Temporal analysis of reported cases of tuberculosis and of tuberculosis-HIV co-infection in Brazil between 2002 and 2012

A comparative analysis of classical and machine learning methods for forecasting TB/HIV co-infection

TB/HIV coinfection poses a complex public health challenge. Accurate forecasting of future trends is essential for efficient resource allocation and intervention strategy development. This study compares classical statistical and machine learning models to predict TB/HIV coinfection cases stratified by gender and the general populations. We analyzed time series data using exponential smoothing and ARIMA to establish the baseline trend and seasonality. Subsequently, machine learning models (SVR, XGBoost, LSTM, CNN, GRU, CNN-GRU, and CNN-LSTM) were employed to capture the complex dynamics and inherent non-linearities of TB/HIV coinfection data. Performance metrics (MSE, MAE, sMAPE) and the Diebold-Mariano test were used to evaluate the model performance. Results revealed that Deep Learning models, particularly Bidirectional LSTM and CNN-LSTM, significantly outperformed classical methods. This demonstrates the effectiveness of Deep Learning for modeling TB/HIV coinfection time series and generating more accurate forecasts.

Tuberculosis forecasting and temporal trends by sex and age in a high endemic city in northeastern Brazil: where were we before the Covid-19 pandemic?

BACKGROUND

The aim of this study was to describe the temporal trend of tuberculosis cases according to sex and age group and evidence the level of disease before the Covid-19 pandemic in a TB high endemic city.

METHODS

This was a time series study carried out in a city in northeast Brazil. The population was composed of cases of tuberculosis, excluding those with HIV-positive status, reported between the years 2002 and 2018. An exploratory analysis of the monthly rates of tuberculosis detection, smoothed according to sex and age group, was performed. Subsequently, the progression of the trend and prediction of the disease were also characterized according to these aspects. For the trends forecast, the seasonal autoregressive linear integrated moving average (ARIMA) model and the usual Box-Jenkins method were used to choose the most appropriate models.

RESULTS

A total of 1620 cases of tuberculosis were reported, with an incidence of 49.7 cases per 100,000 inhabitants in men and 34.0 per 100,000 in women. Regarding the incidence for both sexes, there was a decreasing trend, which was similar for age. Evidence resulting from the application of the time series shows a decreasing trend in the years 2002-2018, with a trend of stability.

CONCLUSIONS

The study evidenced a decreasing trend in tuberculosis, even before the Covid-19 pandemic, for both sex and age; however, in a step really slow from that recommended by the World Health Organization. According to the results, the disease would have achieved a level of stability in the city next years, however it might have been aggravated by the pandemic. These findings are relevant to evidence the serious behavior and trends of TB in a high endemic scenario considering a context prior to the Covid-19 pandemic.

Prevalence and associated factors of tuberculosis treatment abandonment

OBJECTIVE

To estimate the prevalence of tuberculosis treatment abandonment and its associated factors.

METHOD

Cross-sectional study which used cases of tuberculosis in the System of Information on Notification Aggravations (Sistema de Informação de Agravos de Notificação) from 2001 to 2017 in Ceará state. This study included 74,006 cases and the outcome was the closing situation "treatment abandonment". A multivariate analysis was performed to estimate the association between the variables with abandonment.

RESULTS

Throughout the period, the abandonment rate was 12.54%. A higher abandonment prevalence was verified among people who live in the urban zone (PR = 2.45; 95%CI: 2.20-2.74), who are readmitted after abandonment (PR = 2.84; 95%CI: 2.68-3.01), among those notified as recurrent (PR = 1.22; 95%CI: 1.10-1.35) and among drinkers (PR = 1.50; 95%CI: 1.42-1.58). Those who were sputum smear-positive (PR = 1.11; 95%CI: 1.03-1.19) or for whom sputum smear was unperformed (PR = 1.30; 95%CI: 1.20-1.40), coinfection (PR = 2.04; CI95%: 1.89-2.21) and who were not submitted to serology (PR = 1.62; 95%CI: 1.53-1.71) have also a higher prevalence of tuberculosis treatment abandonment.

