We conducted a cross-sectional study in 3 sites in the southeastern part of Honshu (Boso Peninsula), endemic areas for scrub typhus and Japanese spotted fever (6). We included persons who underwent regular checkups during August–November 2020 (Appendix). Questionnaires were distributed during checkups, and the following data were collected: medical history of rickettsioses; environmental exposure to mountains, agriculture, and bushes; and residential addresses. The respondents were asked through questionnaires whether they resided in or had visited mountainous areas, had visited areas with small trees and weeds, or engaged in agricultural work. In addition, we measured the population density and area of each land use (coasts, forests, farmland, rivers or lakes, and wilderness) within a 500-meter radius of the participant’s address (Appendix). The study was approved by the Institutional Review Boards of Nagasaki University and Fukushima Medical University (approval nos. 200305230-2 and 2022-190). Written consent was obtained from all participants.

The primary outcomes were R. typhi seroprevalence and ratio of R. typhi to Orientia tsutsugamushi seroprevalence. O. tsutsugamushi was selected as the comparator outcome because scrub typhus is a notifiable disease and the rickettsiosis most endemic to Japan. Furthermore, we evaluated the seroprevalence of R. japonica, the pathogen of Japanese spotted fever, to determine the possibility of an apparently high seroprevalence of R. typhi because of cross-reactivity in the genus Rickettsia (7) (Appendix). We defined seropositivity as a ratio of >1:40 and defined O. tsutsugamushi seropositivity as a positive result for any of the O. tsutsugamushi serotypes. The sensitivity analyses estimated the seroprevalences at cutoff titers of 1:80 and 1:160.

Because the seropositivity rates of R. typhi and O. tsutsugamushi were regarded as paired binomial data, we tested the difference in their prevalence by using the McNemar test (8) and estimated it using conditional Poisson regression (Appendix). To explore the factors associated with R. typhi seropositivity, we fitted a logistic regression model by using the candidate risk factors. We assessed whether there were differences in the seroprevalence ratios across study sites and conducted the imputation of missing values (Appendix).

Figure 1

Figure 1. Seroprevalence rates of Rickettsia typhi and Orientia tsutsugamushi in study of seroprevalence and predictors of murine typhus, Japan. R. typhiIgG was detected in 11.3%...

Figure 2

Figure 2. Predictors of Rickettsia typhiIgG seropositivity in study of seroprevalence and predictors of murine typhus, Japan. Shown are adjusted odds ratios for age per 10-year increase; population density increase;...

The median age of all participants was 67 years. R. typhi–seropositive participants exhibited a lower population density than R. typhi–seronegative participants, showing a similar trend to O. tsutsugamushi (Table 1; Appendix Table 1). The residential locations of R. typhi–seropositive participants were distributed throughout the Boso Peninsula, and a similar distribution was observed for O. tsutsugamushi–seropositive participants (Appendix Figures 1, 2). Although ≈60% of R. typhi–seropositive participants had titers of <160, 20 participants had titers of >1,280, and 4 had titers of >40,960 (Table 2). Most O. tsutsugamushi–seropositive participants had lower titers, although some exhibited notably high titers (Appendix Table 2).

R. typhi seroprevalence was 11.3% higher than that of O. tsutsugamushi (7.9%) (ratio of seropositivity 1.42; 95% CI 1.20–1.68) (Figure 1). Of the 2,382 residents, 204 were R. japonica seropositive, for a seroprevalence of 8.6%, lower than that of R. typhi. Furthermore, the antibody titer for R. typhi was higher than that for R. japonica in participants who were seropositive for both R. typhi and R. japonica (Appendix Table 3). Results of the sensitivity analyses did not show any changes to the predominance of the R. typhi seroprevalence over the O. tsutsugamushi seroprevalence (Appendix Table 4).

According to the multivariate analysis (Figure 2), the factors associated with R. typhi seropositivity were age (per 10-year increase; adjusted odds ratio [aOR] 2.09 [95% CI 1.80–2.42]), low population density (per 1,000 persons/km² increase; aOR 0.59 [95% CI 0.40–0.86]), and history of bush exposure (aOR 1.39 [95% CI 1.01–1.92]) (Appendix).

We demonstrated robust findings of the predominance of R. typhi seroprevalence over O. tsutsugamushi seroprevalence in rickettsia-endemic areas in Japan. Previously, in those study areas, an epidemiologic study was conducted using now antiquated serologic methods, such as the Weil–Felix test (4), which provided unreliable MT and scrub typhus seroepidemiologic data (9). Contrary to the previous study’s findings, we were able to estimate rickettsiosis seroprevalence and confirm the predominance of MT more precisely using the standard diagnostic test.

This study illustrated that MT is a prevalent and possibly reemerging infection in Japan. Recently, in the United States (10), Greece (11), and Spain (12), the incidence of MT has increased, partly because of improved disease recognition (10) and a change in the transmission route (13). Thus, given the high seroprevalence of R. typhi in Japan, case accumulation is crucial to clarify the possibility of a unique transmission cycle.

The risk factors for R. typhi seropositivity in this study differed from those in previous studies. The increase in R. typhi seroprevalence with decrease in residential population density contradicts the findings of previous studies that showed urban environment as a risk factor (2,14). In addition, exposure to weeded areas was identified as a risk factor, but residential environments, including those near coasts, rivers, and lakes, which have been reported as risk factors (2,3), were not correlated. The differences in risk factors between this study and previous studies might reflect differences in factors related to contact with vectors and reservoirs at each study site.

The first limitation of our study is that seropositivity to R. typhi could indicate cross-reactivity to R. japonica. However, because R. typhi seropositivity was higher than R. japonica seropositivity and the cross-reactivity rate to R. typhi in confirmed Japanese spotted fever patients is ≈20% (15), most patients with R. typhi seropositivity have a true MT infection. Second, we did not consider cross-reactivity within the same group (spotted fever or typhus group) in the genus Rickettsia. However, other diseases caused by this genus have been reported infrequently, except for Japanese spotted fever and MT in domestic infection cases. Third, this study was conducted in persons undergoing routine checkups and might not represent seroprevalence in the general population.

In summary, R. typhi seroprevalence was higher than that of O. tsutsugamushi in rickettsia-endemic areas of Japan, indicating that MT is a neglected and underrecognized condition. This study highlights the need to include MT in the differential diagnosis when examining patients with nonspecific infectious symptoms who are residing in rickettsia-endemic areas. Clinicians should consider comprehensive examinations for rickettsial infections, including MT testing, especially in those with a history of residence in sparsely populated areas or exposure to bushes.