Introduction

Scrub typhus (tsutsugamushi disease; ST) and Japanese spotted fever (JSF) are endemic rickettsial diseases in Japan [1,2]. Early recognition is important because timely antimicrobial therapy is associated with better outcomes [3,4]. However, even in endemic areas, nearly 30% of patients were not correctly diagnosed at the first visit to a participating site [5].

The clinical presentation of these infections overlaps with that of many other infectious and inflammatory conditions, which makes early diagnosis difficult [6]. Rash and eschar can provide useful clues, but they may not be reported by patients, may not be elicited during history taking, or may not be identified on examination [5].

Previous clinical studies have generally evaluated rash and eschar as conventional clinical variables, typically recording whether they were present-or-absent, rather than how they came to clinical attention [4,7]. In practice, whether rash or eschar contributes to correct diagnosis may depend on how these findings come to clinical attention, not merely on their presence or absence. We therefore examined how rash and eschar came to clinical attention when the diagnosis was established, and whether these recognition patterns differed between patients diagnosed at the first visit to a participating site and those diagnosed later.

Methods

Results

Among 217 patients with rickettsial diseases, 156 (71.9%) were correctly diagnosed at the first visit to a participating site and 61 (28.1%) were diagnosed after one or more subsequent visits.

Baseline characteristics

Table 1 summarizes participant characteristics at the time the correct diagnosis was established. Compared with the diagnosed-at-first-visit group, the delayed-diagnosis group more often included patients aged 75 years or older, direct-visit patients, and patients with a longer interval from symptom onset to correct diagnosis. Among 217 patients, 140 were in the direct-visit group and 77 were in the non-direct-visit group. Correct first-visit diagnosis was more common in the non-direct-visit group than in the direct-visit group (63/77 [81.8%] vs 93/140 [66.4%], p = 0.018).

Download:

• PNG
  larger image
• TIFF
  original image

Table 1. Baseline characteristics of the diagnosed-at-first-visit group and the delayed-diagnosis group.

https://doi.org/10.1371/journal.pntd.0014377.t001

First documented incorrect diagnoses

Table 2 shows the first documented incorrect diagnoses before the correct diagnosis was established among 54 of the 61 patients in the delayed-diagnosis group. These incorrect diagnoses were heterogeneous: upper respiratory viral infection (n = 15, 27.8%), bacterial infections such as sepsis, urinary tract infection, or pneumonia (n = 12, 22.2%), and fever of unknown origin (n = 11, 20.4%) were most common. Among these 54 patients, the clinical department associated with the first documented incorrect diagnosis could be determined in 25 patients. These cases most often involved frontline care settings, including emergency medicine, general internal medicine, and the emergency department internal medicine duty service. Of the 54 patients, 40 were in the direct-visit group and 14 were in the non-direct-visit group. Because the number of non-direct-visit patients was small, we did not perform a formal statistical comparison of diagnostic categories between the two groups. Descriptively, upper respiratory viral infection, bacterial infections, and fever of unknown origin were the predominant categories in both groups.

Download:

• PNG
  larger image
• TIFF
  original image

Table 2. First documented incorrect diagnoses in the delayed-diagnosis group.

https://doi.org/10.1371/journal.pntd.0014377.t002

Recognition patterns and correct first-visit diagnosis

In multivariable analysis, several recognition patterns for rash and eschar differed between the diagnosed-at-first-visit group and the delayed-diagnosis group (Fig 1). Among the skin-related variables, rash reported as a chief complaint (aOR 4.95, 95% CI 2.07–11.84), rash elicited during history taking (aOR 5.86, 95% CI 2.80–12.27), and eschar identified on physical examination (aOR 2.80, 95% CI 1.19–6.58) were each independently associated with correct first-visit diagnosis. Eschar was identified on physical examination in 91.5% of the diagnosed-at-first-visit group and in 78.0% of the delayed-diagnosis group. Age 75 years or older was associated with lower odds of correct first-visit diagnosis (aOR 0.40, 95% CI 0.21–0.75). Several additional factors were associated with higher odds of correct first-visit diagnosis: evaluation by general internal medicine at the time of correct diagnosis (aOR 1.85, 95% CI 1.01–3.39), and AST > 33 IU/L (aOR 2.34, 95% CI 1.09–5.02). Conversely, direct visit (aOR 0.46, 95% CI 0.23–0.92), creatinine >1.2 mg/dL (aOR 0.35, 95% CI 0.15–0.81), and lung crackles or infiltrates on chest radiography (aOR 0.38, 95% CI 0.16–0.89) were associated with lower odds of correct first-visit diagnosis. Full adjusted results are shown in S1 Table.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 1. Variables associated with correct diagnosis at the first visit to a participating site.