CONCLUSION

Tuberculosis treatment abandonment is associated to biological and social factors, habits, and health service structure.

Analysis and predication of tuberculosis registration rates in Henan Province, China: an exponential smoothing model study

Background

The World Health Organization End TB Strategy meant that compared with 2015 baseline, the reduction in pulmonary tuberculosis (PTB) incidence should be 20 and 50% in 2020 and 2025, respectively. The case number of PTB in China accounted for 9% of the global total in 2018, which ranked the second high in the world. From 2007 to 2019, 854 672 active PTB cases were registered and treated in Henan Province, China. This study was to assess whether the WHO milestones could be achieved in Henan Province.

Methods

The active PTB numbers in Henan Province from 2007 to 2019, registered in Chinese Tuberculosis Information Management System were analyzed to predict the active PTB registration rates in 2020 and 2025, which is conductive to early response measures to ensure the achievement of the WHO milestones. The time series model was created by monthly active PTB registration rates from 2007 to 2016, and the optimal model was verified by data from 2017 to 2019. The Ljung-Box Q statistic was used to evaluate the model. The statistically significant level is α = 0.05. Monthly active PTB registration rates and 95% confidence interval (CI) from 2020 to 2025 were predicted.

Results

High active PTB registration rates in March, April, May and June showed the seasonal variations. The exponential smoothing winter’s multiplication model was selected as the best-fitting model. The predicted values were approximately consistent with the observed ones from 2017 to 2019. The annual active PTB registration rates were predicted as 49.1 (95% CI: 36.2–62.0) per 100 000 population and 34.4 (95% CI: 18.6–50.2) per 100 000 population in 2020 and 2025, respectively. Compared with the active PTB registration rate in 2015, the reduction will reach 23.7% (95% CI, 3.2–44.1%) and 46.8% (95% CI, 21.4–72.1%) in 2020 and 2025, respectively.

Conclusions

The high active PTB registration rates in spring and early summer indicate that high risk of tuberculosis infection in late autumn and winter in Henan Province. Without regard to the CI, the first milestone of WHO End TB Strategy in 2020 will be achieved. However, the second milestone in 2025 will not be easily achieved unless there are early response measures in Henan Province, China.

Tendência temporal da incidência da coinfecção TB/HIV e testagem de HIV da população idosa brasileira de 2008 a 2018

Resumo Objetivo Analisar a tendência temporal da incidência da coinfecção TB/HIV e a proporção de testagem de HIV na população idosa brasileira e suas regiões, de 2008 a 2018. Método Estudo ecológico, de série temporal, realizado com todos os casos novos de pessoas idosas com coinfecção TB/HIV. Os dados foram extraídos do SINAN-TB e analisados pelo método Prais-Winsten. Resultados Na análise temporal da incidência, o Brasil apresentou tendência estável (VPA 2,1) e as regiões: Sudeste tendência decrescente (VPA -2,15), Nordeste e Norte tendências crescentes (VPA 9,92; VPA 10,18 respectivamente) e Sul e Centro-Oeste tendências estáveis (VPA 0,17; VPA 4,81 respectivamente). No Brasil e nas regiões a proporção de testagem para o HIV mostrou tendências crescentes: Brasil (VPA 12,82), Norte (VPA 20,46), Nordeste (VPA 17,85), Sudeste (VPA 10,29), Sul (VPA 7,11) e Centro-Oeste (VPA 6,10). Dos 3213 casos novos de coinfecção TB/HIV notificados no período estudado, a maior parte (68,66%) foi do sexo masculino, 78,74% na faixa etária de 60 a 69 anos. Quanto à forma clínica a maioria (72,70%) foi do tipo pulmonar. Conclusão Diante da magnitude e implicações da coinfecção TB/HIV na população idosa para os serviços de saúde, esses achados poderão subsidiar profissionais de saúde e gestores na adoção de medidas eficazes no controle dessas doenças.

