Bone Abstracts

Background: Hypophosphatasia (HPP) is a rare metabolic disorder associated with persistently low serum alkaline phosphatase (ALP), with musculoskeletal and dental manifestations. Identifying adults with possible HPP is challenging because presentation can be to a number of services and the low ALP overlooked. The advent of electronic health records (EHR) offers the potential to systematize the diagnosis of rare conditions by using routinely collected data. We describe a structured EHR based phenotyping approach in a diverse community to identify adults with potential undiagnosed HPP and support targeted clinical review with escalation to a specialist MDT.

Methodology: Chelsea and Westminster Hospital NHS Trust is part of an acute provider group serving a diverse community in North West London of over 2 million people. A multi-step filtering pipeline was developed using Cerner Millennium production data and the NHS Admitted Patient Care Data Warehouse. The initial cohort comprised adults aged ≥18 years or over with at least two ALP results below 40 U/L and no readings above 100 U/L Based on existing literature inclusion and exclusion criteria were defined using a curated reference set spanning LOINC, SNOMED-CT, and ICD-10 codes, with diagnosis matching performed across both Cerner and APC/SUS sources. Patients were included if they had at least one HPP-compatible clinical feature and no alternative explanation for low ALP. To address treatment-related confounding, a temporal de-confounding algorithm evaluated consecutive ALP readings around bisphosphonate initiation to identify likely drug-induced low ALP. Socioeconomic deprivation was incorporated through postcode-to-LSOA linkage to Index of Multiple Deprivation (IMD) data.

Results: Over 2 million records were assessed over the period 2018 to 2026. A cohort of 3,707 people were identified with ≥2 readings 40 U/L. Following application of the exclusions above 256 were identified for structured clinical review. 25 patients were discussed at the specialist MDT with 9 patients subsequently undergoing specialist testing.

Conclusion: A strategy to identify potential cases of HPP using routinely collected clinical data appears feasible. The use of standardised criteria enhance the specificity and reproducibility of case identification. The next steps are clinical validation of the identified cases to ascertain the diagnostic effectiveness of this approach.