|
A correction has been published: N Engl J Med 2006;355(9):967.
|
Background Approximately 8 percent of travelers to the developing world require medical care during or after travel. Current understanding of morbidity profiles among ill returned travelers is based on limited data from the 1980s.
Methods Thirty GeoSentinel sites, which are specialized travel or tropical-medicine clinics on six continents, contributed clinician-based sentinel surveillance data for 17,353 ill returned travelers. We compared the frequency of occurrence of each diagnosis among travelers returning from six developing regions of the world.
Results Significant regional differences in proportionate morbidity were detected in 16 of 21 broad syndromic categories. Among travelers presenting to GeoSentinel sites, systemic febrile illness without localizing findings occurred disproportionately among those returning from sub-Saharan Africa or Southeast Asia, acute diarrhea among those returning from south central Asia, and dermatologic problems among those returning from the Caribbean or Central or South America. With respect to specific diagnoses, malaria was one of the three most frequent causes of systemic febrile illness among travelers from every region, although travelers from every region except sub-Saharan Africa and Central America had confirmed or probable dengue more frequently than malaria. Among travelers returning from sub-Saharan Africa, rickettsial infection, primarily tick-borne spotted fever, occurred more frequently than typhoid or dengue. Travelers from all regions except Southeast Asia presented with parasite-induced diarrhea more often than with bacterial diarrhea.
Conclusions When patients present to specialized clinics after travel to the developing world, travel destinations are associated with the probability of the diagnosis of certain diseases. Diagnostic approaches and empiric therapies can be guided by these destination-specific differences.
Much of the current understanding of morbidity profiles among ill returned travelers is based on data that were collected in the 1980s.1,2,3,4 Although some morbidity studies have been designed to examine individual diseases,5,6,7,8,9 specific high-risk destinations,10 and certain types of travelers,11,12,13 a comprehensive, multicenter comparison of the spectrum of illnesses acquired by a broad range of travelers returning from developing regions on all continents has been lacking.14
GeoSentinel sites are specialized travel or tropical-medicine clinics on six continents that collect clinician-based surveillance data on travel-related diseases. These data include a broad sample of travel destinations and of morbidity among persons who have become ill while traveling. We used the GeoSentinel database to verify the assumption that the destination of travel is associated with the probability of each diagnosis among travelers returning from the developing world.
Methods
Geosentinel Sites
The 30 current GeoSentinel sites are specialized travel or tropical-medicine clinics on six continents. They are recruited, on the basis of demonstrated training and experience and a record of publication in travel or tropical medicine, to contribute clinician-based sentinel surveillance data on all ill returned travelers seen (as detailed by the International Society of Travel Medicine, at www.istm.org, and by Freedman et al.15). The sites accounting for the majority of patients seen are within academic centers; several smaller-volume sites (almost all with a current academic affiliation) are in freestanding locations. Intake at the sites reflects a mixed population of patients requiring tertiary care and self-referred patients. Some sites are restricted to outpatient care, and at no site is practice limited to the care of ill travelers.
Data Collection
Criteria for Entry into the GeoSentinel Database
Patients must have crossed an international border within 10 years before presentation and have sought medical advice for a presumed travel-related illness. Anonymous surveillance data (including travel history) that could not be linked back to an individual patient were entered into a database at a central data center. Final diagnoses were assigned codes by the treating clinician from a standardized list of 524 possible individual diagnoses that were also categorized under 21 broad syndromes. (Diagnosis codes grouped according to syndrome are listed in the Supplementary Appendix, available with the full text of this article at www.nejm.org.) All sites used the best available reference diagnostic test in their own country. Patients were assigned as many diagnosis codes as needed. Since most infections are associated with fever, diagnoses predominantly localized to one organ system were included in that organ-system syndrome category and were not attributed to the broader category of systemic febrile disease.
The GeoSentinel data-collection protocol was reviewed by the institutional review board officer at the National Center for Infectious Diseases at the Centers for Disease Control and Prevention and classified as public health surveillance and not as human-subjects research requiring submission to institutional review boards.
Eligibility for Analysis
All travelers who presented to a GeoSentinel site after travel to the developing world were eligible for analysis. Patients with diagnoses that were either laboratory-confirmed or deemed probable were included in the analysis. Eligible patients who were ultimately found after medical assessment to have no underlying disease (1942 patients) and those found to have final diagnoses that were not confirmed or probable (1518 patients) were excluded.
