Plasmodium falciparum parasitaemia is normally estimated by counting on microscopic examination of Giemsa-stained blood smears the number of parasites after a pre-determined number of white blood cells (normally 100–500) has been counted and assuming a standard white cell count (WCC, normally 8000/μl). The following formula is generally applied: parasites/μl = [(number of parasites counted/number of leucocytes counted) × 8000 leucocytes/μl] [World Health Organization (WHO) 2010].
This approximation is expedient as WCC normally cannot be determined in most primary health care settings of the malaria-endemic areas. However, WCC vary greatly and are age-dependent. As opposed to ‘western’ standards (which have 5.0–11.9 WBC × 109/l for all children under 6 years of age), a study in Mozambique reported intervals of 2.2–16.6 in children 1–2 years old, 5.8–5.6 in 2- to 3-year-olds, 5.5–14.7 in 3- to 4-year-olds and 5.6–14.4 in 4- to 5-year-olds (Quintóet al. 2006). Several studies, predominantly in malaria-infected adults, concur that WCC increase in acute malaria (Erhart et al. 2004; McKenzie et al. 2005). Therefore, this estimation may generate systematic errors and lead to incorrect conclusions. The present analysis was conducted to assess the correlation between parasitaemia as generally estimated and parasitaemia calculated based on a patient’s WCC.
Children under 5 years of age were selected because they are the most vulnerable group and constitute the study population in the majority of clinical trials of antimalarial drugs in sub-Saharan Africa. Data on age, parasitaemia and white blood cell total and differential counts were extracted from a database of seven randomised controlled trials of artesunate plus amodiaquine vs. various comparator drugs for treating acute uncomplicated falciparum malaria at 14 sites in 10 sub-Saharan African countries (on average 217 patients per site or 304 patients per country). These studies were conducted between February 1999 in Kenya and December 2006 in Senegal. All were approved by local or national and WHO ethics committees. The designs of these clinical trials were similar and based on the WHO (2006) protocol, as described by Zwang et al. (2009). Here, analyses were restricted to the patients under 5 years of age with recorded WCC at baseline (n = 3044, i.e. 26% of the 11 700 patients enroled in the trials).
In the original studies, parasitaemia (number/μl) was quantified by a standard approximation method (40 × number of parasites per 200 WCCs on thick film, i.e. assuming 8000 WCCs/μl) in six of the seven studies included in this analysis (Adjuik et al. 2002; Mårtensson et al. 2005; Karema et al. 2006; Dorsey et al. 2007; Djimdéet al. 2008; Sirima et al. 2009). The 7th study counted P. falciparum, assuming an average WCC of 7500/μl (Ndiaye et al. 2009). For the purpose of coherence, we also recalculated the estimated parasitaemia based on 8000 WCCs for the present study.
Parasitaemia on admission was recalculated for all studies assuming 200 WCCs for all patients vs. each patients’ own actual WCC on admission. The Wilcoxon signed rank test was used to compare both methods of estimating parasitaemia on paired patient data; the Wilcoxon rank test was used for non-paired data. For multivariate analysis, we used a linear model between age (in years) and parasitaemia (log-transformed) with a random effect for study site. P-values under 0.05 were considered significant. The statistical programme used was STATA (version 10; Stata Corp.).
The geometric mean of the actual pre-treatment parasitaemia was 18 334 per μl overall (Table 1). Combining data from all sites over the entire 0–4 year age range, no significant difference (+2%, P = 0.277) was detected between estimated and actual parasitaemia. However, statistically significant differences were detected when data were stratified by age and site. Using paired analysis, parasitaemia estimated using standardised WCC was significantly lower than when computed using actual WCC in children under 1 year of age (−14%, P = 0.001) and slightly higher in 4-year-olds (+8%, P = 0.076). Differences were detected at 11 of the 14 sites between the two methods, with five sites overestimating (Zanzibar and Rwanda where WCC were lower) and six underestimating parasitaemia (no significant difference at three sites) (Table 2). In the sites with significant differences between the two methods, the overall percentage overestimation of the geometric mean was twice (+37%) as much as that of the underestimation (−19%) corresponding approximately to the difference in WCC (10.6 and 6.6 × 109/l, respectively, P = 0.001) and in age (2 vs. 3 years, P = 0.001). Overall in these sites, the difference in parasitaemia estimated based on actual (17 852/μl) and standardised WCC (18 210/μl) was not significant (+2%, P = 0.313, paired analysis).
