Machine learning algorithms using national registry data to predict loss to follow-up during tuberculosis treatment.

BACKGROUND: Identifying patients at increased risk of loss to follow-up (LTFU) is key to developing strategies to optimize the clinical management of tuberculosis (TB). The use of national registry data in prediction models may be a useful tool to inform healthcare workers about risk of LTFU. Here we developed a score to predict the risk of LTFU during anti-TB treatment (ATT) in a nationwide cohort of cases using clinical data reported to the Brazilian Notifiable Disease Information System (SINAN). METHODS: We performed a retrospective study of all TB cases reported to SINAN between 2015 and 2022; excluding children (< 18 years-old), vulnerable groups or drug-resistant TB. For the score, data before treatment initiation were used. We trained and internally validated three different prediction scoring systems, based on Logistic Regression, Random Forest, and Light Gradient Boosting. Before applying our models we splitted our data into training (~ 80% data) and test (~ 20%) sets, and then compared the model metrics using the test data set. RESULTS: Of the 243,726 cases included, 41,373 experienced LTFU whereas 202,353 were successfully treated. The groups were different with regards to several clinical and sociodemographic characteristics. The directly observed treatment (DOT) was unbalanced between the groups with lower prevalence in those who were LTFU. Three models were developed to predict LTFU using 8 features (prior TB, drug use, age, sex, HIV infection and schooling level) with different score composition approaches. Those prediction scoring systems exhibited an area under the curve (AUC) ranging between 0.71 and 0.72. The Light Gradient Boosting technique resulted in the best prediction performance, weighting specificity and sensitivity. A user-friendly web calculator app was developed ( https://tbprediction.herokuapp.com/ ) to facilitate implementation. CONCLUSIONS: Our nationwide risk score predicts the risk of LTFU during ATT in Brazilian adults prior to treatment commencement utilizing schooling level, sex, age, prior TB status, and substance use (drug, alcohol, and/or tobacco). This is a potential tool to assist in decision-making strategies to guide resource allocation, DOT indications, and improve TB treatment adherence.

Causes of death in children with congenital Zika syndrome in Brazil, 2015 to 2018: A nationwide record linkage study.

BACKGROUND: Children with congenital Zika syndrome (CZS) have severe damage to the peripheral and central nervous system (CNS), greatly increasing the risk of death. However, there is no information on the sequence of the underlying, intermediate, immediate, and contributing causes of deaths among these children. The aims of this study are describe the sequence of events leading to death of children with CZS up to 36 months of age and their probability of dying from a given cause, 2015 to 2018. METHODS AND FINDINGS: In a population-based study, we linked administrative data on live births, deaths, and cases of children with CZS from the SINASC (Live Birth Information System), the SIM (Mortality Information System), and the RESP (Public Health Event Records), respectively. Confirmed and probable cases of CZS were those that met the criteria established by the Brazilian Ministry of Health. The information on causes of death was collected from death certificates (DCs) using the World Health Organization (WHO) DC template. We estimated proportional mortality (PM%) among children with CZS and among children with non-Zika CNS congenital anomalies (CA) by 36 months of age and proportional mortality ratio by cause (PMRc). A total of 403 children with confirmed and probable CZS who died up to 36 months of age were included in the study; 81.9% were younger than 12 months of age. Multiple congenital malformations not classified elsewhere, and septicemia unspecified, with 18 (PM = 4.5%) and 17 (PM = 4.2%) deaths, respectively, were the most attested underlying causes of death. Unspecified septicemia (29 deaths and PM = 11.2%) and newborn respiratory failure (40 deaths and PM = 12.1%) were, respectively, the predominant intermediate and immediate causes of death. Fetuses and newborns affected by the mother's infectious and parasitic diseases, unspecified cerebral palsy, and unspecified severe protein-caloric malnutrition were the underlying causes with the greatest probability of death in children with CZS (PMRc from 10.0 to 17.0) when compared to the group born with non-Zika CNS anomalies. Among the intermediate and immediate causes of death, pneumonitis due to food or vomiting and unspecified seizures (PMRc = 9.5, each) and unspecified bronchopneumonia (PMRc = 5.0) were notable. As contributing causes, fetus and newborn affected by the mother's infectious and parasitic diseases (PMRc = 7.3), unspecified cerebral palsy, and newborn seizures (PMRc = 4.5, each) were more likely to lead to death in children with CZS than in the comparison group. The main limitations of this study were the use of a secondary database without additional clinical information and potential misclassification of cases and controls. CONCLUSION: The sequence of causes and circumstances involved in the deaths of the children with CZS highlights the greater vulnerability of these children to infectious and respiratory conditions compared to children with abnormalities of the CNS not related to Zika.

