Table 3 Impact of stochastic variation when estimating parasite density from thick or thin films with nominated parasite density. Results from simulations of counting parasites on thick films against 200 WBC and 500 WBC (assuming 8000 WBC per µL; (A)) or on a thin film per 1000 RBC and 2000 RBC (assuming 5 × 10⁶ RBC per µL; (B)). Parasite density estimates obtained from a single read by counting parasites per 500 WBC or 2000 RBC tend to be closer to the actual density. When the parasite density is estimated from two readings, fewer paired reads are discordant per Obare criteria [30]. Parasite density estimates were defined as being discordant if there is a < 10% chance of observing the two read densities if both were random samples from the theoretical probability distribution with mean equal to the average of the first and second read densities, given that at least one of the first or second reads has a density for a particular species above 200 parasites/µL

(A) Thick film counts

Against 200 WBC

Distribution of single read (n = 1000)

Mean (sd)

199.8 (89.1)

250.8 (101.6)

496.7 (138.9)

983.5 (200.1)

5018.5 (565.1)

10,041.7 (904.6)

16,001.9 (1357.9)

25th and 75th percentiles

120 and 261

189 and 314

396 and 585

839 and 1121

4619 and 5397

9412 and 10,624

15,080 and 16,956

5th and 95th percentiles

76 and 355

82 and 429

272 and 731

673 and 1320

4121 and 5989

8610 and 11,604

13,822 and 18,339

Simulation of 2 microscopy readings (n = 10,000)

% with both readings ≤ 200/uL

34.3

13.9

0.02

0.0

0.0

0.0

0.0

% requiring 3rd read (Obare criteria)

12.7

15.4

15.3

15.4

23.1

28.7

36.9

Against 500 WBC

Distribution of single read (n = 1000)

Mean (sd)

199.7 (57.0)

248.9 (64.5)

499.5 (89.2)

998.3 (133.8)

5004.5 (354.2)

9995.1 (579.5)

15,964.0 (883.0)

25th and 75th percentiles

158 and 238

204 and 287

441 and 559

910 and 1085

4756 and 5247

9607 and 10,372

15,395 and 16,556

5th and 95th percentiles

112 and 300

144 and 365

351 and 648

780 and 1213

4441 and 5605

9076 and 10,992

14,534 and 17,524

Simulation of 2 microscopy readings (n = 10,000)

% with both readings ≤ 200/µL

25.3

4.9

0.0

0.0

0.0

0.0

0.0

% requiring 3rd read (Obare criteria)

1.9

3.0

2.5

2.8

5.5

9.8

16.9

(B) Thin film counts

Against 1000 RBC

Distribution of single read (n = 1000)

Mean (sd)

16,328.4 (8892.7)

19,979.2 (9541.3)

32,539.0 (12,592.1)

63,489.3 (17,600.1)

127,011.1 (23,853.1)

200,712.7 (32,038.7)

252,236.6 (33,445.7)

25th and 75th  %iles

9609 and 22,386

13,915 and 24,919

23,747 and 39,969

51,450 and 75,794

110,391 and 142,273

179,076 and 221,459

229,845 and 274,556

5th and 95th  %iles

4617 and 33,049

4818 and 37,912

14,020 and 53,447

34,967 and 93,502

89,799 and 168,045

147,499 and 254,466

194,879 and 306,865

Simulation of 2 microscopy readings (n = 10,000)

% requiring 3rd read (Obare criteria)

11.5

12.1

13.6

14.5

11.9

14.5

10.6

Against 2000 RBC

Distribution of single read (n = 1000)

Mean (sd)

16,116.9 (6051.3)

20,140.1 (6869.9)

3773.0 (9055.7)

64,819.4 (12,493.7)

128,216.7 (17,857.2)

199,747.5 (22,125.0)

249,103.6 (25,238.4)

25th and 75th  %iles

11,872 and 19,666

15,956 and 24,070

25,668 and 37,926

56,610 and 72,317

115,908 and 140,332

183,871 and 215,118

231,741 and 265,957

5th and 95th  %iles

7076 and 26,527

9397 and 31,699

18,248 and 47,561

45,282 and 86,739

99,791 and 159,231

164,356 and 235,178

209,136 and 291,824

Simulation of 2 microscopy readings (n = 10,000)

% requiring 3rd read (Obare criteria)

2.5

3.1

4.6

3.7

3.5

3.3

3.5