|Age (year)||N||WBC (× 109/l)||Parasitaemia calculated on actual WCC||Parasitaemia estimated on standardised 8000 WCC||Relative difference (%)||P|
|<1||220||10 167||15 125||13 031||−16||0.001|
|1||370||9634||18 365||17 822||−3||0.195|
|2||863||9095||19 905||19 988||0||0.238|
|3||866||8507||17 920||18 749||4||0.395|
|4||725||8072||18 090||19 603||8||0.076|
|Total||3044||8827||18 334||18 681||2||0.277|
|Site||N||Age (years)||WCC (× 109/l)||Parasitaemia calculated on actual WCC||Parasitaemia estimated on standardised 8000 WCC||P|
|Burkina Faso-Pouytenga||582||2.0||1.2||11.5||4.8||18 797||14 151||0.001|
|Cameroon-Mendong||72||2.9||0.9||8.2||2.1||57 345||54 353||0.146|
|Gabon-Lambéréné||71||3.2||0.9||8.2||2.9||26 492||27 597||0.920|
|Kenya-Migori||28||2.4||1.6||9.9||4.2||30 864||27 402||0.007|
|Madagascar-Tsiroanomandidy||55||3.3||0.8||7.7||2.9||10 145||10 456||0.409|
|Mali-Bancoumana||94||3.0||0.9||10.6||3.6||31 787||23 813||0.001|
|Mali-Bougoula||669||2.3||1.2||10.4||4.4||19 164||15 935||0.001|
|Rwanda-Kicukiro||174||2.9||1.0||6.8||2.6||27 999||35 295||0.001|
|Rwanda-Mashesha||218||2.8||1.0||5.4||1.6||14 331||21 942||0.001|
|Rwanda-Rukara||250||2.3||1.0||4.0||1.6||17 641||37 676||0.001|
|Senegal-Mlomp||204||3.1||0.9||11.2||4.6||25 846||20 129||0.001|
|Uganda-Kampala||249||3.2||0.8||9.1||3.8||11 682||11 035||0.020|
|Zanzibar-Kivunge||278||2.4||1.2||6.3||3.1||11 777||16 479||0.001|
|Zanzibar-Micheweni||100||2.0||1.2||6.5||1.5||13 432||17 026||0.001|
|Total||3044||2.5||1.2||8.8||4.5||18 334||18 681||0.277|
Using a linear random effects model, there was a trend for older children having higher log-transformed parasitaemia when using standardised WCC (r = 0.019, P = 0.058) but when using the actual WCC, age was not related to parasitaemia (r = −0.001, P = 0.954) (Figure 1).
This study shows that, in children under 5 years of age with acute uncomplicated malaria living in sub-Saharan African countries with moderate to intense malaria transmission, the parasite biomass needed for malaria to be clinically manifest is not significantly different across the age range 1–4 years when calculated using the patient’s actual WCC. Overall, parasitaemia on presentation did not change significantly with age. One must not forget although that these, like any other trials, had set limits of parasitaemia to qualify for inclusion. Here, the upper limit of parasitaemia was 200 000/μl in all studies but the lowest limit varied from 500/μl (Dorsey et al. 2007) to 1000/μl (Adjuik et al. 2002; Ndiaye et al. 2009; Sirima et al. 2009) to 2000/μl (Mårtensson et al. 2005; Karema et al. 2006; Djimdéet al. 2008). Therefore, it is not possible to categorically set the pyrogenic threshold of P. falciparum parasitaemia in African children under 5 years of age from this dataset.
The admission parasitaemia as customarily estimated in the published series assuming 8000 WBCs/μl increased with age, which would be interpreted as a sign of mounting immunity with older children being able to tolerate higher parasitaemia. However, this appears to be an artefact, as it does no longer hold true when each subject’s own WCC are used to calculate parasitaemia – no significant association between age and parasitaemia was observed. This discrepancy occurs because although the two methods agree overall for all under-fives, WCC are higher in younger children.
These changes in WCC with age cause the standard approximation method to significantly underestimate parasitaemia in infants. The equipoise between the two methods of estimation in our study was at 2 years of age; thereafter, parasitaemia became (non-statistically significant) higher with standardised than with actual WCC. It should also be noted that here the average WCC were approximately 8800 and >8000 (the figure assumed in the standard method) at all ages.
A cross-sectional study in 240 Nigerian children aged 1–8 years found that the assumed parasite counts with the standard method overestimated parasitaemia but did not provide details as to possible age-related differences (Jeremiah & Uko 2007). Also, children in that study were asymptomatic, and parasite counts were considerably lower than those recorded in the symptomatic children of our study.
Most laboratories in malaria-endemic rural areas would not have the capacity to measure WCC; our data indicate that in routine practice using standard WCC counts seems altogether acceptable. Moreover, with WCC declining with age, the standardised 8000 WCC generates adequate estimates in older children and adults.
However, clinical trials of uncomplicated falciparum malaria should express parasitaemia calculated on the basis of the patients’ own baseline WCC. The differences observed between the two methods, although overall minimal, are important especially in the younger children. Using actual WCC will increase precision in estimating parasitaemia and improve our understanding of malaria and response to treatment, including haematological changes in acute and convalescent malaria. A larger sample size is desirable to confirm that this observation can be widely generalised.