Forest plot of adjusted odds ratios (aORs) with 95% confidence intervals (CIs) for variables associated with correct first-visit diagnosis among 217 patients with rickettsial diseases. The vertical dashed line indicates an aOR of 1.0. Rash as the chief complaint, rash elicited during history taking, eschar detected on physical examination, evaluation by general internal medicine at the time of correct diagnosis, and AST > 33 IU/L were associated with correct first-visit diagnosis, whereas age ≥ 75 years, direct visit, creatinine >1.2 mg/dL, and lung crackles or infiltrates on chest radiography were associated with lower odds of correct first-visit diagnosis. aORs were estimated using logistic regression models adjusted for age as a continuous variable, sex, and clinical department at the time of correct diagnosis. When age ≥ 75 years was evaluated as an exposure variable, continuous age was omitted from the adjustment set; when clinical department was evaluated as an exposure variable, clinical department was omitted from the adjustment set. Because variables were assessed at the time the correct diagnosis was established, these associations should be interpreted cautiously. Full results are shown in S1 Table.

https://doi.org/10.1371/journal.pntd.0014377.g001

All patients in whom an eschar was elicited during history taking were correctly diagnosed at the first visit to a participating site. Therefore, an adjusted odds ratio could not be estimated for this variable. In additional exploratory analyses, rash-related recognition patterns remained positively associated with correct first-visit diagnosis across age groups and study periods, although several subgroup estimates were less precise because of smaller sample sizes (S2 and S3 Tables). Baseline characteristics by direct-visit status are shown in S4 Table. Similar analyses stratified by direct-visit status are shown in S5 Table. The proportion of correct first-visit diagnosis also differed across grouped clinical departments (S6 Table).

Discussion

In this endemic setting, the way rash and eschar came to clinical attention differed between the diagnosed-at-first-visit group and the delayed-diagnosis group. Specifically, compared with the delayed-diagnosis group, the diagnosed-at-first-visit group more often had rash recognized as a chief complaint or during history taking, and eschar recognized on physical examination. Notably, eschar was identified on physical examination in 78.0% of patients in the delayed-diagnosis group at the time the correct diagnosis was established. This pattern is consistent with the parent study, in which patients more often noticed rash than eschar spontaneously [5]. Together, these observations suggest that early recognition may depend not only on the presence of rash or eschar, but also on whether these findings are actively sought during clinical evaluation. These supplementary analyses also suggested that the overall direction of the rash-related associations was broadly preserved across study periods, age groups, and direct-visit groups, although several subgroup estimates were imprecise.

The broad spectrum of incorrect diagnoses documented before the correct diagnosis was established reflects the diagnostic challenge that precedes the recognition of rash and eschar. These diagnoses were heterogeneous, consistent with the nonspecific early presentation of rickettsial diseases (Table 2) [4,12]. Fever was common in both groups but was not independently associated with correct first-visit diagnosis, highlighting its limited discriminatory value in this setting. Older age was associated with lower odds of correct first-visit diagnosis, though the underlying reasons remain unclear. Possible explanations include less specific symptom reporting in older adults, which may reflect pre-existing cognitive impairment, subtle changes in attention or consciousness during acute illness, as well as greater diagnostic competition from common infections in this age group and under-recognition of skin findings during initial frontline evaluations. These interpretations remain speculative.

In endemic areas, targeted questioning about rash and thorough skin inspection for eschar are practical steps toward earlier recognition of ST and JSF in routine care. In ST, eschars may occur in soft skin folds and covered areas such as the axillae, inguinal region, genital area, and trunk [13]. In JSF, attention to the extremities and to rash involving the palms and soles may aid recognition [5].

Lung crackles or infiltrates on chest radiography and elevated creatinine were associated with lower odds of correct first-visit diagnosis. Because these variables were assessed at the time the correct diagnosis was established rather than at the initial visit, we could not determine whether they contributed to initial misdiagnosis, reflected subsequent disease progression, or both. Rickettsial diseases can involve the lungs and kidneys [14–16]. From a clinical perspective, however, such findings may have anchored clinicians toward alternative diagnoses such as pneumonia, urinary tract infection, or sepsis. These findings do not exclude ST or JSF from consideration in endemic settings. In contrast, elevated AST was associated with higher odds of correct first-visit diagnosis. AST was assessed at the time the correct diagnosis was established and may reflect disease severity or diagnostic context rather than an independent contributor to early recognition.

