A systematic review of sub-microscopic Plasmodium vivax infection

Abstract

Background: An accurate estimate of Plasmodium vivax prevalence is essential for the successful implementation of malaria control and elimination programmes. Prevalence estimates both inform control strategies and are used in their evaluation. Light microscopy is the main method for detecting Plasmodium parasitaemia in the peripheral blood, but compared to molecular diagnostics, such as polymerase chain reaction (PCR), has limited sensitivity.

Methods: A systematic review and meta-analysis was conducted to assess the effect of detection method on the prevalence of P. vivax and to quantify the extent to which P. vivax infections are undetected by microscopy. Embase, Medline and the Cochrane Database were searched for studies reporting prevalence by PCR and by microscopy and that contained all of the following key words: vivax, PCR, and malaria. Prevalence estimates and study meta-data were extracted systematically from each publication. Combined microscopy:PCR prevalence ratios were estimated by random effects meta-analysis. Sensitivity and specificity of microscopy were calculated using PCR as the gold standard.

Results: Of 874 studies reviewed, 40 met the criteria for inclusion contributing 54 prevalence pairs. The prevalence of P. vivax infection measured by PCR was consistently higher than the prevalence measured by microscopy with sub-patent parasitaemia. The mean prevalence of infection detected by microscopy was 67 % (95 % CI 59-73 %) lower than the prevalence detected by PCR. The detection of sub-patent parasitaemia did not vary according to the microscopy method (thick or, thick and thin smears), the PCR prevalence (as a measure of the true P. vivax prevalence), the type of blood used or DNA extraction method.

Conclusions: Quantifying P. vivax parasitaemia by PCR rather than microscopy consistently increased prevalence estimates by a factor of 2.3. Whilst the sensitivity of microscopy can be improved by better methods, molecular methods have potential to be scaled up to improve the detection of P. vivax transmission reservoirs.

Fig. 1

PRISM flow diagram. Study selection (PRISM flow diagram)

Fig. 2

PCR P. vivax prevalence versus microscopy P. vivax prevalence, monoinfections and mixed infections. Scatter plot of the prevalence of P. vivax infection as determined by PCR versus prevalence of P. vivax as detected by microscopy. Prevalence pairs including only monoinfections (infections with only P. vivax) are shown in open circles and the prevalence pairs including all infections (P. vivax monoinfections and P. vivax infections where another species is also detected) are shown in filled circles. The line shows the expected association if both techniques were equally sensitive

Fig. 3

Sub-microscopic P. vivax prevalence by region. Box plots showing the median and the IQR of sub-microscopic P. vivax prevalence by region (Asia Thailand, Sri Lanka, Lao PDR, Vietnam, Malaysia and Cambodia; Asia Pacific Indonesia and Papua New Guinea; South America Brazil, Venezuela and Peru)

Fig. 4

Ratios of the prevalence of P. vivax determined by microscopy to that determined by PCR. Forest plot displaying ratios (represented as closed diamonds) and 95 % confidence intervals (horizontal lines) of the prevalence of P. vivax detected by microscopy compared to that detected by PCR in 36 pair observations. The open diamond and broken vertical line represents the combined estimate from a random effects (D + L overall) meta-analysis, the combined estimate from a fixed effect (I − V overall) is also shown. The unbroken vertical line is at null value (1)

Fig. 5

Scatter plot of the microscopy: PCR prevalence ratio versus the underlying PCR prevalence