The general distributions of sex, age, and relative parasite proportions did not differ greatly between case-patients (n = 94) and controls (n = 260) for all of Ontario and the GTA subset (Table 1). The spatial distribution of the case-patients and controls was consistent with Ontario’s settlement patterns, with most of the observations located in southern Ontario and the GTA (Figure 1). The downtown core of Toronto had a large concentration of controls, whereas more case-patients lived in suburban areas.

Significant clustering of case-patients occurred in suburban Toronto during the summer months (Figure 1). A significant space–time cluster was identified during May 15, 2008–November 6, 2008, in the region northwest of Toronto near Brampton (p<0.01; relative risk 3.54). This area is near the main international airport in the GTA. In this area during the summer months, tested persons were 3.5× more likely to receive a diagnosis of malaria than were persons outside of this area and period. No significant clustering of individual parasite species (P. falciparum or P. vivax) was identified.

Cases were not significantly more likely to occur in the summer months (June–September) than in the nonsummer months for the GTA and for all of Ontario (χ²; p = 0.14 and p = 0.11, respectively). This result is consistent with that of the space–time cluster analysis, which found that the significant space–time cluster extended through summer into autumn. Visual observation of monthly incidence, however, did not indicate any strong or clear trends, possibly because of the influence of low observation numbers, and incidence and seasonality in malaria-endemic countries from which the disease is imported (Figure 2).

Population density was highly variable in the study area, with more densely populated areas located near the downtown core of the GTA. Median household income, in contrast, exhibited negligible clustering except for low-income suburbs east and west of the downtown core. The proportion of residents who were immigrants from malaria-endemic regions was highest in the suburban areas to the east and west of Toronto (Figure 3). More than 94% of case-patients for whom travel history was recorded indicated recent travel to Africa or Asia (Figure 4), with travel to Africa reported most frequently. These results are consistent with known spatial ranges of malaria-endemic areas (30).

Case-patients were significantly more likely than controls to be male and live in low-income neighborhoods with a higher proportion of residents who immigrated from malaria-endemic regions (Table 2). The proportion of residents who were immigrants from malaria-endemic countries was, for example, >2× as high in neighborhoods with case-patients than in neighborhoods where controls were identified. The average income of case-patient neighborhoods was ≈$28,000, compared with ≈$31,000 in control neighborhoods (Student t test, p = 0.04). Case-patients with a travel history (n = 47) were more likely to report having traveled to Africa and to be living in a neighborhood with a higher proportion of residents who are immigrants from malaria-endemic countries than controls with a travel history (Table 3).

The regression results are consistent with univariate analyses: case-patients were more likely to be male (odds ratio [OR] 2.24, 95% CI 1.24–4.05) and live in neighborhoods with high levels of immigration from malaria-endemic countries (OR 1.09, 95% CI 1.06–1.12) (Table 4). Population density was a weak predictor, but was retained in the model to account for potential confounding with proportion of residents who are immigrants from malaria-endemic regions of Africa. This variable had a wide range (0%–58%) and showed the strongest effect on case status: for every 1% increase in the proportion of residents who are immigrants from malaria-endemic regions, the odds of being a case-patient increased by 7%. The odds of being a case-patient were >17× higher in neighborhoods in the top quartile of proportion of residents who are immigrants from malaria-endemic areas than in the neighborhoods in the lowest quartile. There were slight negative effects from median household income and population density. For every 1 unit increase in either, the odds of being a case-patient decreased by 1%.

The pseudo R² of the model was 0.19, with a predictive accuracy of 70% correctly classified by using a cutoff of 0.25 (67% sensitivity, 71% specificity). The model fit the data well (Hosmer-Lemeshow test, p = 0.76), and leverage values did not indicate any outliers that required removal. Several high positive residuals were found, because of the low sensitivity of the model that used a default cutoff (0.5). This low sensitivity reflects the limited number of variables available for analysis. The model was developed, however, for explanatory rather than predictive purposes; low sensitivity reflects the small number of covariates and use of neighborhood-level variables.

The distribution of parasite species infection in case-patients reflects the patterning of global malaria; P. vivax cases are associated with travel to or immigration from Asia, and P. falciparum cases are associated with travel to Africa (Table 3). Globally, a significant spatial patterning of parasite species exists, with P. falciparum found predominantly in Africa and P. vivax found predominantly in Asia (30). Case-patients infected with P. falciparum, for example, were significantly more likely to report recent travel to Africa than case-patients infected with P. vivax (Fisher exact test, p<0.01). Conversely, case-patients infected with P. vivax were more likely to report travel to malaria-endemic regions of Asia (Fisher exact test, p<0.01). Case-patients with P. vivax lived in neighborhoods with a significantly higher proportion of residents who are immigrants from malaria-endemic regions (Mann-Whitney, p<0.01). In contrast, however, P. falciparum case-patients did not necessarily live in neighborhoods with a high proportion of residents who are immigrants from malaria-endemic regions of Africa (Mann-Whitney, p = 0.33). This finding may be due to the much lower overall level of immigration from Africa.

