GeoSentinel is a global network aimed at surveying ill international travelers and was established in 1995 through the International Society of Travel Medicine and the Centers for Disease Control and Prevention. The 41 current GeoSentinel sites contribute clinician-based, anonymous information on all ill travelers seen that is entered in a structured query language database at a central data center. Data collected include demographic information, recent travel history, reason for travel, outpatient or inpatient status, whether the patient was seen during or after travel, whether the patient traveled independently versus organized travel (i.e., travel with a tour group), whether the patient had pretravel encounter with a healthcare provider, patient symptoms, and diagnosis. All GeoSentinel sites use the best reference diagnostic methods available in the country where the site is located. The diagnosis, which may be reported as confirmed, probable, or suspected, is chosen from a standardized list of >500 possible etiologic and syndromic diagnoses, 11 of which refer to rickettsial diseases. Patients may receive several diagnoses, some of which describe well-known complications of rickettsial diseases, e.g., acute renal failure, acute encephalitis, acute respiratory distress syndrome, and death (3,4). The GeoSentinel questionnaire does not contain queries about case management, but in December 2008 all sites were requested to estimate the percentage of their patients with rickettsial diseases who received treatment with antirickettsial drugs (e.g., doxycycline) during the study period.

Data entered into the GeoSentinel database from June 1996 through December 2008 were reviewed. Only travelers seen with confirmed or probable diagnoses were included in the analysis. Cases of rickettsial diseases were defined according to the criteria mentioned below. A subanalysis, comparing SFG rickettsioses with other diseases in ill returnees from sub-Saharan Africa, was also performed (Figure 1).

A microbiologic diagnosis of recent rickettsial disease was in most cases based on immunofluorescence antibody (IFA) tests and, occasionally, on PCR and Western blot. Diagnostic criteria of recent infection by IFA tests were either a 4-fold increase of immunoglobulin (Ig) G or IgM titers in paired serum samples drawn >10 days apart, or elevated IgG and/or IgM titers in single samples consistent with recent infection as interpreted by the local microbiology laboratory. Because IFA tests cannot distinguish between species within the same Rickettsia biogroup, diseases caused by Rickettsia species were dichotomized as SFG rickettsiosis and TG rickettsiosis.

We included all cases of confirmed rickettsial disease defined as a traveler with pertinent travel history and clinical signs, and microbiologic test results supporting recent infection. For SFG rickettsiosis and scrub typhus we also included probable cases defined as a traveler with pertinent travel history and clinical signs including >1 inoculation eschars, and inconclusive or unavailable microbiologic test results. Cases were categorized in 9 disease groups: SFG rickettsiosis, TG rickettsiosis, indeterminate SFG/TG rickettsiosis (defined as cases in whom neither clinical signs nor microbiologic tests could distinguish between SFG and TG rickettsiosis), scrub typhus, ehrlichiosis, anaplasmosis, acute and chronic Q fever (based on the presence of antibodies to the 2 distinct antigenic phases of C. burnetii by IFA test [5]), and bartonellosis. Countries were grouped according to the United Nations region system, and for sub-Saharan Africa also in subregions; for clarification, the southern Africa subregion included Botswana, Lesotho, Namibia, South Africa, and Swaziland. Classification of reason for travel related to current illness was grouped into 4 categories: tourism, business, visiting friends and relatives, and other reason, which included studies or military, missionary, or foreign aid deployment.

All data were analyzed using SAS software, version 9 (SAS Institute, Cary, NC, USA). Comparison of travelers to sub-Saharan Africa with SFG rickettsioses with all other ill travelers to the same destination used the Student t test for continuous variables and Cochran-Mantel-Haenszel statistic for binary variables. The association between possible risk factors and a diagnosis of SFG rickettsioses acquired in sub-Saharan Africa was measured with odds ratios and 95% confidence intervals. Variables included in the univariate analysis were mean age, male gender, travel to southern Africa, travel from March through May, travel >1 week, no pretravel clinic visit, independent travel, and tourism as reason for travel. Variables with a p value <0.05 in the final multivariate logistic regression model were considered significant.

Proportionate morbidity for SFG rickettsiosis acquired in sub-Saharan Africa was defined as the number of cases as a proportion of all ill returned travelers from the region. Analysis of case reports over time was based on monthly proportionate morbidity, i.e., the number of patients with SFG rickettsiosis as a proportion of the number of all ill returned travelers attending any GeoSentinel site in that month. The monthly seasonality was based on aggregate data for that month over all years. The annual proportionate morbidity for SFG rickettsiosis acquired in southern Africa from 1999 through 2008 was estimated in a similar way based on relevant cases for each year. The Cochran-Armitage test was used for testing trend over years. This statistic tests for trend in binomial proportions across levels of a single factor or covariate and is appropriate for a contingency table where the response variable has 2 levels and the explanatory variable is ordinal. A 1-sided p value <0.05 would indicate a decreasing (with a negative statistic) or increasing (with a positive statistic) trend.