Machine learning algorithms using national registry data to predict loss to follow- up during tuberculosis treatment.

Background: Identifying patients at increased risk of loss to follow-up (LTFU) is key to developing strategies to optimize the clinical management of tuberculosis (TB). The use of national registry data in prediction models may be a useful tool to inform healthcare workers about risk of LTFU. Here we developed a score to predict the risk of LTFU during anti-TB treatment (ATT) in a nationwide cohort of cases using clinical data reported to the Brazilian Notifiable Disease Information System (SINAN). Methods: We performed a retrospective study of all TB cases reported to SINAN between 2015-2022; excluding children (<18 years-old), vulnerable groups or drug-resistant TB. For the score, data before treatment initiation were used. We trained and internally validated three different prediction scoring systems, based on Logistic Regression, Random Forest, and Light Gradient Boosting. Before applying our models we split our data into train (~80% data) and test (~20%), and then we compare model metrics using a test data set. Results: Of the 243,726 cases included, 41,373 experienced LTFU whereas 202,353 were successfully treated and cured. The groups were different with regards to several clinical and sociodemographic characteristics. The directly observed treatment (DOT) was unbalanced between the groups with lower prevalence in those who were LTFU. Three models were developed to predict LTFU using 8 features (prior TB, drug use, age, sex, HIV infection and schooling level) with different score composition approaches. Those prediction scoring system exhibited an area under the curve (AUC) ranging between 0.71 and 0.72. The Light Gradient Boosting technique resulted in the best prediction performance, weighting specificity, and sensibility. A user-friendly web calculator app was created (https://tbprediction.herokuapp.com/) to facilitate implementation. Conclusions: Our nationwide risk score predicts the risk of LTFU during ATT in Brazilian adults prior to treatment commencement. This is a potential tool to assist in decision-making strategies to guide resource allocation, DOT indications, and improve TB treatment adherence.

Tuberculosis treatment outcomes of diabetic and non-diabetic TB/HIV co-infected patients: A nationwide observational study in Brazil.

Background: Tuberculosis (TB) is a worldwide public health problem, especially in countries that also report high numbers of people living with HIV (PLWH) and/or diabetes mellitus (DM). However, the unique features of persons with TB-HIV-DM are incompletely understood. This study compared anti-TB treatment (ATT) outcomes of diabetic and non-diabetic TB/HIV co-infected patients. Methods: A nationwide retrospective observational investigation was performed with data from the Brazilian Tuberculosis Database System among patients reported to have TB-HIV co-infection between 2014 and 2019. This database includes all reported TB cases in Brazil. Exploratory and association analyses compared TB treatment outcomes in DM and non-DM patients. Unfavorable outcomes were defined as death, treatment failure, loss to follow-up or recurrence. Multivariable stepwise logistic regressions were used to identify the variables associated with unfavorable ATT outcomes in the TB-HIV population. Results: Of the 31,070 TB-HIV patients analyzed, 999 (3.2%) reported having DM. However, in these TB-HIV patients, DM was not associated with any unfavorable treatment outcome [adjusted Odds Ratio (aOR): 0.97, 95% CI: 0.83-1.12, p = 0.781]. Furthermore, DM was also not associated with any specific type of unfavorable outcome in this study. In both the TB-HIV group and the TB-HIV-DM subpopulation, use of alcohol, illicit drugs and tobacco, as well as non-white ethnicity and prior TB were all characteristics more frequently observed in persons who experienced an unfavorable ATT outcome. Conclusion: DM is not associated with unfavorable TB treatment outcomes in persons with TB-HIV, including death, treatment failure, recurrence and loss to follow up. However, consumption habits, non-white ethnicity and prior TB are all more frequently detected in those with unfavorable outcomes in both TB-HIV and TB-HIV-DM patients.