Impact of the change in the antitubercular regimen from three to four drugs on cure and frequency of adverse reactions in tuberculosis patients from Brazil: A retrospective cohort study

BACKGROUND

The Ministry of Health in Brazil included ethambutol in the intensive phase of sensible tuberculosis (TB) treatment in March 2010, due to the increasing drug resistance, and implemented the fixed dose combination in the TB treatment guidelines.

METHODS

A retrospective cohort study was performed to determine the impact of change from three to four drugs schemes on the TB cure and frequency of adverse drug reactions (ADRs) in TB patients. To answer this question, we used data from 730 randomly selected patients who received anti-TB treatment between January 2007 and December 2014 in a reference center from Salvador, Brazil.

FINDINGS

TB patients who received the RHEZ regimen (n = 365) developed ADRs more frequently than those treated with the RHZ (n = 365) (86 [23.6%] vs. 55 [15.1%]; p = 0.01). This difference in ADR incidence was even higher in patients above 30 years-old (64 [74.4%] vs. 36 [65.5%]; p = 0.01). The overall number of ADR episodes was greater in patients from the RHEZ group than in the group that received RHZ (170 [61.4%] vs. 107 [38.6%]; p = 0.03). Multivariable logistic regression analysis adjusted for age, alcohol use and diabetes demonstrated that patients receiving the RHEZ regimen had increased odds of developing ADRs than those undertaking the RHZ scheme (odds ratio [OR]: 1.61, 95% confidence interval [CI]: 1.10-2.35; p = 0.015). The overall cure rate was similar between the distinct treatment groups.

CONCLUSION

The patients treated with the four-drug regimen exhibited increased risk of ADRs compared to those who received the three-drug regimen, and especially in patients older than 30 years of age.

Temporal pattern of tuberculosis cure, mortality, and treatment abandonment in Brazilian capitals

Objective:

to analyze the temporal pattern of tuberculosis cure, mortality, treatment abandonment in Brazilian capitals.

Method:

this is an ecological study whose data source was the Information System of Notifiable Diseases for Tuberculosis (Sistema de Informação de Agravos de Notificação para Tuberculose). For analysis of temporal evolution, regressions by join points were performed considering the annual percentage variation and the significance of the trend change with 95% confidence interval.

Results:

542,656 cases of tuberculosis were found, with emphasis on a 3% decrease per year in the cure rate for Campo Grande (interval: −5.0 - −0.9) and a 3.5% increase for Rio de Janeiro (interval: 1.9 - 4.7). Regarding abandonment, it decreased 10.9% per year in Rio Branco (interval: −15.8 - −5.7) and increased 12.8% per year in Fortaleza (interval: 7.6 - 18.3). For mortality, a decreasing or stationary tendency was identified, with a greater decrease (7.8%) for Porto Velho (interval:−11.0 - −5.0) and a lower one (2.5%) in Porto Alegre (interval:−4.5 - −0.6).

Conclusion:

the rates of cure and abandonment are far from the ones recommended by the World Health Organization, showing that Brazilian capitals need interventions aimed at changing this pattern.

Sociodemographic and health profile of TB/HIV co-infection in Brazil: a systematic review

OBJECTIVE

to ascertain the epidemiological profile of TB/HIV co-infection in Brazilian scenarios.

METHOD

this is a systematic review conducted via electronic search in databases PubMed, EMBASE, LILACS and SciELO, having as inclusion criterion articles based on the Brazilian scenario of TB/HIV co-infection.

RESULTS

of the total 174 studies, 15 were selected, revealing the epidemiological profile of the co-infection in different scenarios: male, economically active age, low education level, brown/black ethnicity, low income, heterosexual, pulmonary clinical form, alcoholism and Directly Observed Therapy.

CONCLUSION

the sociodemographic and epidemiological profile of people with TB/HIV co-infection has an expected occurrence pattern, which corroborates articles found in the literature, either at national level or by grouping the studies according to region or state.

Clinical and epidemiological features of tuberculosis in children and adolescents

OBJECTIVE

To analyze the clinical and epidemiological features of tuberculosis in children and adolescents in an infectious diseases reference hospital.