Comparative Morbidity According to Travel Destination
We analyzed the six developing regions of the world where the majority of these illnesses were acquired (Figure 1). Recent, temporally clustered travel to more than one country within a region often makes it difficult to attribute the illness to a specific country. The place of exposure during travel was therefore determined to be the single region visited or, for travelers who entered more than one region, was determined according to the data field for "most likely place of exposure" if that information could be designated as certain by the clinician by virtue of incubation period or known patterns of endemicity (Figure 1). To calculate a cumulative morbidity profile for all ill returned travelers, data from patients for whom a single region of exposure could be determined were grouped and then combined with data for ill travelers to multiple regions in whom ascertainment of the relevant place of exposure was impossible.
|
The data analysis was performed with the use of SAS software, version 8 (SAS Institute). The primary variable analyzed was proportionate morbidity, calculated as the number of patients with a specific diagnosis or group of diagnoses as a proportion of all ill travelers returning from a developing region. All results are given in terms of the number of patients per 1000 patients in the group. Analyses of specific etiologic organisms according to region were performed only within an individual syndrome group. This analytic approach prevented a very frequent diagnosis within a region, such as malaria among travelers to sub-Saharan Africa, from overwhelming the contributions of various causes associated with other syndrome groups, such as diarrhea, that supplied proportionately fewer cases. Chi-square tests were used to compare these proportions among regions. Because of the large numbers of statistical tests performed, a conservative two-sided significance level of P<0.01 was chosen.
Results
Overall Morbidity
The characteristics of 17,353 patients whose data were reported to GeoSentinel from June 1996 through August 2004 and who met the inclusion criteria are shown in Table 1. These patients, with travel exposures in 230 countries, presented to GeoSentinel sites located in Europe (49 percent), the United States or Canada (33 percent), Israel (8 percent), Australia or New Zealand (8 percent), and other sites (3 percent). More than 50 percent of the ill travelers had obtained documented advice from a medical provider before traveling. Diagnoses for 67.0 percent of all the returned travelers fell into 4 of 21 major syndrome categories: systemic febrile illness, acute diarrhea, dermatologic disorders, and chronic diarrhea (Table 2). The median duration of the most recent trip ranged from 14 days for the Caribbean to 35 days for South America. Although most of the patients (64 percent) were seen within a month after travel, 10 percent had indolent diseases or diseases with a long incubation period and were not seen until more than six months after travel (Table 1). The effects of pretravel care were not measured.
|
|
Destination Analysis
For 13,836 of the ill returned travelers (80 percent), the relevant place of exposure could be determined to be one of the six major regions that receive travelers in the developing world. Most of the patients had traveled as tourists, except for travelers to sub-Saharan Africa, which had the highest proportions of business travelers and travelers visiting friends and relatives (Table 1). Unlike those returning from all other regions, less than a third of ill persons returning from the Caribbean had received advice before traveling.
Significant differences in proportionate morbidity (P<0.01) were seen among the travel regions for 16 of the 21 syndrome categories (Table 2). In particular, systemic febrile illness was found disproportionately among patients presenting to GeoSentinel sites after travel to sub-Saharan Africa or Southeast Asia and acute diarrhea among those presenting after travel to south central Asia. Dermatologic disorders were seen disproportionately less commonly among travelers returning from sub-Saharan Africa or south central Asia. Diagnoses contributing to death included severe and complicated malaria, pulmonary embolism, pneumonia, and pyogenic abscess.
Regional Morbidity According to Specific Diagnosis
Destination-specific variations in the proportionate morbidity associated with etiologic diagnoses within each of the top four syndromic categories are presented in Table 3. Etiologic diagnoses were not commonly ascertained for chronic diarrhea, so they are not listed.
|
For all regions except Southeast Asia, parasite-induced diarrhea was more common among ill returned travelers than was bacterial diarrhea. Patients with bacterial diarrhea presented most commonly after travel in Southeast Asia; campylobacter was the predominant cause. Of the parasitic causes, giardiasis was reported disproportionately among travelers returning from south central Asia.
Overall, insect bites were the most common cause of dermatologic problems, followed by cutaneous larva migrans, allergic reactions, and skin abscesses. However, cutaneous larva migrans was the most common dermatologic disorder among patients presenting after travel to the Caribbean, among whom it was seen much more commonly than among those returning from sub-Saharan Africa, south central Asia, or Central or South America. Bacterial skin infections, including skin abscesses, were found more commonly among patients returning from sub-Saharan Africa, south central Asia, or Southeast Asia than among those returning from the other three regions. Leishmaniasis was found mostly among patients who had traveled to South America or, to a lesser extent, Central America. Myiasis was reported most frequently among patients who had traveled to South America or Central America.
