Fig 1.
Malaria co-infection in pre-treatment samples (n = 504).
(A) Co-infection prevalence of asexual parasites. Double infection with Pf denotes the proportion of samples PCR positive for Pf and at least one other species. Triple infection with Pf denotes the proportion of samples PCR positive for Pf and at least two other species. Other bars are similarly defined. 95% confidence intervals were calculated using Wilson’s binomial method. (B) Asexual parasite density in co-infected samples. (C) Venn diagram of asexual parasite co-infection. (D) Co-infection prevalence of gametocytes. (E) Gametocyte density in co-infected samples. (F) Venn diagram of gametocyte co-infection.
Fig 2.
Excess asexual parasite co-infection in pre-treatment samples.
(A) In the pre-treatment samples, Pf asexual prevalence was 23.8% (120/504) and Pv asexual prevalence was 47.4% (239/504). If these parasites were randomly distributed, we would expect co-infection prevalence of 0.238 * 0.474 = 11.3% (red and yellow striped region). (B) Co-infection prevalence of 14.7% (74/504) was observed, in excess of what is expected by random mixing. (C) Expected versus observed co-infection prevalence for the six pairwise combinations of asexual parasites. The dashed line represents the scenario where observed co-infection prevalence equals expected prevalence. Mono-coloured points denote observed prevalence rates of the four species. The multi-coloured data points fall above this line. The solid line denotes a regression model fitted through these points, with 95% confidence intervals shown in grey.
Fig 3.
Co-infection in pre-treatment samples stratified by village.
(A) For the six pairwise combinations of two malaria species, the observed and expected co-infection asexual prevalence is plotted for each of the five villages. The dashed line represents the scenario where observed co-infection prevalence equals expected prevalence. The multi-coloured data points tend to fall above this line. The solid line denotes a regression model fitted through these points, with 95% confidence intervals shown in grey. (B) Observed and expected triple infection. If the prevalence of Pm is XPm, then the expected prevalence of Pf, Pv and Pm co-infection is XPf * XPv * XPm. The multi-coloured points fall above the dashed line indicating greater observed than expected prevalence. The solid line denotes a regression model fitted through these points. (C) Gametocyte co-infection. (D) Gametocyte triple infection.
Table 1.
Estimated risk of co-infection in pre-treatment samples.
*The excess risk of triple infection accounting for the known pairwise clustering due to heterogeneity (e.g. 1.26 = 2.42/1.382).
Fig 4.
Effect of co-infection on parasite density in pre-treatment samples.
The y-axis denotes the measured parasite density and the x-axis denotes the confounding effect of co-infection. Each square denotes the fold change in parasite density due to co-infection. For example, for Pm asexual parasites, co-infection with Pf asexual parasites leads to a 6.57 (2.9, 14.8) fold increase in Pm asexual parasite density. Grey squares denote interactions where it was not possible to estimate an effect. Otherwise, all estimated effects are presented regardless of statistical significance. Orange squares denote significant associations with P values < 0.05. Red squares denote significant associations with P values < 0.05 after the Benjamini-Hochberg adjustment for multiple hypothesis testing.
Fig 5.
Longitudinal analysis of P. malariae and the effect of co-infection.
The prevalence of (A) P. malariae asexual parasites, and (B) P. malariae gametocytes. The prevalence of all P. malariae infections is shown in green, and the prevalence of P. malariae and P. falciparum co-infection is shown in red. The grey shaded region denotes the period of treatment. 5561 samples were included over the entire time period. 123 samples were positive for P. malariae asexual parasites, and 61 of these were co-infected with P. falciparum asexual parasites. 43 samples were positive for P. malariae gametocytes, and 18 of these were co-infected with P. falciparum gametocytes. (C) The observed proportion of P. malariae asexual infections that are co-infected with P. falciparum is shown in red. The expected proportion of P. malariae infections co-infected with P. falciparum under an assumption of random mixing is shown in black.