METHOD

A documental and retrospective study was carried out with 88 medical files in an infectious diseases reference hospital in the state of Ceará. Data were analyzed by univariate, bivariate and multivariate approaches.

RESULTS

It was found that, depending on the tuberculosis type, its manifestations may vary. The logistic regression model considered only pulmonary tuberculosis due to a number of observations and included female sex (95% CI: 1.4-16.3), weight loss (95% CI: 1.8-26.3), bacilloscopic screening (95% CI: 1.5-16.6) and sputum collected (95% CI: 1.4-19.4) as possible predictors.

CONCLUSIONS

Children and adolescents present different manifestations of the disease depending on the tuberculosis type that affects them. Knowing the most common features of each condition could enhance early diagnosis and, consequently, result in adequate treatment and care.

Characteristics of cases of tuberculosis coinfected with HIV in Minas Gerais State in 2016

This study aimed to characterize the cases of tuberculosis (TB) co-infected with the human immunodeficiency virus (HIV) in Minas Gerais State, Brazil, after the notification sheet modification, and to verify the association between the new variables and the treatment outcome. It is an analytical cross-sectional study with TB/HIV cases notified in the year 2016 to the Brazilian Information System for Notifiable Diseases (Sistema de Informação de Agravos de Notificação). Descriptive statistics, chi-square test, and multiple correspondence analysis were performed to verify the association between the outcome, ageand associated diseases. Of the 180 cases, most were male (75.6%) between 30 and 49 years old (63.3%), mixed ethnicity (black and white) (49.4%), 94.4% had the Acquired Immunodeficiency Syndrome (AIDS) and 60.6% had pulmonary TB. The molecular test was not performed at the time of diagnosis in 70.5% of the cases. Homeless people (4.4%) and prisoners (3.9%) featured prominently among the special populations. People between 40 and 49 years old without concurrent diseases were cured in 40.0% of the cases; 18.9% abandoned the treatment due to smoking, drug abuse and mental illness in the age group between 20 and 29 years old. The deaths were associated with the age group between 30 and 39 years old and the occurrence of AIDS. The results have shown that the groups considered vulnerable (drug users, smokers and people with mental illness) abandoned the treatment, the notification upon death from AIDS in adults was late and some treatments were inadequate. The epidemiological surveillance, prevention and assistance strategies towards cases of TB/HIV must be improved in order to achieve the goal of the Brazilian National Plan to end Tuberculosis as a Public Health Problem until 2035 in the state.

Current use and acceptability of novel diagnostic tests for active tuberculosis: a worldwide survey

OBJECTIVE

To determine the current use and potential acceptance (by tuberculosis experts worldwide) of novel rapid tests for the diagnosis of tuberculosis that are in line with World Health Organization target product profiles.

METHODS

A multilingual survey was disseminated online between July and November of 2016.

RESULTS

A total of 723 individuals from 114 countries responded to the survey. Smear microscopy was the most commonly used rapid tuberculosis test (available to 90.9% of the respondents), followed by molecular assays (available to 70.7%). Only a small proportion of the respondents in middle- and low-income countries had access to interferon-gamma-release assays. Serological and lateral flow immunoassays were used by more than a quarter (25.4%) of the respondents. Among the respondents who had access to molecular tests, 46.7% were using the Xpert assay overall, that proportion being higher in lower middle-income countries (55.6%) and low-income countries (76.6%). The data also suggest that there was some alignment of pricing for molecular assays. Respondents stated they would accept novel rapid tuberculosis tests if available, including molecular assays (acceptable to 86.0%) or biomarker-based serological assays (acceptable to 81.7%). Simple biomarker-based assays were more commonly deemed acceptable in middle- and low-income countries.

CONCLUSIONS

Second-generation molecular assays have become more widely available in high- and low-resource settings. However, the development of novel rapid tuberculosis tests continues to be considered important by tuberculosis experts. Our data also underscore the need for additional training and education of end users.