Clinical Scorecard: Utilizing Information Content as a Screening Mechanism for Identifying Rare Diseases in Health Systems
At a Glance
Category
Detail
Condition
Rare Diseases (RD)
Key Mechanisms
Use of information content (IC) derived from SNOMED CT clinical terminology to screen and identify rare disease patients by detecting atypical clinical profiles in electronic health records (EHR).
Target Population
Patients within large health systems, exemplified by a longitudinal dataset of over 1.2 million patients in Singapore.
Care Setting
Hospital and health system levels utilizing electronic health records.
Key Highlights
Rare diseases affect 3.5–5.9% of the global population but often face diagnostic delays averaging 5–6 years with multiple specialist consultations.
Information content (IC) applied to SNOMED CT data can distinguish rare disease patient profiles from the first clinical encounter with ~95% sensitivity.
The method surfaced 71 underdiagnosed rare diseases in the population, most of genetic origin, demonstrating potential for early screening and identification.
Guideline-Based Recommendations
Diagnosis
Leverage SNOMED CT clinical terminology for granular documentation of patient data to improve rare disease identification.
Apply information-theoretic metrics such as information content (IC) thresholds to flag patients with atypical clinical profiles suggestive of rare diseases.
Management
Implement screening protocols using IC-based methods to prioritize patients for further diagnostic evaluation and specialist referral.
Recognize the limited treatment options available for most rare diseases and focus on timely diagnosis to enable appropriate care planning.
Monitoring & Follow-up
Monitor flagged patients longitudinally using EHR data to detect evolving clinical patterns consistent with rare diseases.
Maintain data quality and consistency in coding practices to enhance the accuracy of IC-based screening.
Risks
Potential follow-up burden due to false positives; balancing sensitivity (~95%) with precision (~20%) is necessary.
Fragmented and inconsistent clinical data across providers may limit screening effectiveness.
Patient & Prescribing Data
Rare disease patients identified through SNOMED–Orphanet mappings within a large health system dataset.
Only approximately 5% of rare diseases have approved treatments, underscoring the importance of early identification to optimize care pathways despite limited therapeutic options.
Clinical Best Practices
Utilize standardized and granular clinical terminologies like SNOMED CT integrated into EHR systems for comprehensive patient data capture.
Apply information-theoretic approaches such as information content metrics to detect atypical clinical presentations indicative of rare diseases.
Develop and implement screening thresholds that balance sensitivity and precision to manage follow-up workload effectively.
Address data fragmentation by promoting interoperability and consistent coding practices across healthcare providers.
Incorporate mappings between SNOMED CT and rare disease nomenclatures like Orphanet to enhance identification accuracy.