Malaria case-patients in Ontario were found predominantly in the GTA, with more case-patients in suburban areas outside the city center. Malaria case-patients were more likely to live in neighborhoods with a high proportion of residents who emigrated from malaria-endemic areas. Case-patients were more likely than controls to report travel to areas with endemic malaria, and there was concordance between the parasite species and the region of travel. The association between parasite species and geography held when neighborhood immigration was examined as well—cases of P. vivax malaria corresponded with immigration from malaria-endemic areas of Asia, and cases of P. falciparum malaria were found in areas with immigration from malaria-endemic areas of Africa.

Case-patients were more likely to be male than female. This finding is consistent with the literature on sex and travel-associated diseases, which finds that men are more likely than women to have vector-borne diseases, including malaria (31). Whether this finding is due to a biologic predisposition or to behavioral differences associated with increased transmission risk is not known (31). Whether potential differences in patterns of health care use between men and women could have affected the results in this study also is not known. Further research into how men and women seek pretravel and posttravel medical care in Ontario would improve the understanding of this result.

When assessing imported malaria cases, underreporting is a substantial issue. In Ontario, underreporting of malaria is likely to range from 10% to 40% (5,32). In Ontario, 226 malaria cases were documented by laboratories in 1998 (32). However, in that same year, only 160 cases in Ontario were reported to Health Canada (32). No recent published or quantified estimates of actual incidence are available, and the true level of current underreporting of malaria in Ontario and Canada more generally remains unclear. The data used in the study are strengthened by the inclusion of negative and positive malaria test results. The use of controls enables the factors that correlate with malaria cases to be better understood, unbiased by underlying patterns of malaria testing.

The odds of becoming infected and importing malaria to Canada were >17× higher for residents of neighborhoods with high immigration from malaria-endemic areas than those with low immigration from malaria-endemic countries. These results, combined with the geographic findings of the cluster analysis, are supported by census data on immigration. Brampton (the location of the significant space–time cluster of malaria cases), Markham, and Ajax all showed a significant increase in the proportion of residents born outside Canada in the 2006 census (25). As immigration to the neighborhoods in the suburbs of the GTA grows, more imported malaria cases could result. These results have implications for potential preventative measures that could be taken before persons travel abroad. The use of cluster detection methods has been proposed in other areas to target prevention methods (33,34). Targeted prevention could focus on hospitals and clinics found in the cluster area and in regions with a high proportion of residents who are immigrants from malaria-endemic areas, with the goal of implementing targeted travel medicine screenings and education regarding malaria prophylaxis.

Predeparture travel medicine services are not covered under Ontario’s provincial insurance plan (35). However, the use of travel clinics is not associated with income in Canada (36). As a high-risk group, VFRs are characterized by differing preventive care choices and lower use of pretravel medical services (37). In Canada, 1 study found that VFRs consulted family practitioners more often than travel clinics and that the family practitioners were more likely to prescribe inappropriate chemoprophylaxis (38). Another reason for the increased risk for imported malaria among VFRs could be differing perceptions of the risk for malaria, leading to different behavior in malaria-endemic areas (16,21,39). The recent growth of immigrants in Brampton and related Toronto suburbs may merit targeted programming by physicians and public health agencies, including translation of public health prevention materials (25).

Imported malaria in Canada is of public health significance given current, and likely substantially underestimated, cases of imported disease in Canada. Our characterization of imported malaria cases in Ontario highlights increased risk for infection in immigrant and low-income neighborhoods, with associations between immigration from countries where malaria transmission and particular pathogen types are endemic. These results suggest a pattern of inequality in tropical health outcomes in Ontario with implications for prevention and control of current incidence.

Imported malaria cases represent the potential for geographic emergence of a range of travel-associated diseases. The current landscape of emerging infectious diseases demands that we improve our understanding of the mechanisms by which diseases are imported and the characteristics of neighborhoods and populations at highest risk. The results of this study point to potential individual- and neighborhood-level risk factors that might be relevant to emerging infectious diseases more generally. As new infectious diseases emerge, identifying the processes of emergence and the areas and persons at greatest risk is critical. An analysis of the local patterns of imported malaria in Ontario can help the public health community better understand the ways in which global travel and immigration can affect the spatial distribution of other travel-related diseases.