In Japan, general internal medicine is a hospital-based specialty that evaluates undifferentiated and complex medical conditions, including fever of unknown origin. In this setting, it typically receives referrals from primary care physicians and emergency departments. The associations with direct visit and general internal medicine likely reflect the diagnostic pathway rather than direct causal effects. Some non-direct-visit patients, and some patients seen by general internal medicine, may have arrived with accumulated clinical information from prior evaluation. For example, failure to respond to initial antibiotics or a persistent fever with rash may have narrowed the differential diagnosis toward rickettsial disease. Consistent with this interpretation, correct first-visit diagnosis at the participating sites was more common in the non-direct-visit group than in the direct-visit group (81.8% vs 66.4%, p = 0.018). However, some non-direct-visit patients who were classified as being in the diagnosed-at-first-visit group had already experienced a delay in diagnosis from their first point of care. These variables should therefore be interpreted as indicators of diagnostic context, not specialty-specific advantages.

We analyzed ST and JSF together because our aim was to evaluate shared processes related to first-visit recognition, not to distinguish disease-specific clinical features. Both infections also require similar initial clinical consideration in endemic settings.

This study has several limitations. Most importantly, predictor variables were assessed at the time of correct diagnosis, not at the first visit. We therefore could not determine whether these findings had been present but overlooked or became more apparent over time. In the delayed-diagnosis group, the initial visit sometimes occurred outside the participating sites, and even when first-visit records were available, documentation of clinical findings was often incomplete when rickettsial disease was not initially suspected. Second, because this secondary analysis relied on medical-record review across retrospective and prospectively collected cases, the availability and consistency of clinical documentation were limited. For the same reason, data on empirical antimicrobial treatment at the initial visit were not available in a sufficiently complete and standardized form across all patients in the delayed-diagnosis group. Third, direct visit and clinical department likely reflected diagnostic context rather than independent causal effects, particularly because non-direct-visit patients may have arrived at participating sites with additional clinical information or a narrowed differential diagnosis. Fourth, because this study was conducted in endemic settings in Japan, the findings may not be fully generalizable to non-endemic settings or to health systems with different patterns of health care access. Fifth, ST and JSF were analyzed together, and disease-specific differences may not have been fully captured. Additionally, the relatively small number of patients in the delayed-diagnosis group (n = 61) may have limited statistical power, particularly for subgroup analyses. Although the dataset included both retrospective and prospective periods, no uniformly applied protocol for eliciting rash, eschar, or related symptoms was in place during either period. Differences in data capture intensity between the two periods therefore cannot be entirely excluded. For non-direct-visit patients, information about the specialty of the physician at the first point of care outside the participating sites was not consistently available, precluding a systematic analysis of this factor. In addition, the content of referral documentation from prior physicians was not systematically captured in the study dataset, and the potential contribution of information shared from outside physicians to the initial diagnostic impression at the participating sites could not be assessed. Future prospective studies that systematically document rash and eschar at the initial visit, regardless of clinical suspicion, would be needed to determine whether recognition of these findings at the first encounter directly improves diagnostic accuracy. Despite these limitations, this study suggests that the diagnostic process may depend not only on the presence of rash and eschar but also on how these findings come to clinical attention.

Conclusions

In endemic settings, patients with ST or JSF who were correctly diagnosed at the first visit to a participating site more often had rash and eschar brought to clinical attention in specific ways. Directed inquiry about skin symptoms and careful examination for eschar remain clinically important in routine care.

Supporting information

Acknowledgments

We acknowledge the investigators and clinical staff of the parent study, “Distinguishing Japanese Spotted Fever and Scrub Typhus, Central Japan, 2004–2015” [5], at Kameda Medical Center, Awa Regional Medical Center, and Kameda Family Clinic–Tateyama for their contributions to data collection and management. We also thank the staff of the Chiba Prefectural Institute of Public Health and other laboratory personnel involved in PCR and serological testing for rickettsial diseases in the parent study. Those acknowledged had no role in the present secondary analysis, including data analysis, interpretation, or manuscript preparation. We thank Hiromi Fujita, PhD, for valuable discussions and advice. Language editing and phrasing assistance were provided using artificial intelligence tools (ChatGPT, OpenAI) to improve grammar, clarity, and fluency. The authors reviewed and verified all AI-generated content to ensure accuracy and integrity.