The databases used for the present analysis include 22 years of records from a local malaria reference center in Montreal, Canada (the McGill University Centre for Tropical Diseases [TDC]) and up to 13 years of quality assurance and notifiable disease surveillance databases of the provincial and federal governments of Canada, France, India, Switzerland, the United States, and the United Kingdom. TDC is a clinical and laboratory facility that provides care to 800 to 1,100 new patients per year (approximately 60% primary cases and 40% consult cases), drawn mainly from the Montreal region. The TDC database from 1981 to 2002 has allowed previous detailed reviews of changing patterns of malaria in its patient populations (8–10). Malaria-relevant data captured include category of traveler (tourist, immigrant, refugee, expatriate, missionary, and volunteer), countries visited, and malaria species. A diagnosis of malaria is made if parasites are noted on a blood smear (thin, thick, or buffy coat) or if, in the last 5 years, the patient had a positive result on a malaria antigen-capture test (e.g., Macromed [Nova Century Scientific, Inc., Burlington, Ontario, Canada], ICT Malaria P.f. [ICT Diagnostics, Brookvale, New South Wales, Australia], or OptiMAL [Flow Inc., Portland, OR]). While active surveillance studies during this period included polymerase chain reaction (PCR) as a screening tool, PCR-positive cases were not included in any of the passive surveillance statistics unless they were also independently confirmed by either malaria antigen capture or smear.

Provincial reportable disease databases have included, in the past 10 years, patient characteristics such as age, sex, and malaria species, but not the likely country of acquisition. Because 90% of all malaria cases in Canada were reported by the Provinces of British Columbia (Monica Naus [British Columbia Centre for Disease Control], pers. comm.), Ontario (Lorraine Schiedel [Ontario Ministry of Health and Long-Term Care], pers. comm.), and Québec (Colette Colin [Ministère de la santé et des services sociaux, Québec], pers. comm.), the present analysis focuses on their data, primarily for 1990–2002 (11). Quality assurance data for the province of Québec (1994–2002) were provided by TDC and the Laboratoire de Santé Publique du Québec. The federal government’s notifiable disease database from 1990 to 2002 is a compilation of selected information from individual provincial databases and includes patient age and sex for each report but no malaria species or country of acquisition (Carole Scott [Division of Disease Surveillance, Health Canada], pers. comm.).

International malaria surveillance data (1990–2002) were acquired from the World Health Organization (WHO) Regional Office for South East Asia (Rakish Mani Rastogi, pers. comm.), the WHO Regional Office for Europe (12), and the United States (13–24). Malaria rates for all countries were based on population data of the U.S. Census Bureau (25).

Trends in Canadian immigration and refugee data for the years 1990–2002 were provided by Citizen and Immigration, Canada (Karen Tremblett [Medical Services Branch, Citizen and Immigration Canada], pers. comm.), data on language by Statistics Canada (26), and travel patterns of Canadians to the tropics by the World Tourism Organization, Madrid (27).

Malaria importations into Canada can occur by either immigration or travel, and changing malaria attack rates in the countries of exposure are likely to influence the incidence of imported disease. Changes in Canadian immigration and refugee patterns from 1990 to 2002 are notable for a threefold increase in annual immigrant numbers from the Indian subcontinent and relatively stable numbers from sub-Saharan Africa. Neither combined nor separate provincial immigration and refugee patterns explain the important swings in annual Canadian malaria rates.

While the geographic origins of immigrants and refugees do not immediately explain the epidemic changes in P. vivax malaria seen in the mid-1990s, their nonrandom aggregation in certain provinces allows additional insights. African immigrants and refugees have settled all across Canada in every province in numbers that paralleled the province’s population. Immigrants and refugees from the Indian subcontinent did not: 84% settled in Ontario and British Columbia, the provinces with the most pronounced P. vivax epidemics. Canadian travelers to malaria-endemic areas have gradually but steadily increased during the past 15 years, most notably with a threefold increase to Southeast Asia and Central and South America, a twofold increase to the important malarial region of south Asia, and a smaller increase to Africa. Travel patterns did not offer an explanation for either the P. vivax epidemics in British Columbia and Ontario in the late 1990s or the P. falciparum epidemic in Québec from 2000 to 2001. The World Tourism Organization data do not break down Canadian travel by traveler’s province of origin; however, comparing U.S. malaria surveillance data with TDC surveillance data, both of which track the likely country of origin of a malaria case, Québec travelers acquire most African malaria in French-speaking African countries (69%), a minor source of malaria for Americans (18%). English-speaking Ontario and British Columbia likely have more “American” travel patterns than francophone Québecers. However, no fluctuations were seen in rates of travel to either East or West Africa or to the Indian subcontinent, the major source of Canada-acquired P. vivax malaria, which would explain the impressive change in Canadian malaria reporting from 1995 through 1997.

