Optimal sampling designs for estimation of Plasmodium falciparum clearance rates in patients treated with artemisinin derivatives

Table 4 Summary of results from reference and alternative schedules

Schedule	Prop OE (of 2746 profiles)	(IQR)	CV (%)	Std dev log(HL)	Prop NZ tlag	Median (IQR) NZ tlag	Prop tail	Prop (%) profiles misclassified, with HL cutoff of:
Schedule	Prop OE (of 2746 profiles)	(IQR)	CV (%)	Std dev log(HL)	Prop NZ tlag	Median (IQR) NZ tlag	Prop tail	3 h	4 h	5 h	6 h
Ref	0%	0 (0–0)	54.5	0.52	26.5 (n = 729)	6 (6–6.25)	34% (n = 933)	0	0	0	0
A1	49.9%	0.01 (-0.15-0.21)	51.6	0.503	22.1 (n = 606)	6 (6–6.27)	9.36% (n = 257)	7.7	3.9	2.7	2.2
A2	60.2%	0.06 (-0.10-0.29)	51.1	0.482	17.4 (n = 478)	6 (6–6.03)	9.87% (n = 271)	8.6	4.4	3	2.2
A3	59.8%	0.09 (-0.09-0.34)	50.9	0.484	12.5 (n = 344)	12 (12–12)	9.91% (n = 272)	9.5	4.5	3	2.1
A4	66.3%	0.15 (-0.06-0.46)	47.7	0.435	6.96 (n = 191)	12 (12–12)	9.25% (n = 254)	10	4.6	3.4	2.2

For each of the reference and alternative schedules (A1-A4) applied to real patient data: the proportion of profiles that overestimate the HL (relative to the reference HL); median difference in HL (IQR); coefficient of variation of HL (CV) as a percentage; standard deviation of log (HL); the proportion of profiles with a non-zero lag phase (tlag); the median (IQR) tlag for profiles with a non-zero tlag; the proportion of profiles with a tail and the proportions of profiles that are misclassified when the HL estimates are dichotomized using HL cut-offs of three, four, five and six hours. IQR = interquartile range; Ref = reference; OE = overestimation (proportion of HLs from alternative schedule which exceeded reference HL); NZ = non-zero; prop = proportion.