Illness in Travelers Visiting Friends and Relatives: What Can Be Concluded?

TO THE EDITOR—We were interested to read the GeoSentinel report by Leder et al. [1] that focused on an important group of international travelers who, to date, have not been clearly defined in terms of demographic characteristics and travel-related morbidity. We suggest that there are significant issues related to the design, analysis, interpretation, and conclusions of the study that require comment. Although Leder and colleagues acknowledge several limitations in their report, practitioners who are not familiar with the nature of the GeoSentinel program and/or who do not work with migrant travelers may not fully appreciate the significance of these limitations.

First, although the classification of travelers into 3 groups looks appealing, the classifications have been applied retroactively to the data, and the consequences of this are significant. The retrospective cohort nature of the study design limits the interpretation of outcomes to a cohort association and diminishes the generalizability of the conclusions to wider practice outside of the participating GeoSentinel centers.

Second, there is no design evidence that the recategorization of travelers into “immigrant visiting friends and relatives,” “traveler visiting friends and relatives,” and “tourist,” as defined within the report, is either robust or reliably discriminating for travel-related risk or for health outcomes.

Third, the data recruitment allows for the introduction of both patient referral and selection bias. This may create epidemiological associations that may not be representative of travelers outside of the study group. GeoSentinel sites are often academic or tertiary care centers, and are predominantly based in America; thus, they may be biased towards recruiting tourists rather than travelers visiting friends and relatives. Patterns of access to medical service by migrants may differ from those of the host population [2]. Allowable health insurance coverage and issues of willingness to pay for services in the visited nation [3] may influence pretravel and posttravel service use by travelers visiting friends and relatives. Insurance coverage may be linked to the study's observations of early clinical presentation by tourist travelers, compared with the travelers visiting friends and relatives (who have limited insurance) .

Other design considerations include the acquisition of diseases, such as malaria, which are primarily related to the destination rather than the reason for travel. Analysis of travel to regions of West Africa and East and southern Africa would have been more reflective of actual risk than reasons for travel. There is evidence that travelers visiting friends and relatives are overrepresented as travelers [4] to both Asia and sub-Saharan Africa, and the relative high proportion of disease prevalence in the group may be a reflection of greater exposure to and not increased likelihood of disease. The differing pattern of morbidity among the groups of travelers and immigrants visiting friends and relatives may relate to their economic status, access to and use of services, and medical care–seeking behavior, rather than to travel-associated risk.

All of these factors combined are design issues that we believe makes studies like the Leder et al. [1] study difficult to extend beyond the participating GeoSentinel clinics. Nonetheless, the report by Leder and colleagues and similar studies highlight the importance of defining and determining population-based risk factors in cohorts of travelers. Existing limitations of current data at this time do not allow the associations of outcomes that are demonstrated in the report to be extended to all travelers and immigrants visiting friends and relatives or migrant travelers.

• © 2007 by the Infectious Diseases Society of America