife.82469. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Burke RM, Nliwasa M, Feasey HRA, et al. Community-based active case-finding interventions for tuberculosis: a systematic review. Lancet Public Health. 2021;6:e283–99. doi: 10.1016/S2468-2667(21)00033-5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Bohlbro AS, Hvingelby VS, Rudolf F, et al. Active case-finding of tuberculosis in general populations and at-risk groups: a systematic review and meta-analysis. Eur Respir J. 2021;58:2100090. doi: 10.1183/13993003.00090-2021. [DOI] [PubMed] [Google Scholar]
- 6.Stuck L, van Haaster AC, Kapata-Chanda P, et al. How “Subclinical” is Subclinical Tuberculosis? An Analysis of National Prevalence Survey Data from Zambia. Clin Infect Dis. 2022;75:842–8. doi: 10.1093/cid/ciab1050. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Stuck L, Klinkenberg E, Abdelgadir Ali N, et al. Prevalence of subclinical pulmonary tuberculosis in adults in community settings: an individual participant data meta-analysis. Lancet Infect Dis. 2024;24:726–36. doi: 10.1016/S1473-3099(24)00011-2. [DOI] [PubMed] [Google Scholar]
- 8.Teo AKJ, MacLean E-H, Fox GJ. Subclinical tuberculosis: a meta-analysis of prevalence and scoping review of definitions, prevalence and clinical characteristics. Eur Respir Rev. 2024;33:230208. doi: 10.1183/16000617.0208-2023. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Grimes DA, Schulz KF. Bias and causal associations in observational research. Lancet. 2002;359:248–52. doi: 10.1016/S0140-6736(02)07451-2. [DOI] [PubMed] [Google Scholar]
- 10.Tourangeau R, Yan T. Sensitive questions in surveys. Psychol Bull. 2007;133:859–83. doi: 10.1037/0033-2909.133.5.859. [DOI] [PubMed] [Google Scholar]
- 11.WHO consolidated guidelines on tuberculosis: module 2: screening: systematic screening for tuberculosis disease. [16-Feb-2025]. https://www.who.int/publications/i/item/9789240022676 Available. Accessed. [PubMed]
- 12.Liu YE, Mabene Y, Camelo S, et al. Mass incarceration as a driver of the tuberculosis epidemic in Latin America and projected effects of policy alternatives: a mathematical modelling study. Lancet Public Health. 2024;9:e841–51. doi: 10.1016/S2468-2667(24)00192-0. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Sacchi FPC, Praça RM, Tatara MB, et al. Prisons as reservoir for community transmission of tuberculosis, Brazil. Emerg Infect Dis. 2015;21:452–5. doi: 10.3201/eid2103.140896. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Cords O, Martinez L, Warren JL, et al. Incidence and prevalence of tuberculosis in incarcerated populations: a systematic review and meta-analysis. Lancet Public Health. 2021;6:e300–8. doi: 10.1016/S2468-2667(21)00025-6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Nguyen HV, Tiemersma E, Nguyen NV, et al. Disease Transmission by Patients With Subclinical Tuberculosis. Clin Infect Dis. 2023;76:2000–6. doi: 10.1093/cid/ciad027. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Wang Z, Li H, Song S, et al. Transmission of tuberculosis in an incarcerated population during the subclinical period: A cross-sectional study in Qingdao, China. Front Public Health. 2023;11:1098519. doi: 10.3389/fpubh.2023.1098519. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Mangu CD, Clowes P, van den Hombergh J, et al. New admissions and asymptomatic TB cases seem to fuel TB epidemic in prisons, a cross sectional survey in Tanzania. PLOS Glob Public Health . 2024;4:e0002372. doi: 10.1371/journal.pgph.0002372. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Bezerra W da SP, Jung E, Croda MG, et al. Role of tuberculosis symptoms in transmission risk to cell contacts in prisons. BMJ Public Health. 2025;3:e003218. doi: 10.1136/bmjph-2025-003218. