Comparing yield and relative costs of WHO TB screening algorithms in selected risk groups among people aged 65 years and over in China, 2013

OBJECTIVE To calculate the yield and cost per diagnosed tuberculosis (TB) case for three World Health Organization screening algorithms and one using the Chinese National TB program (NTP) TB suspect definitions, using data from a TB prevalence survey of people aged 65 years and over in China, 2013. METHODS This was an analytic study using data from the above survey. Risk groups were defined and the prevalence of new TB cases in each group calculated. Costs of each screening component were used to give indicative costs per case detected. Yield, number needed to screen (NNS) and cost per case were used to assess the algorithms. FINDINGS The prevalence survey identified 172 new TB cases in 34,250 participants. Prevalence varied greatly in different groups, from 131/100,000 to 4651/ 100,000. Two groups were chosen to compare the algorithms. The medium-risk group (living in a rural area: men, or previous TB case, or close contact or a BMI <18.5, or tobacco user) had appreciably higher cost per case (USD 221, 298 and 963) in the three algorithms than the high-risk group (all previous TB cases, all close contacts). (USD 72, 108 and 309) but detected two to four times more TB cases in the population. Using a Chest x-ray as the initial screening tool in the medium risk group cost the most (USD 963), and detected 67% of all the new cases. Using the NTP definition of TB suspects made little difference. CONCLUSIONS To "End TB", many more TB cases have to be identified. Screening only the highest risk groups identified under 14% of the undetected cases,. To "End TB", medium risk groups will need to be screened. Using a CXR for initial screening results in a much higher yield, at what should be an acceptable cost.