The two surveillance sources of India and the United States were also reviewed for malaria incidence trends. American malaria surveillance includes the likely country of origin of a malaria case. An obvious increase in P. vivax cases from India was seen in the United States, from 150 cases to 371 and down to 123, during 1995 to 1997. This increase paralleled the epidemic peak seen in Canada, primarily in Ontario and British Columbia. In India, an epidemic of P. vivax malaria occurred during this same period (1995–1997) in the Punjabi states of Punjab and Haryana (Figure 7). With negligible changes in travel destination or immigration numbers to explain the 1995–1997 epidemic in Canada, the explanation is probably an increased P. vivax attack rate in Canadians traveling to the Punjab, where a P. vivax epidemic occurred and ended at the same time as the Canadian epidemic.

Canadian notifiable diseases surveillance data generated by local, provincial, and federal sources provided evidence for the occurrence of two as-yet unreported malaria epidemics in Canada in the last decade. One was a P. vivax epidemic, the epicenter of which was almost certainly in the Punjab, India. The second was a P. falciparum epidemic in Québec related to an increased influx of Central African refugees from Tanzanian refugee camps. At the time, neither of these epidemics was brought to the attention of health practitioners in travel clinics through publication or other standard channels. Consequently, possible explanations and potential interventions were not discussed.

Trends in immigration do not explain the malaria incidence changes seen in Canada. These trends differ for each province both in terms of country of origin and numbers. However, the major fluctuations in federal and provincial malaria rates from 1990 to 2002, and, in particular, during the epidemic years, were not found to be directly linked to provincial immigration numbers or to the travel destinations of Canadians in general. Unfortunately, no mechanism records the destinations of travelers from specific provinces. Ontario and British Columbia are home to 86% of the Punjabi-speaking Canadian population. If provincial travel destination data were available, it would likely show that these provinces were the source of most Canadian travelers to the Indian Punjab (27).

Working back from individual case data in each province seems to be the most accurate way to identify countries where large numbers of imported malaria may originate. Country of likely origin of the malaria should be indicated on all requisitions for malaria laboratory diagnosis, and this information and the malaria species should be reported to provincial and then federal surveillance bodies. The fact that the 1995–1997 epidemic was primarily due to P. vivax, the predominant malaria species in India, and that it occurred at the same time as the P. vivax epidemic in the Indian Punjabi states of Punjab and Haryana, is strong evidence to conclude that the Canadian epidemic was an extension of the Punjab epidemic. This association is supported by the abrupt halt of both Canadian and Punjabi epidemics in the same year.

The surveillance process for notifiable diseases in Canada and in other countries where malaria is now an imported disease should be reviewed. Specific conditions, such as the frequency of analysis of surveillance data, need to be discussed and agreed on by collectors of these data at each level of government. Without a firm plan in place for analysis and dissemination of results, the validity, not to mention the utility of the entire surveillance system, is placed in jeopardy. One approach could be the American emerging infections programs, a link between public health, academic, and clinical communities (32).

For surveillance data to be useful and cost-effective, it must be both available in a timely fashion and interpretable. Local surveillance systems have obvious benefits when increased water- and foodborne infections or vaccine-preventable diseases lead to quick public health action. Malaria surveillance differs in two major ways from these classical scenarios. Malaria is an imported disease, and no immediate intervention (e.g., vaccine, chemical disinfectant, and handwashing) will affect an epidemic. As with sexually transmitted infections, the control of a malaria epidemic in Canadian travelers requires public education. In the United States, both malaria speciation and country of likely acquisition of the malaria case are part of surveillance. Such information, if part of the Canadian system, would allow rapid appreciation of the etiology of epidemics such as those reviewed here, which would potentially lead to appropriate public health response.