Skip to main content

Table 6 Univariable and multivariable risk factors for PCR-confirmed recrudescent failures at day 28

From: The effect of dosing strategies on the therapeutic efficacy of artesunate-amodiaquine for uncomplicated malaria: a meta-analysis of individual patient data

   Univariable analysis Multivariable analysis b Population attributable risk c
(N = 9,058)
Variable Total n [ n ] a Crude HR [95% CI] p -Value Adjusted HR [95% CI] P -Value Freq. PAR
Age (y) 9,095 (265) 0.92 [0.89-0.96] <0.001 - - - -
Amodiaquine dose (5 mg/kg) 9,095 (265) 0.94 [0.84-1.04] 0.220 0.94 [0.84-1.05] 0.280 - -
Enrolment clinical variables        
Parasitemia (per 10-fold) 9,095 (265) 1.46 [1.16-1.84] 0.001 1.39 [1.1-1.74] 0.005 10.4% 3.7%
Parasitemia >100,000 parasites/μl 9,095 (265) 1.41 [0.98-2.05] 0.066 - - - -
Fever (temp > 37.5°C) 8,755 (252) 1.05 [0.78-1.41] 0.760 - - - -
Hemoglobin (g/dl) 6,895 (237) 0.93 [0.87-1.00] 0.055 - - - -
Anemia (Hb < 10 g/dl) 6,895 (237) 1.37 [1.04-1.81] 0.028 - - - -
Gametocytes presence 4,790 (99) 1.04 [0.54-1.98] 0.910 - - - -
Underweight (WAZ < −2)d 6,260 (616) 0.87 [0.61-1.26] 0.470 - - - -
Gender        
Female (reference) 4,231 (126) 1 - - - - -
Male 4,702 (124) 0.91 [0.71-1.16] 0.450 - - - -
Age category        
≥12 y (reference) 1,135 (12) 1 - - - - -
<1 y 782 (31) 3.15 [1.46-6.78] 0.004 3.93 [1.76-8.79] 0.001 8.6% 20.9%
1 to <5 y 5,645 (199) 3.62 [1.83-7.18] <0.001 4.47 [2.18-9.19] <0.001 62.3% 69.2%
5 to <12 y 1,533 (23) 1.90 [0.91-3.98] 0.088 2.03 [0.96-4.28] 0.064 16.9% 15.1%
Drug formulation        
FDC (reference) 4,135 (70) 1 - - - - -
Co-blistered NFDC 1,256 (21) 1.02 [0.52-2.00] 0.950 1.38 [0.75-2.57] 0.300 13.9% 5.1%
Loose NFDC-25 1,291(70) 3.62 [1.79-7.30] <0.001 3.51 [2.02-6.12] <0.001 14.3% 25.8%
Loose NFDC-30e        
In Rukara/Kailahun/Kisumuf 461 (59) 8.41 [3.24-21.84] <0.001 7.75 [4.07-14.76] <0.001 5.1% 26.3%
Rest of the sites 1,952 (45) 1.34 [0.77-2.34] 0.300 1.47 [0.91-2.38] 0.110 21.1% 8.3%
Treatment supervision        
Fully supervised (reference) 8,530 (245) 1 - - - - -
Partially supervised 565 (20) 1.37 [0.45-4.17] 0.580 - - - -
Parasite clearance        
Day3 Parasitemia 8,788 (252) 2.17 [0.88-5.35] 0.092 - - - -
Region        
Africa (reference)g 8,624 (245) 1 - - - - -
Asia 434 (20) 1.27 [1.83-3.55] 0.700 7.39 [3.45-15.86] <0.001 4.8% 21.6%
S. Americah 37 (0) - - - -   
  1. aNumber of patients [n] for each variable/levels of factor with number of PCR-confirmed recrudescence [n] by day 28.
  2. bVariance of the random effect = 0.22. Adding hemoglobin (AHR = 0.94 [95% CI: 0.88-1.02]; P = 0.064), day 3 parasite positivity (AHR = 2.04 [95% CI:0.83-5.00]; P = 0.107) to a model containing age, parasitemia, AQ dose, region and formulation led to a non-significant likelihood ratio test, and hence those variables were not kept for multivariable analysis. Although anemia (AHR = 1.35 [95% CI: 1.02-1.78]; P = .034) was found to be significant, a large proportion of patients had missing values. Hence, random imputation was performed for anemia, hemoglobin and gametocytemia, which showed that they were not significant in the presence of other variables (Additional file 6: Text S6, Figure 1). To examine the robustness of the parameter estimates, a sensitivity analysis was carried out by removing one study site at a time which showed that the overall coefficient of variation of parameter estimates in the multivariable model was small (all CV <10%) (Additional file 6: Text S6, Table 3).
  3. cOverall PAR for model = 92.6%.
  4. dUnderweight for age defined only in children < 5 years.
  5. eCompared to FDC, patients treated with loose NFDC-30 were at higher risk of recrudescence (AHR = 2.89 [95% CI: 1.49-5.59]; P = 0.002) when all the sites were combined.
  6. Pairwise comparisons.
  7. Co-blistered NFDC v loose NFDC-25 (AHR = 2.50 [95% CI: 1.18-5.44]; P = 0.016).
  8. Co-blistered NFDC v loose NFDC-30 in Rukara/Kailahun/Kisumu (AHR = 5.61 [95% CI: 2.48-12.69]; P < 0.001).
  9. Co-blistered NFDC v loose NFDC-30 in rest of the sites (AHR = 1.07 [95% CI: 0.54-2.10]; P = 0.850).
  10. Loose NFDC-25 v loose NFDC-30 in Rukara/Kailahun/Kisumu (AHR = 2.21 [95% CI: 1.03-4.71]; P = 0.041).
  11. Loose NFDC-25 v loose NFDC-30 in rest of the sites (AHR = 0.42 [95% CI: 0.23-0.77]; P = 0.005).
  12. fThe test for proportional hazards did not hold true for this category. The overall assumption of proportional hazards held true globally and individually for each of the covariates when these three sites were excluded from the model. The coefficients of the remaining model parameters were similar with and without these three sites kept in the model. The assumption of proportionality was tested for each of the studies separately with at least five failures (Additional file 6: Text S6, Table 3) and found to be satisfactory.
  13. gWithin Africa, there were no differences between East and West Africa: AHR = 1.14 [0.62-2.15]; P = 0.690.
  14. hHazards ratio could not be estimated as there were no PCR-confirmed failures in South America.