2023-11-05-世界卫生组织-Methods_used_by_WHO_to_estimate_the_global_burden_of_TB_disease_68页_907kb

WHO Methods to Estimate the Global Burden of TB Disease (2000–2022)

Core Content Summary

The World Health Organization (WHO) employs a variety of methods to estimate the global burden of tuberculosis (TB) disease, including incidence and mortality, and to account for drug-resistant TB and the impact of the COVID-19 pandemic. These methods are tailored to different data availability and quality scenarios across countries.

Main Methods for TB Incidence Estimation (2000–2019)

WHO uses five main methods to estimate TB incidence:

1. Method 1 - Results from TB prevalence surveys

   • Uses prevalence data combined with disease duration estimates.
   • Assumes epidemic equilibrium to estimate incidence.
   • Two approaches are used: dividing prevalence by average duration and using compartments (susceptible, untreated, treated) to estimate transitions.
   • Limitations include the difficulty in verifying assumptions about disease duration and under-reporting.

2. Method 2 - National TB prevalence survey + country-specific dynamic model

   • Applied in India, which accounted for 27% of global TB cases in 2019.
   • Combines prevalence survey data with a dynamic model that incorporates drug sales, mortality estimates, and HIV prevalence.
   • Model structure includes compartments for those awaiting diagnosis and those who have sought care.
   • The model was calibrated using updated mortality data and reviewed in 2023.

3. Method 3 - Case notification data + expert opinion on case detection gaps

   • Uses case notifications and expert estimates of under-reporting and under-diagnosis.
   • Incidence is calculated using the formula: $I = \frac{f(N)}{1 - g}$, where $N$ is case notifications and $g$ is the detection gap.
   • Assumes incidence follows a horizontal trend if insufficient data is available.
   • Limitations include small expert sample sizes and potential over-reporting due to systematic screening.

4. Method 4 - Notifications in high-income countries adjusted by a standard factor

   • Applies a standard adjustment factor (85%) to account for under-reporting and under-diagnosis.
   • Used for 139 countries, including most high-income countries and some upper-middle-income countries.
   • Mortality in high-income countries is estimated from vital registration data.

5. Method 5 - Inventory studies and capture-recapture modelling

   • Used in 8 countries: China, Egypt, Indonesia, Iraq, the Netherlands, the United Kingdom, and Yemen.
   • Requires at least three data sources and assumes a closed population.
   • Capture-recapture is used to estimate the proportion of unreported cases.
   • Limitations include the need for multiple data sources and potential biases in data collection.

Estimation of HIV-Positive TB Incidence

• Provider-initiated testing and counselling with ≥50% HIV testing coverage is the primary source of HIV prevalence in TB patients.
• However, this data is subject to selection biases.
• Additional data sources include sero-surveys and HIV sentinel surveillance systems.

Mortality Estimation

• Mortality is estimated using national vital registration systems where available.
• In other countries, mortality is derived indirectly from incidence and case fatality rates.
• Specific estimates for India and HIV-positive populations are provided, using a combination of prevalence data, mortality surveys, and expert opinions.

Uncertainty in Estimation

• Uncertainty is estimated using statistical methods, including the Beta distribution and variance calculations.
• The method of moments is used to derive parameters for incidence estimation.
• Uncertainty intervals are provided for most estimates, reflecting the reliability of the data and assumptions.

Impact of the COVID-19 Pandemic on TB Burden (2020–2022)

• Dynamic models are used to estimate the impact of reduced TB services during the pandemic.
• Country-specific and region-specific models are applied to countries with significant reductions in case notifications.
• Region-specific models are used for 23 countries with ≥10% cumulative reduction in case notifications from 2020 to 2021.
• Models account for disruptions in TB detection, treatment, and care.

Disaggregation by Age and Sex (2022)

• TB incidence and mortality are disaggregated by age and sex.
• This allows for a more detailed understanding of disease burden across different demographics.

Drug-Resistant TB Estimation

• WHO estimates the proportion of TB cases with rifampicin resistance (RR-TB) and the incidence of MDR/RR-TB.
• These estimates are based on inventory studies and capture-recapture models.
• The proportion of RR-TB cases with fluoroquinolone resistance is also estimated.

Attributable Risk for TB (2022)

• Attributable risk is calculated using risk ratios, exposed population, and population attributable fractions.
• This helps in understanding the contribution of risk factors to TB incidence.

Key Findings

• The five methods used for TB incidence estimation are tailored to different data availability and quality.
• Dynamic models are essential for countries with significant changes in TB case notifications, particularly during the pandemic.
• Expert opinion plays a critical role in estimating case detection gaps and incidence trends.
• Capture-recapture and inventory studies provide more accurate incidence estimates but are limited in scope due to data requirements.
• HIV prevalence among TB cases is a key factor in estimating incidence and mortality for HIV-positive populations.

Limitations and Considerations

• Methods based on prevalence surveys are subject to assumptions that are difficult to verify.
• Expert opinion methods are prone to biases and limited data.
• Capture-recapture is only feasible in a few high-burden countries due to the need for multiple data sources.
• The impact of the pandemic on TB services is estimated using dynamic models, which require calibration and are subject to their own limitations.

Conclusion

WHO's methods for estimating TB burden are diverse and context-specific, reflecting the varying data quality and availability across countries. These methods are continuously reviewed and updated to improve accuracy and reliability, particularly in light of the challenges posed by the pandemic and drug resistance.