The Effect of Diabetes and Prediabetes on Antituberculosis Treatment Outcomes: A Multicenter Prospective Cohort Study.

BACKGROUND: It is unclear whether diabetes or prediabetes affects unfavorable treatment outcomes and death in people with tuberculosis (PWTB). METHODS: Culture-confirmed, drug-susceptible PWTB, enrolled in the Regional Prospective Observational Research in Tuberculosis (RePORT)-Brazil cohort between 2015 and 2019 (N = 643) were stratified based on glycemic status according to baseline glycated hemoglobin. Unfavorable tuberculosis (TB) outcome was defined as treatment failure or modification, recurrence, or death; favorable outcome was cure or treatment completion. We corroborated the findings using data from PWTB reported to the Brazilian National System of Diseases Notification (SINAN) during 2015-2019 (N = 20 989). Logistic regression models evaluated associations between glycemic status and outcomes. RESULTS: In both cohorts, in univariate analysis, unfavorable outcomes were more frequently associated with smoking, illicit drug use, and human immunodeficiency virus infection. Diabetes, but not prediabetes, was associated with unfavorable outcomes in the RePORT-Brazil (adjusted relative risk [aRR], 2.45; P < .001) and SINAN (aRR, 1.76; P < .001) cohorts. Furthermore, diabetes was associated with high risk of death (during TB treatment) in both RePORT-Brazil (aRR, 2.16; P = .040) and SINAN (aRR, 1.93; P = .001). CONCLUSIONS: Diabetes was associated with an increased risk of unfavorable outcomes and mortality in Brazilian PWTB. Interventions to improve TB treatment outcomes in persons with diabetes are needed.

Tuberculosis Burden and Determinants of Treatment Outcomes According to Age in Brazil: A Nationwide Study of 896,314 Cases Reported Between 2010 and 2019.

Approximately 1.4 million people die annually worldwide from tuberculosis. Large epidemiologic studies can identify determinants of unfavorable clinical outcomes according to age, which can guide public health policy implementation and clinical management to improve outcomes. We obtained data from the national tuberculosis case registry; data were reported to the Brazilian National Program (SINAN) between 2010 and 2019. Clinical and epidemiologic variables were compared between age groups (child: <10 years, young: 10-24years, adult: 25-64years, and elderly: ≥65years). Univariate comparisons were performed together with second-generation p-values. We applied a backward stepwise multivariable logistic regression model to identify characteristics in each age group associated with unfavorable TB treatment outcomes. There were 896,314 tuberculosis cases reported during the period. Tuberculosis incidence was highest among adult males, but the young males presented the highest growth rate during the period. Directly observed therapy (DOT) was associated with protection against unfavorable outcomes in all age groups. The use of alcohol, illicit drugs, and smoking, as well as occurrence of comorbidities, were significantly different between age groups. Lack of DOT, previous tuberculosis, race, location of tuberculosis disease, and HIV infection were independent risk factors for unfavorable outcome depending on the age group. The clinical and epidemiological risk factors for unfavorable tuberculosis treatment outcomes varied according to age in Brazil. DOT was associated with improved outcomes in all age groups. Incidence according to age and sex identified adults and young males as the groups that need prevention efforts. This supports implementation of DOT in all populations to improve tuberculosis outcomes.

Prevalence and Clinical Profiling of Dysglycemia and HIV Infection in Persons With Pulmonary Tuberculosis in Brazil.

