Scenario | VCs | THCs | CHs | Total |
---|
Scenario 1 |
 1. Initial visit | 50.2%*15.5% | 29.8%*38.8% | 16.0%*38.8% | 25.6% |
 2. Referral from VC | – | 50.2%*7.8%*38.8% | 50.2%*4.9%*38.8% | 2.5% |
 3. Referral from THCa | – | – | 50.2%*7.8%*1.2%*38.8%+ 29.8%*16.0%*38.8% | 1.8% |
Total | 7.8% | 13.1% | 9.0% | 29.8% |
Scenario 2 |
 1. Initial visit | 100%*15.5% | 0.0% | 0.0% | 15.5% |
 2. Referral from VC | – | 100%*7.8%*38.8% | 100%*4.9*38.8% | 4.9% |
 3. Referral from THC | – | – | 100%*7.8%*1.2%*38.8% | 0.0% |
Total | 15.5% | 3.0% | 5.0% | 20.5% |
Scenario 3 |
 1. Initial visit | 0.0% | 100%*38.8% | 0.0% | 38.8% |
 2. Referral from VC | – | – | – |  |
 3. Referral from THC | – | – | 100%*16.0%*38.8% | 6.2% |
Total | 0.0% | 38.8% | 6.2% | 45.0% |
- Note: Based on the appointment data of the patients, three groups were constructed within each scenario: initial visits, referral from a VC, and referral from a THC. We did not collect data on CHs, and, thus, we assume that CHs had correct treatment rates equal to those of the THCs, which is a conservative estimate, as CH providers have been shown to have higher rates of correct treatment than do THC providers when managing disease s[7]. We combined facility-level data (Table 2) and facility sorting behaviors (Supplementary Fig. 1) to calculate the probability of correct management on a facility level. The probability of correct management within the healthcare system was calculated by summing the results of the three levels