Of the 231 cases of SFG rickettsioses (146 confirmed and 85 probable) a total of 136 (58.9%) were men; the mean age was 43.4 years (median 45 years, range 33–53 years). Tourists comprised 182 (78.8%), 28 (12.1%) had traveled for business, 6 (2.6%) were visiting friends and relatives, and 14 (6.1%) had traveled for other reasons. One hundred ninety-seven (87.6%) case-patients were infected in sub-Saharan Africa; South Africa (n = 135), Zimbabwe (n = 13), and Tanzania (n = 7) were the 3 most commonly reported countries of exposure (Table 1). The median time from travel to reporting to a GeoSentinel site was 8 days (range 4–12 days). Two case-patients, a 23-year-old French woman tourist to Mongolia in June 2008, and a 27-year-old Swiss woman student traveler to Corsica, France, in October 2008, were infected by Rickettsia slovaca confirmed by PCR and Western blot. There were no deaths among the travelers with SFG rickettsioses, but 22 (9.6%) were hospitalized, and multiorgan failure with acute respiratory distress syndrome developed in 2 Israeli men 47 and 73 years of age, respectively, after they traveled to India in October 2004.

Analysis of demographic and exposure variables for the 197 travelers to sub-Saharan Africa with SFG rickettsiosis compared with 11,690 travelers to the same region with other diagnoses is shown in Table 2. A multivariate logistic regression model was performed to analyze the effects on SFG rickettsiosis of risk factors including age, gender, travel to southern Africa, travel from March to May, travel duration longer than 7 days, no pretravel clinic visit, independent travel, and travel for tourism. Older age, male gender, travel to southern Africa, travel from March to May, and travel for tourism were found to be independently associated with SFG rickettsiosis. An analysis of monthly proportionate morbidity identified a peak of cases in March, April, and May (Figure 2). The proportionate morbidity of cases with SFG rickettsiosis in ill returnees from southern Africa from 1999 through 2008 was 139/1,017 (13.7%). The overall Cochran-Armitage trend test did not show statistically significant change (p = 0.877, by 2-sided trend test). As shown in Figure 3, the annual proportionate morbidity increased and decreased twice over the years and did not maintain a monotone increasing or decreasing pattern.

The 10 travelers with TG rickettsiosis, of whom 6 were men, had a mean age of 28.0 years (median 25.5 years, range 20–50 years). Four case-patients were tourists, 2 were visiting friends and relatives, 1 was a business traveler, and 3 had traveled for other purposes. Southeast Asia, including 3 travelers infected in Indonesia, was the most common region of exposure (Table 1). The median time from travel to reporting a site was 6 days (range 3–14 days). Five patients were hospitalized and no complications were noted. One patient, a 23-year-old Japanese man visiting friends and relatives in Bali, Indonesia, had R. typhi infection confirmed by PCR.

Indeterminate SFG/TG rickettsiosis was diagnosed in 4 male tourists, 15, 50, 51, and 67 years of age, respectively. Two case-patients were infected in Southeast Asia (Table 1). No complications were recorded.

The 16 travelers with scrub typhus (5 confirmed and 11 probable), of whom 10 were men, had a mean age of 36.6 years (median 32.5 years; range 26–67 years). Nine patients were infected in Southeast Asia (including Thailand, n = 4) and 5 in south-central Asia (including India, n = 3). Twelve patients were tourists, 1 was visiting friends and relatives, and 3 had traveled for business. The median time from travel to reporting to a GeoSentinel site was 9 days (range 2–25 days). Six case-patients were hospitalized. Encephalitis developed in 2 Singaporean travelers: a 42-year-old man, a tourist infected in Thailand in December 2007, died, and a 51-year-old man infected in Taiwan while traveling on business in April 2008 survived.

An uncomplicated case of human granulocytic anaplasmosis was seen in August 2008 in a 32-year-old US woman tourist to the Netherlands. The diagnosis was based on detection of intracellular morulae.

The 11 travelers with acute Q fever, including 4 men, had a mean age of 53.0 years (median 59 years; range 20–64 years). Sub-Saharan Africa, including 2 case-patients infected in Tanzania, was the most common region of exposure (Table 1). Seven patients were tourists, 3 were business travelers, and 1 had traveled for other reasons. The median time from travel to reporting to a site was 21 days (range 5–52 days). Isolated fever (n = 7) and respiratory symptoms (n = 4) were the most common clinical manifestations. Four patients were hospitalized. All 11 patients with Q fever had uncomplicated clinical courses.

The 7 travelers with bartonellosis, including 3 males, had a mean age of 33.6 years (median 32 years, range 9–56 years). Three patients were infected in the Caribbean (Table 1). Four were tourists and 3 had traveled for other purposes. The median time from travel to reporting to a GeoSentinel site was 25 days (range 7–39 days). Bartonella henselae infection was diagnosed in all case-patients by serologic or immunohistochemical analysis. Six travelers had uncomplicated cat-scratch disease, and 1 had bacillary angiomatosis.