Table 2 Potential use of serological surveys to help inform programmatic response for three anticipated use cases and nine scenarios

Use case/scenario	Seroconversion rate	Programmatic Response/Examples
Surveillance for elimination of trachoma after halting MDA or during post-validation (80% prior probability of No action needed)
1	SCR ≤ 2.2	No action needed. Morocco-Boumalne Dades-2019 (Fig. 2)
2	SCR > 2.2 & <4.5	No action needed. Additional monitoring may be considered. Morocco-Agdaz-2019 (Fig. 2)
3	SCR ≥ 4.5	Additional monitoring required (clinical, serology, PCR).
Baseline survey to assess trachoma endemicity (50% prior probability of No action needed)
4	SCR ≤ 1.6	No action needed. Togo-Anie-2017, Togo-Keran-2017 (Fig. 2)
5	SCR > 1.6 & <3.8	No action needed. Additional monitoring may be considered. Sudan-Kotom-2019 (Fig. 5)
6	SCR ≥ 3.8	Consider initiating MDA. Sudan-El Seraif-2019, Sudan-Saraf Omrah-2019 (Fig. 5)
Unusual epidemiology based on clinical and PCR markers (50% prior probability of No action needed)
7	SCR ≤ 1.6	Additional monitoring may be considered (serology, PCR) to assess etiology of clinical signs. Papua New Guinea-West New Britain-2015 (Table 1, Fig. 5)
8	SCR > 1.6 & <3.8	Additional monitoring may be considered. Ethiopia-Dera-2017 (Table 1, Fig. 5) Vanuatu-Torba/Malampa/Penama/Shefa/Tafea/Sanma-2016 (Table 1, Fig. 5)
9	SCR ≥ 3.8	Additional monitoring required (serology, PCR).

The scenarios vary the seroconversion rate (SCR) estimated from Pgp3 IgG responses in children aged 1–5 years, per the primary analysis. Illustrative thresholds for the SCR have been provided as examples for how thresholds could be used to guide programmatic decision-making and were chosen for each use case using 90% posterior probability that action is not needed and 90% posterior probability that action is needed. In surveillance for elimination, illustrative thresholds reflect an informative prior assumption of 80% that no action is needed. For baseline survey and unusual epidemiology scenarios, illustrative thresholds reflect an uninformative prior (50% in each category). There were no examples in the present dataset of scenarios 3 and 9, but such results are possible.

Search