Endocrine Abstracts

Objectives: Hypophosphatasia (HPP) is a rare genetic disorder. Early diagnosis is challenging due to the disease’s complexity and low physician awareness.1 This study aimed to demonstrate how a digital health approach that scans electronic health records (EHR) may lead to earlier diagnosis of HPP.

Methods: Patients with HPP were identified in the Optimum Patient Care Research Database (OPCRD), a UK database of 13.7 million patients’ primary care EHR. Cases were identified by the presence of the SNOMED-CT diagnostic code for HPP. The clinical features that make up the diagnostic criteria for HPP were mapped to the appropriate SNOMED-CT codes.1 All EHR with a diagnostic code for HPP were examined for the presence of clinical features of HPP in advance of their diagnostic date. Descriptive statistics of HPP clinical features, categorised by organ system, were performed including EHR count and time before diagnosis. Each clinical feature was summarised by EHR count, mean and median time (months) to diagnosis.

Results: The total number of EHR with an HPP diagnostic code was 201. The total number of EHR with HPP features before diagnostic code with ages between 0 – 18 years old was 61. A total of 22 clinical features suggestive of HPP were identified and categorised into 10 organ systems. The clinical feature with the highest number of EHR unique counts was fractures (n=15), with a mean and median time to diagnosis of 280 months and 238 months, respectively. Six clinical features had an EHR count ≥5: mouth ulcer (13; [151 mean months], [49 median months]), low alkaline phosphatase (10; [52], [6]), delayed motor milestones (9; [104], [86]), rickets (7; [0], [0]), carious teeths (5; [37], [26]) and seizures (5; [180], [17]). A total of nine clinical features identified had an EHR count of <5.

Conclusions: HPP clinical features can be identified in patients’ EHR in advance of diagnosis. The identified clinical features may be used to develop phenotypical prediction tools to help identify patients at risk of HPP. Further work is needed to build an algorithm and validate these results with control comparisons.