Travelers returning from each of the regions presented with intestinal nematode infestations, primarily involving ascaris and intestinal strongyloides. Of all ill returned travelers who presented with nondiarrheal gastrointestinal diagnoses, 40 per 1000 seen at our sites were reported to have acute hepatitis A, 20 per 1000 hepatitis B, 6 per 1000 hepatitis C, 13 per 1000 hepatitis E, and 36 per 1000 unspecified hepatitis. The sample size was too small to allow us to assess regional differences for hepatitis. Schistosomiasis (due to Schistosoma mansoni or S. haematobium) was seen predominantly among travelers returning from sub-Saharan Africa (196 per 1000). Acute brucellosis, leptospirosis, cysticercosis, filariasis, histoplasmosis, and echinococcosis occurred too infrequently to allow comparison among regions.
Rare Diagnoses
Clinicians evaluating returned travelers frequently entertain rare or exotic diagnoses. Travel-related cases of Ebola virus disease, Japanese encephalitis, rabies, tetanus, diphtheria, plague, tularemia, murine typhus, Rift Valley fever, poliomyelitis, primary amebic meningoencephalitis, anthrax, or yellow fever are reported sporadically in the literature. No cases of any of these diagnoses occurred among the 17,353 travelers whose data were analyzed in this study or among any of the 25,023 patients whose records were included in any category in our database but were excluded from this study. Among the 17,353 patients in our cohort, each of the following diagnoses occurred only once: Angiostrongylus cantonensis infestation, A. costaricensis infestation, hantavirus infection, cholera, melioidosis, Ross River virus infection, African trypanosomiasis, legionellosis, and meningococcal meningitis.
Proportionate Morbidity According to Region
A summary profile of proportionate morbidity among ill returned travelers according to diagnosis or diagnosis group, expressed in terms of the proportion per 1000 ill returned travelers, is presented in Figure 2. Shown is the proportion, not the incidence rate, of each of the top 22 specific diagnoses among all ill returned travelers and among travelers returning from each of the regions.
|
Our database represents a large sample of ill travelers returning from the developing world. GeoSentinel sites are located primarily within academic centers, so some of the patients seen at each site present on the basis of consultations or outside referrals. Patients with imported illnesses that are neither self-limited nor mild are generally seen at such practices at some point during their care. At the same time, these practices are also the initial point of entry for many returned travelers who had pretravel medical preparation at the same clinic or for those who are affiliated with corporations, religious organizations, aid agencies, or governmental entities with which the clinic has an ongoing relationship or contractual arrangement. With significant growth in international travel, a travel or tropical-medicine clinic or infectious-diseases practice generally emerges in a community as the local specialized resource for providers and as the primary entry point for increasingly sophisticated travelers. Our data provide a reference for likely diagnoses, stratified according to destination, among travelers seeking care at or referred to such practices.
The data do not represent a comprehensive epidemiologic analysis of all illness in all travelers. Similarly, they do not represent a sample of illnesses in returned travelers such as those who would be seen at a nonspecialized, primary care practice, where mild or self-limited conditions would occur with higher frequency. Diagnoses with frequencies that may be underrepresented in GeoSentinel include diseases with a short incubation period, such as dengue; many cases of such diseases manifest during travel, so cases reported in returned travelers represent those acquired at the end of a trip. Finally, the small proportion of returned travelers with severe acute disease who need immediate hospitalization and are initially admitted through an emergency department may be underrepresented if the diagnosis is not considered related to travel or if care was sought outside the network. Travelers to destinations self-perceived to be particularly risky may present to a specialized unit earlier or with milder illness than travelers to more familiar destinations. Our data do not allow us to estimate incidence rates or to provide numerical risk for travel to particular destinations.
On an aggregate basis, the GeoSentinel database is a sample of illnesses acquired during exposure in 230 countries by sentinel travelers presenting in 13 countries for care. Nevertheless, biases for certain behavior-related diagnoses probably exist because of the geographic distribution and proportionate contributions of the individual sites. Patient intake at each site reflects local or national differences in the makeup of the traveling population, the distribution of travel destinations, and access to medical care. In addition, accommodations, eating habits, and other risk behaviors at a given destination may reflect the national and cultural background of the traveler. European travelers are more heavily represented in the sample than are North Americans (49 percent vs. 33 percent), and those from other regions (18 percent) are relatively underrepresented. Travel-related sexually transmitted diseases (proportionate morbidity, 8 per 1000 patients) and infections with the human immunodeficiency virus (2 per 1000 patients) were reported, but they are probably underrepresented in tropical-disease units; persons with self-recognized risk exposures or characteristic genitourinary symptoms may seek care in other settings.
At the same time, the benefit of a multicenter perspective over smaller, single-site studies, which have often reflected local patterns of travel, is demonstrated by the analysis of dermatologic disease. The most widely cited study of dermatologic disorders in travelers included 269 persons returning to France.16 In that study, more than 60 percent of the patients had returned from travel to francophone countries of the Caribbean (Martinique and Guadeloupe), from French Guiana, and from sub-Saharan Africa. The finding that cutaneous larva migrans, pyodermas, and arthropod bites were among the top diagnoses agrees with our own overall findings in all ill travelers. However, the significant destination-specific differences that we are able to describe for conditions such as cutaneous larva migrans, leishmaniasis, myiasis, swimmer's itch, and animal bites were not assessed in the earlier study.
The geographic dispersion of our sites, the full range of possible diagnoses, and the range of diagnostic tests necessary to identify the causes of tropical diseases preclude centralized laboratory testing for all patients. We used the best available national reference diagnostics for confirmed diagnoses. Probable diagnoses were restricted to patients who had an indisputable physical finding (e.g., tickborne rickettsiosis, cutaneous larva migrans, myiasis, or tungiasis), a response to a highly specific therapeutic agent, or a classic clinical presentation and exposure history, with other possible causes definitively ruled out by laboratory analysis. The last situation is particularly applicable to dengue, for which results of serologic tests are often difficult to obtain immediately and are often negative during the acute illness at the time of clinical presentation; in addition, patients with dengue often do not return for follow-up when faced with the cost of expensive diagnostic evaluation of a self-limited illness, particularly when the symptoms have resolved.
Codes available for a full range of syndromic diagnoses were used to classify patients consistently when a specific etiologic diagnosis was not assignable. Reasons included the frequency of self-limited acute infections and the effectiveness of empiric treatment or self-treatment for many syndromes, as compared with the expense and difficulty of diagnostic evaluation for exotic agents. For the syndromic groups of acute diarrhea and systemic febrile illness, specific etiologic diagnoses were not made in 40 percent or more of cases. Other studies have also found similarly high rates of undiagnosed fever17,18 and diarrhea19 in similar patients. Nonetheless, the clinicians in our study are experienced in disease recognition and diagnosis and have access to usually available diagnostic methods; the treatable causes of illness that are routinely sought in an initial clinic visit would probably not have been missed.
Our data on diarrheal illness illustrate important differences among studies that evaluate illness occurring during travel as compared with that seen after travel. All etiologic studies of travelers' diarrhea performed in the visited country during travel have shown a predominance of bacterial causes and a relative paucity of parasitic causes.19,20,21 With respect to post-travel illness seen at our sites, diarrhea caused by parasites predominated over bacterial diarrhea in the overall analysis of ill returned travelers and in the analysis of every individual region except Southeast Asia. The likely explanation is that parasite-induced diarrhea tends to have a longer incubation period than bacterial diarrhea and to be more chronic and less treatable by the empirical antibiotic agents carried by travelers or prescribed at an initial medical encounter. Similar considerations most likely apply to other syndrome groups as well.
The high frequency of rickettsial disease reflects the emergence of Rickettsia africae in recent years in southern Africa, as has been reported in several single-site studies.22 Dengue in travelers is well reported in studies from the 1980s onward, but in contrast to earlier data, it now appears to occur more frequently than malaria among travelers returning from any region except Africa and Central America. This change may reflect the use of effective antimalarial chemoprophylaxis by travelers. The proportionate morbidity associated with dengue is especially high among travelers returning from Southeast Asia and the Caribbean.
The current data provide a reference for likely diagnoses in returned travelers, with stratification according to destination but not according to prophylaxis designed before travel. The profile of proportionate morbidity generated by these data is important for guiding post-travel diagnosis and empiric therapy as well as for prioritizing pretravel intervention strategies.
Supported by a cooperative agreement (U50/CCU412347) from the Centers for Disease Control and Prevention (CDC), Atlanta, and by an initial pilot grant from the International Society of Travel Medicine (both to GeoSentinel, the global surveillance network of the International Society of Travel Medicine and of the CDC).
Dr. Freedman reports having received consulting fees from Shoreland, Sanofi Pasteur, GlaxoSmithKline, and Salix Pharmaceuticals and owning equity in Shoreland. Dr. Weld reports owning equity in Amgen and Chiron. Dr. Kozarsky reports having received consulting fees from Berna Products and lecture fees from GlaxoSmithKline. Dr. Keystone reports having received consulting fees from Sanofi Pasteur and GlaxoSmithKline and speaking fees from Roche and GlaxoSmithKline. No other potential conflict of interest relevant to this article was reported.
The views expressed in this article are those of the authors and do not necessarily represent those of the CDC.
We are indebted to Elena Axelrod, Indira Srinivasan, Sue Baldwin, Christina Bernadotte, Raj Srinivas, and members of the staff at each GeoSentinel site for data management or computer programming; and to Brenda Bagwell and Adam Plier for ongoing administrative support.
Source Information
From the Division of Geographic Medicine, University of Alabama at Birmingham, Birmingham (D.O.F.); the Division of Global Migration and Quarantine, Centers for Disease Control and Prevention, Atlanta (L.H.W., P.E.K., R.R., M.S.C.); the Division of Infectious Disease, Emory University, Atlanta (P.E.K., T.F.); the Department of Tropical and Infectious Diseases, University of Munich, Munich, Germany (F.S.); the Division of Infectious Diseases, University of Toronto, Toronto (J.S.K.); and CIWEC Clinic, Kathmandu, Nepal (P.P.).
Dr. Fisk is deceased.
Address reprint requests to Dr. Freedman at the Division of Geographic Medicine, 1530 Third Ave. S., BBRB 203, Birmingham, AL 35294, or at freedman{at}uab.edu.
References
In addition to the authors, members of the GeoSentinel Surveillance Network include the following: Kaiser Permanente, Honolulu — V. Ansdell (Oct. 1997 to Jan. 2003 only); Boston University, Boston — E. Barnett; Royal Melbourne Hospital, Melbourne, Australia — G. Brown and J. Torresi; University of Brescia, Brescia, Italy — G. Carosi and F. Castelli; Harvard University, Cambridge, Mass. — L. Chen and M. Wilson; Cornell University, New York — B. Connor; Hôpital Nord, Marseille, France — J. Delmont and P. Parola; Mount Sinai Medical Center, New York — A. Gurtman; University of Utah, Salt Lake City — D. Hale and S. Gelman; Hudson River Health Care, Peekskill, N.Y. — N. Piper-Jenks; University of Washington, Seattle — E. Jong; Travellers Medical and Vaccination Centres of Australia, Adelaide — R. Kass (Dec. 1997 to March 2001 only); University of Toronto, Toronto — K. Kain; Orlando Regional Health Center, Orlando, Fla. — C. Licitra; University of Geneva, Geneva — L. Loutan and F. Chappuis; Fresno International Travel Medical Center, Fresno, Calif. — M. Lynch; Tulane University, New Orleans — S. McLellan; Travel Clinic Services, Johannesburg, South Africa — R. Muller; National Institutes of Health, Bethesda, Md. —T. Nutman and A. Klion; Catholic University of Chile, Santiago — C. Perret and F. Valdivieso; Johns Hopkins University, Baltimore — B. Sack and R. MacKenzie; Sheba Medical Center, Tel Hashomer, Israel — E. Schwartz; Travellers Health and Vaccination Centre, Auckland, New Zealand — M. Shaw; University of Zurich, Zurich, Switzerland — R. Steffen and P. Schlagenhauf; Albert Einstein School of Medicine, Bronx, N.Y. — M. Wittner; and Royal Free Hospital, London — J. Zuckerman (Sept. 2000 to May 2003 only).
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Related Letters:
Illness in Returned Travelers
Bishai D. M., Baker T., Freedman D. O., Weld L. H., Kozarsky P.
Extract |
Full Text |
PDF
N Engl J Med 2006;
354:1851, Apr 27, 2006.
Correspondence
This article has been cited by other articles:
HOME | SUBSCRIBE | SEARCH | CURRENT ISSUE | PAST ISSUES | COLLECTIONS | PRIVACY | TERMS OF USE | HELP | beta.nejm.org Comments and questions? Please contact us. The New England Journal of Medicine is owned, published, and copyrighted © 2009 Massachusetts Medical Society. All rights reserved. |
Previous
