Linking cohort-based data with electronic health records: a proof-of-concept methodological study in Hong Kong

Abstract

ObjectivesData linkage of cohort-based data and electronic health records (EHRs) has been practised in many countries, but in Hong Kong there is still a lack of such research. To expand the use of multisource data, we aimed to identify a feasible way of linking two cohorts with EHRs in Hong Kong.MethodsParticipants in the ‘Children of 1997’ birth cohort and the Chinese Early Development Instrument (CEDI) cohort were separated into several batches. The Hong Kong Identity Card Numbers (HKIDs) of each batch were then uploaded to the Hong Kong Clinical Data Analysis and Reporting System (CDARS) to retrieve EHRs. Within the same batch, each participant has a unique combination of date of birth and sex which can then be used for exact matching, as no HKID will be returned from CDARS. Raw data collected for the two cohorts were checked for the mismatched cases. After the matching, we conducted a simple descriptive analysis of attention deficit hyperactivity disorder (ADHD) information collected in the CEDI cohort via the Strengths and Weaknesses of ADHD Symptoms and Normal Behaviour Scale (SWAN) and EHRs.ResultsIn total, 3473 and 910 HKIDs in the birth cohort and CEDI cohort were separated into 44 and 5 batches, respectively, and then submitted to the CDARS, with 100% and 97% being valid HKIDs respectively. The match rates were confirmed to be 100% and 99.75% after checking the cohort data. From our illustration using the ADHD information in the CEDI cohort, 36 (4.47%) individuals had ADHD–Combined score over the clinical cut-off in the SWAN survey, and 68 (8.31%) individuals had ADHD records in EHRs.ConclusionsUsing date of birth and sex as identifiable variables, we were able to link the cohort data and EHRs with high match rates. This method will assist in the generation of databases for future multidisciplinary research using both cohort data and EHRs.

Improving Cohort-Hospital Matching Accuracy through Standardization and Validation of Participant Identifiable Information

“…For our consented cohort using only demographic summary variables, we achieved high matching rates to birthing hospitals records using efficient, secure methods. These matching rates were comparable to studies using combinations of unique identifiers (lacking in our dataset) and the child’s and mother’s demographic information, e.g., the studies from UK [ 14 ] and Hong Kong [ 10 ]. We have found no other reported studies that have achieved this.…”

“…Firstly, such cohorts typically collect molecular, phenotypic, and participant-reported data that administrative datasets lack. Secondly, hospitals clinical data may be especially important for the pregnancy and newborn periods, as (1) these are the only life points at which hospital care is essentially universal in many countries [ 3 , 6 , 8 , 9 , 10 ]; (2) there is substantial variation in care [ 11 ]; and (3) these periods shape subsequent outcomes for the entire maternal and child life course [ 12 ]. By accessing deep clinical data, consented cohorts could fill many gaps in both predicting health care and understanding its impacts on health and wellbeing [ 13 ].…”

“…High matching rates between cohorts and hospital records have been achieved when unique identifiers are available, for example true matching rates above 97% for a UK study [ 14 ] and nearly 100% for a Hong Kong study [ 10 ]. Unfortunately, many consented cohorts do not hold hospital unique identifiers; participants may not know them, and it may not be easily possible to obtain them from services due to privacy and legal considerations [ 4 , 5 ].…”

Validation of Diagnostic Coding for Asthma in an Electronic Health Record System in Hong Kong

“…19 The CDARS captures medical information including diagnosis, drug prescription details, demographics, admissions, medical procedures, and laboratory results. Despite the fact that the diagnostic coding was shown to be highly accurate in some diagnoses in local studies, [20][21][22] the accuracy of diagnostic code for asthma has not been validated. In the current study, we aimed to assess the validity of International Classification of Diseases, 9th Revision (ICD-9) code algorithms for identifying asthma in CDARS.…”

The role of cytology in densely populated territories: An experience from Hong Kong

“…In terms of data extraction, the Clinical Data Analysis and Reporting System allows for territory‐wide, systematic exporting of clinical and laboratory data, 4 and all pathology departments within the Hospital Authority department use a uniform electronic laboratory information system. Cytology services are integral to all anatomical pathology units in the public sector of Hong Kong, and they share the same laboratory information system with surgical specimens.…”