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Dos Santos PCP, da Silva Santos A, de Oliveira RD, et al. Pooling Sputum Samples for Efficient Mass Tuberculosis Screening in Prisons. Clin Infect Dis. 2022;74:2115–21. doi: 10.1093/cid/ciab847. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Harris PA, Taylor R, Thielke R, et al. Research electronic data capture (REDCap)—A metadata-driven methodology and workflow process for providing translational research informatics support. J Biomed Inform. 2009;42:377–81. doi: 10.1016/j.jbi.2008.08.010. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.R Core Team R: a language and environment for statistical computing
- 22.Layton MC, Henning KJ, Alexander TA, et al. Universal radiographic screening for tuberculosis among inmates upon admission to jail. Am J Public Health. 1997;87:1335–7. doi: 10.2105/ajph.87.8.1335. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Telisinghe L, Fielding KL, Malden JL, et al. High tuberculosis prevalence in a South African prison: the need for routine tuberculosis screening. PLoS One. 2014;9:e87262. doi: 10.1371/journal.pone.0087262. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Sossen B, Richards AS, Heinsohn T, et al. The natural history of untreated pulmonary tuberculosis in adults: a systematic review and meta-analysis. Lancet Respir Med. 2023;11:367–79. doi: 10.1016/S2213-2600(23)00097-8. [DOI] [PubMed] [Google Scholar]
- 25.Richards AS, Sossen B, Emery JC, et al. Quantifying progression and regression across the spectrum of pulmonary tuberculosis: a data synthesis study. Lancet Glob Health. 2023;11:e684–92. doi: 10.1016/S2214-109X(23)00082-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Teo AKJ, Luu KB, Garden F, et al. The relationship between a known diagnosis of tuberculosis and symptom reporting: implications for case detection strategies. Eur Respir J. 2025;66:2402521. doi: 10.1183/13993003.02521-2024. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Ribeiro AP. Health conditions of prisoners in the state of Rio de Janeiro, Brazil. Ciênc Saúde Coletiva. 2016;21:2031–40. doi: 10.1590/1413-81232015217.08552016. [DOI] [PubMed] [Google Scholar]
- 28.Soares Filho MM, Bueno PMMG. Demography, vulnerabilities and right to health to Brazilian prison population. Ciênc Saúde Coletiva. 2016;21:1999–2010. doi: 10.1590/1413-81232015217.24102015. [DOI] [PubMed] [Google Scholar]
- 29.Mendelsohn SC, Mulenga H, Tameris M, et al. Screening for asymptomatic tuberculosis among adults with household exposure to a patient with pulmonary tuberculosis. Infectious Diseases (except HIV/AIDS) 2025 doi: 10.1101/2025.01.20.25320843. Preprint. [DOI] [PMC free article] [PubMed]
- 30.Jeong Y-J, Park JS, Kim HW, et al. Characteristics of subclinical tuberculosis compared to active symptomatic tuberculosis using nationwide registry cohort in Korea: prospective cohort study. Front Public Health. 2023;11:1275125. doi: 10.3389/fpubh.2023.1275125. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Grint DJ, Dhillon J, Butcher PD, et al. Xpert MTB/RIF Cycle Threshold as a Marker of Tuberculosis (TB) Disease Severity: Implications for TB Treatment Stratification. Clin Infect Dis. 2026;82:e126–34. doi: 10.1093/cid/ciaf527. [DOI] [PubMed] [Google Scholar]
- 32.Churchyard GJ, Houben RMGJ, Fielding K, et al. Implications of subclinical tuberculosis for vaccine trial design and global effect. Lancet Microbe. 2024;5:100895. doi: 10.1016/S2666-5247(24)00127-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Long NH, Diwan VK, Winkvist A. Difference in symptoms suggesting pulmonary tuberculosis among men and women. J Clin Epidemiol. 2002;55:115–20. doi: 10.1016/s0895-4356(01)00455-3. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
Data are available on reasonable request.