The association between gold mining and malaria in Guyana: a statistical inference and time-series analysis

Abstract

Background: Guyana reported a significant rise in malaria between 2008 and 2014. As there was no evidence of impairment of national malaria control strategies, public health authorities attributed the surge to a temporal increase in gold mining activity in forested regions. However, systematic analysis of this association is lacking because of the difficulties associated with collecting reliable data for both malaria and mining. We aimed to investigate the association between the international gold price and Plasmodium falciparum malaria transmission in Guyana between 2007 and 2019. We also aimed to evaluate the association between P falciparum cases and the El Niño-Southern Oscillation pattern, which has previously been suggested as a major driver of malaria.

Methods: We used national malaria surveillance data from Guyana to estimate the correlation over time between the international gold price and reported P falciparum infections in individuals who were likely to be involved in mining activities (ie, men and boys aged between 15 and 50 years who were living in mining regions) for each month between 2007 and 2019. We compared the estimates with those obtained from individuals who were unlikely to be directly involved in mining activities (ie, women, children aged 12 years and younger, and adults aged over 70 years) and estimates obtained from individuals living in non-mining regions. We also evaluated the correlation between P falciparum infections and the El Niño-Southern Oscillation pattern in the same subpopulations and time period. Lastly, we evaluated the performance of a statistical model formulated to estimate P falciparum infections in real time using the international gold price as the predictor variable.

Findings: The proportion of P falciparum malaria cases in temporary residents, which was used as a proxy for circulating individuals involved in gold mining, was highest during the years of peak gold price (ie, between 2008 and 2014). Cases of malaria in all demographic groups showed a strong positive correlation with the gold price, but only in regions with mining camps (0·88 [95% CI 0·84-0·89] for boys and men aged between 15 and 50 years and 0·80 [0·73-0·85] for the aggregated population of women, children aged 12 years and younger, and adults older than 70 years). The highest correlation occurred earlier in men and boys aged between 15 and 50 years, the demographic most likely to be miners, suggesting that transmission in mining camps is followed by infections in the community. On the basis of these findings, we were able to reliably forecast P falciparum malaria trends using only the gold price as the predictor variable. A 1% increase in gold price was associated with a 2·13% increase in P falciparum infections after 1 month in the mining populations, and with a 1·63% increase after 2 months in the non-mining populations. Lastly, La Niña climatic events showed an additional, smaller positive correlation with malaria transmission.

Interpretation: Our analysis provides evidence that the P falciparum malaria surge observed in Guyana between 2008 and 2014 was likely to have been driven mainly by an increase in gold mining, while climate factors might have contributed synergistically. We propose that the international gold price over time is a useful indicator of malaria trends. We conclude that the feasibility of malaria elimination in Guyana, and in other areas in the Amazon where malaria and gold mining overlap, should be evaluated against the challenges posed by rapidly rising gold prices.

Funding: Ramón Areces Foundation, National Institutes of Health, and National Institute of General Medical Sciences.

Figure 1

Distribution of Plasmodium falciparum cases and gold price in Guyana, 2007–19 (A) Time series of P falciparum malaria cases in Guyana by month (grey) and detrended gold price per month (yellow). (B) Distribution of P falciparum cases between 2007 and 2019 by age group and sex in mining regions (regions 1, 7, 8, and 10). (C) Monthly proportion of malaria cases (represented by red points) in non-resident men and boys in gold mining regions aged between 15 and 50 years by gold price quartile rank. Boxplots represent mean and range between the 2·5th and 97·5th percentiles.

Figure 2

Stratified correlations between Plasmodium falciparum malaria and gold price time series in Guyana, 2007–19 (A) Number of P falciparum cases in mining regions (regions 1, 7, 8, and 10) disaggregated by mining populations (shown in green) and non-mining populations (shown in brown). (B) Case time series in non-mining regions (regions 2, 5, 6, and 9) disaggregated by mining populations (shown in green) and non-mining populations (shown in brown). (C) Scatter plot showing monthly cases in mining populations (green dots) with a 1-month shift and non-mining populations (red dots) in mining regions versus monthly adjusted gold price with a 2-months shift.

Figure 3

Model forecasting Plasmodium falciparum infections in Guyana 2008 and 2019 using the gold price as the predictor variable Prediction intervals of expected number of P falciparum malaria cases per month in Guyana, estimated by fitting a generalised linear model that used only gold price as the predictor. Ribbons represent 25th–75th percentiles (shown in dark green) and 2·5th and 97·5th percentiles (shown in light green). Lines represent true reported cases. Red area shows the period (2014) when the bed net distribution was strengthened (appendix p 8).