Key result
Clinical prediction models for scrub typhus in pediatric AES demonstrate strong performance with 0.83 c-statistic.
Why the study?
Scrub typhus and other doxycycline-treatable infections are significant contributors to AES in India, but limited surveillance in South India has hindered their recognition and inclusion in treatment protocols.
Can clinical prediction models accurately diagnose scrub typhus and identify doxycycline-treatable causes in children with acute encephalitis syndrome?
Observational (n=587)
Yes
Can clinical prediction models accurately diagnose scrub typhus and identify doxycycline-treatable causes in children with acute encephalitis syndrome?
Clinical prediction models using simple clinical and laboratory parameters can accurately identify scrub typhus and doxycycline-treatable causes in children with acute encephalitis syndrome in South India.
Answers are grounded in this paper and cite their sources.
May aid targeted doxycycline decisions in endemic paediatric AES; leaves open external validation and outcome trials.
Background: Scrub typhus and other doxycycline-treatable infections are significant contributors of acute encephalitis syndrome (AES) in India. Limited surveillance in South India has hindered their recognition and the inclusion of doxycycline in treatment protocols. We aimed to systematically investigate infectious aetiologies of AES in children from Karnataka, India, and develop clinical prediction models for diagnosing scrub typhus and guiding clinical decisions for doxycycline therapy. Methods: This multicentre, prospective study enrolled children aged >28 days to 18 years with AES presenting to three tertiary care hospitals in Bengaluru, India. Primary outcomes were microbiological diagnosis of AES and clinical prediction models for diagnosing scrub typhus and identifying patients with doxycycline-treatable causes. Models were developed using multivariable logistic regression, internally validated, and simplified into point-scoring systems. Model performance was evaluated using c-statistics, calibration slopes, and calibration-in-the-large, adhering to TRIPOD guidelines. Findings: Between February 2020 and February 2023, 714 children were screened, of whom 587 were included. Of these, 315 (54%) had a microbiological diagnosis. Scrub typhus accounted for 138/315 (44%), and doxycycline-treatable causes were diagnosed in 193/315 (61%) of these cases. Key predictors associated with both scrub typhus and doxycycline-treatable causes were age, illness duration, lymphadenopathy, oedema, hepatomegaly, lymphocyte count, platelet count, and serum albumin levels. Adjusted c-statistics were 0.83 (95% CI: 0.78-0.87) for the scrub typhus model and 0.75 (95% CI: 0.7-0.81) for the doxycycline model, with calibration slopes of 0.85 (0.82-0.88) and 0.83 (0.78-0.87), respectively. CITL values were -0.03 (-0.06-0) and 0.05 (0.02-0.09). Points-based scores predicted probabilities ranging from 5% to 99.8% (scrub typhus model) and 20%-99% (doxycycline-treatable model). Interpretation: Scrub typhus was the most common microbiological diagnosis, and most patients had a doxycycline-treatable cause, underscoring the need to prioritise doxycycline in empirical treatment protocols in South India. The models demonstrated strong performance; however external validation is necessary for broader applicability.
No takes yet. Share an insight, caveat, or question.
Damodar et al. (2025) conducted an observational in paediatric acute encephalitis syndrome (n=587). Clinical prediction models was evaluated on microbiological diagnosis of AES and clinical prediction models for diagnosing scrub typhus and identifying patients with doxycycline-treatable causes (c-statistic 0.83, 95% CI 0.78-0.87). Clinical prediction models for scrub typhus in paediatric acute encephalitis syndrome demonstrated strong performance with an adjusted c-statistic of 0.83.
Synapse has enriched 10 related papers. Consider them for comparative context: