Diagnostic sensitivity

Shazia Ruybal-Pesántez

Last updated: 30 Mar 2023

Source: vignettes/data_diagnostics.Rmd

This report aims to:

  • Explore the sensitivity to detect malaria by microscopy, rapid diagnostic test (RDT) and polymerase chain reaction (PCR)
  • Gain insights to help with future model development.

Prevalence by microscopy vs. PCR

Whittaker et al. (2021)

Data: PCR_micro_full_whittaker2021

The data come from a systematic review by Whittaker et al. (2021) looking into global patterns of submicroscopic P. falciparum infection. This review collates data from 166 studies containing 551 distinct cross-sectional surveys in which surveyed individuals had malaria infection assessed by both PCR and microscopy from a given location. Of these surveys, 164 were done in a specific age-group (0–5 years, 6–15 years, and >15 years) and 387 were from cross-sectional surveys done in populations that spanned more than one age-group.

Below we can take a look at the variables included in the dataset:

head(PCR_micro_full_whittaker2021)
## # A tibble: 6 × 19
##   Name      Year Globa…¹ Country Admin_1 Hist_…² Curr_…³ PCR_N…⁴ PCR_N…⁵ PCR_P…⁶
##   <chr>    <dbl> <chr>   <chr>   <chr>     <dbl>   <dbl>   <dbl>   <dbl>   <dbl>
## 1 Proietti  2011 East A… Uganda  Apac      0.746   0.625     241     133 5.52e-1
## 2 Souza     2012 South … Brazil  NA       NA      NA          74       1 1.35e-2
## 3 Souza     2012 South … Brazil  NA       NA      NA          74       2 1.35e-2
## 4 Souza     2012 South … Brazil  NA       NA      NA         134       1 1.35e-2
## 5 Atkinson  2012 Asia&O… Solomo… NA       NA      NA        1843       1 5.43e-4
## 6 Pegha M…  2016 West A… Gabon   Ogooue…   0.341   0.189     277     161 5.81e-1
## # … with 9 more variables: Microscopy_N_Tested <dbl>,
## #   Microscopy_N_Positive <dbl>, Micro_Prev <dbl>, PCR_Method <chr>,
## #   PCR_Method_Raw <chr>, Microscopy_Fields <dbl>, Microscopy_Leucocytes <dbl>,
## #   Sampling_Season <chr>, Sampling_Season_Raw <chr>, and abbreviated variable
## #   names ¹​Global_Region, ²​Hist_Trans, ³​Curr_Trans, ⁴​PCR_N_Tested,
## #   ⁵​PCR_N_Positive, ⁶​PCR_Prev

The authors found that microscopy detected 44.9% (95% CI 42.0–47.8) of all PCR-detectable infections. This varied by setting, with more submicroscopic infections (60-70%) in areas with lower PCR prevalence such as South America, but <20% submicroscopic infections in areas with the highest PCR prevalence such as West Africa. The authors found that both historical and current transmission levels are important determinants of the submicroscopic reservoir size.

PCR_micro_full_whittaker2021 %>% 
  ggplot(aes(x = PCR_Prev, y = Micro_Prev, group = Global_Region, color = Global_Region)) +
    geom_point() +
    geom_smooth(method = "loess", formula = "y ~ x", se = F) +
    scale_x_continuous(labels = scales::percent_format(accuracy = 1)) +
    scale_y_continuous(labels = scales::percent_format(accuracy = 1)) +
    scale_color_manual(values = c("hotpink3", "goldenrod3", "lightblue3", "olivedrab3")) +
    labs(x = "PCR prevalence (%)",
         y = "Microscopy prevalence (%)") +
    theme_minimal()

Data: PCR_micro_age_whittaker2021

The authors also aggregated the surveys carried out in specific age groups into three categories: young children (0–5 years), n=49 sureveys, older children (6–15 years) n=62 surveys, and adults (>15 years) n=53 surveys. This age-disaggregated dataset is also available, below we can take a look at the variables included in this dataset:

head(PCR_micro_age_whittaker2021)
## # A tibble: 6 × 21
##   Name      Year Globa…¹ Country Admin_1 Hist_…² Curr_…³ PCR_N…⁴ PCR_N…⁵ PCR_P…⁶
##   <chr>    <dbl> <chr>   <chr>   <chr>     <dbl>   <dbl>   <dbl>   <dbl>   <dbl>
## 1 Proietti  2011 East A… Uganda  Apac      0.746   0.625      57      38   0.667
## 2 Proietti  2011 East A… Uganda  Apac      0.746   0.625      62      43   0.694
## 3 Proietti  2011 East A… Uganda  Apac      0.746   0.625     122      52   0.426
## 4 Atelu     2016 West A… Ghana   Upper …   0.653   0.493      29       4   0.138
## 5 Atelu     2016 West A… Ghana   Upper …   0.653   0.493      38       8   0.211
## 6 Atelu     2016 West A… Ghana   Upper …   0.653   0.493     142      17   0.120
## # … with 11 more variables: Microscopy_N_Tested <dbl>,
## #   Microscopy_N_Positive <dbl>, Micro_Prev <dbl>, Age_Group <fct>,
## #   Age_Group_Raw <chr>, PCR_Method <chr>, PCR_Method_Raw <chr>,
## #   Microscopy_Fields <dbl>, Microscopy_Leucocytes <dbl>,
## #   Sampling_Season <chr>, Sampling_Season_Raw <chr>, and abbreviated variable
## #   names ¹​Global_Region, ²​Hist_Trans, ³​Curr_Trans, ⁴​PCR_N_Tested,
## #   ⁵​PCR_N_Positive, ⁶​PCR_Prev

The authors found that a greater proportion of submicroscopic infections was observed in adults compared to young and older children. This was less pronounced in areas with higher transmission.

PCR_micro_age_whittaker2021 %>% 
  ggplot(aes(x = PCR_Prev, y = Micro_Prev, group = Age_Group, color = Age_Group)) +
    geom_point() +
    geom_smooth(method = "loess", formula = "y ~ x", se = F) +
    scale_x_continuous(labels = scales::percent_format(accuracy = 1)) +
    scale_y_continuous(labels = scales::percent_format(accuracy = 1)) +
    labs(x = "PCR prevalence (%)",
         y = "Microscopy prevalence (%)") +
    theme_minimal()