Table 3. Comparing Site-Level ANC-RT with and without Continuation from ANC-UAT Sites.

Mean difference and MAE (Mean Absolute Error) between estimated and “true” value of quantities. Results given for each data scenario at 1, 3, and 5 years of simulated ANC-RT data. Each setting (of data scenario and year) uses 50 replicates under different seeds.

Quantity	“True” Value	ANC-RT Data Scenario	Mean Difference			MAE

			Years of RT Data			Years of RT Data

			1	3	5	1	3	5
Prevalence in 2011 (ρ₂₀₁₁)	6.615%	50 Resampled	−0.150%	−0.121%	0.099%	0.465%	0.442%	0.425%

		50 with Continuity	−0.167%	−0.106%	0.112%	0.480%	0.460%	0.421%

Prevalence Change 2010–2011 (ρ₂₀₁₁–ρ₂₀₁₀)	−0.193%	50 Resampled	0.019%	0.019%	0.010%	0.023%	0.023%	0.017%

		50 with Continuity	0.020%	0.018%	0.009%	0.024%	0.022%	0.017%

Prevalence in 2016 (ρ₂₀₁₆)	5.738%	50 Resampled	−0.093%	−0.073%	0.110%	0.391%	0.371%	0.365%

		50 with Continuity	−0.107%	−0.059%	0.118%	0.401%	0.388%	0.359%

Prevalence Change 2015–2016 (ρ₂₀₁₆–ρ₂₀₁₅)	−0.179%	50 Resampled	0.009%	0.007%	0.001%	0.015%	0.015%	0.013%

		50 with Continuity	0.010%	0.007%	0.000%	0.016%	0.015%	0.013%

Calibration btw UAT and RT (β_RT)	−0.1000	50 Resampled	0.0438	0.0352	0.0281	0.0602	0.0539	0.0494

		50 with Continuity	−0.0098	−0.0191	−0.0211	0.0331	0.0262	0.0276

Calibration btw UAT and pregnant women (α_UAT)	0.2402	50 Resampled	−0.0273	−0.0229	−0.0378	0.0466	0.0467	0.0533

		50 with Continuity	0.0177	0.0115	0.0118	0.0536	0.0480	0.0597

Shared Non-Sampling Variance of UAT and RT ( $δ_{S P}^{2}$ )	0.01301	50 Resampled	0.00125	−0.00028	−0.00008	0.00426	0.00157	0.00119

		50 with Continuity	0.00072	−0.00042	−0.00007	0.00322	0.00181	0.00118