BACKGROUND: There are scarce data on the prevalence and disease presentation of HIV in patients with tuberculosis (TB) and dysglycemia (diabetes [DM] and prediabetes [PDM]), especially in TB-endemic countries. METHODS: We assessed the baseline epidemiological and clinical characteristics of patients with culture-confirmed pulmonary TB, enrolled in a multicenter prospective cohort in Brazil (RePORT-Brazil) during 2015-2019. Dysglycemia was defined by elevated glycated hemoglobin and stratified as PDM or DM. Additionally, we used data from TB cases obtained through the Brazilian National Notifiable Diseases Information System (SINAN), during 2015-2019. In SINAN, diagnosis of diabetes was based on self-report. Logistic regression models were performed to test independent associations between HIV, dysglycemia status, and other baseline characteristics in both cohorts. RESULTS: In the RePORT-Brazil cohort, the prevalence of DM and of PDM was 23.7 and 37.8%, respectively. Furthermore, the prevalence of HIV was 21.4% in the group of persons with TB-dysglycemia and 20.5% in that of patients with TBDM. In the SINAN cohort, the prevalence of DM was 9.2%, and among the TBDM group the prevalence of HIV was 4.1%. Logistic regressions demonstrated that aging was independently associated with PDM or DM in both the RePORT-Brazil and SINAN cohorts. In RePORT-Brazil, illicit drug use was associated with PDM, whereas a higher body mass index (BMI) was associated with DM occurrence. Of note, HIV was not associated with an increased risk of PDM or DM in patients with pulmonary TB in both cohorts. Moreover, in both cohorts, the TBDM-HIV group presented with a lower proportion of positive sputum smear and a higher frequency of tobacco and alcohol users. CONCLUSION: There is a high prevalence of dysglycemia in patients with pulmonary TB in Brazil, regardless of the HIV status. This reinforces the idea that DM should be systematically screened in persons with TB. Presence of HIV does not substantially impact clinical presentation in persons with TBDM, although it is associated with more frequent use of recreational drugs and smear negative sputum samples during TB screening.

Novel stepwise approach to assess representativeness of a large multicenter observational cohort of tuberculosis patients: The example of RePORT Brazil.

BACKGROUND: A major goal of tuberculosis (TB) epidemiological studies is to obtain results that can be generalized to the larger population with TB. The ability to extrapolate findings on the determinants of TB treatment outcomes is also important. METHODS: We compared baseline clinical and demographic characteristics and determinants of anti-TB treatment outcomes between persons enrolled in the Regional Prospective Observational Research in Tuberculosis (RePORT)-Brazil cohort between June 2015 and June 2019, and the registry of TB cases reported to the Brazilian National TB Program (Information System for Notifiable Diseases [SINAN]) during the same time period. Multivariable regression models adjusted for the study site were performed using second-generation p-values, a novel statistical approach. Associations with unfavorable treatment outcomes were tested for both RePORT-Brazil and SINAN cohorts. FINDINGS: A total of 1,060 culture-confirmed TB patients were enrolled in RePORT-Brazil and 455,873 TB cases were reported to SINAN. Second-generation p-value analyses revealed that the cohorts were strikingly similar with regard to sex, age, use of antiretroviral therapy and positive initial smear sputum microscopy. However, diabetes, HIV infection, and smoking were more frequently documented in RePORT-Brazil. Illicit drug use, the presence of diabetes, and history of prior TB were associated with unfavorable TB treatment outcomes; illicit drug use was associated with such outcomes in both cohorts. CONCLUSIONS: There were important similarities in demographic characteristics and determinants of clinical outcomes between the RePORT-Brazil cohort and the Brazilian National registry of TB cases.

Effect of acute and chronic exposure to ammonia on different larval instars of Anopheles darlingi (Diptera: Culicidae).

Anopheles darlingi (Diptera: Culicidae) is the most important vector of malaria in South America and has already been found in peri-urban areas that commonly contain toxic nitrogenous compounds, such as ammonia. The adaptation of mosquitoes to polluted breeding sites can increase their distribution and affect the dynamics of vector-borne diseases such as malaria. Therefore, the present study investigated the tolerance of larval instars of An. darlingi to ammonia under acute and chronic exposure conditions. Anopheles darlingi larval mortality, development time, and pupal and adult production using larvae of the 1st (L1) and 3rd (L3) instar were assessed as both acute and chronic effects of exposure to different concentrations of ammonia. Lethal concentrations (LCs) for L1 larvae were lower than LCs for L3 larvae. In general, higher ammonia concentrations caused an increase in larval mortality, especially in chronically exposed L1 larvae. The larval development time in L1 and L3 was longer with chronic treatment and decreased with increasing concentrations of ammonia. The number of pupae was very low for acutely exposed L1 and L3 larvae. Likewise, the probability of adult production decreased with increasing ammonia concentrations. This is the first report on the tolerance of An. darlingi to pollutants.