Natural Language Processing for Clinical Laboratory Data Repository Systems: Implementation and Evaluation for Respiratory Viruses.

BACKGROUND: With the growing volume and complexity of laboratory repositories, it has become tedious to parse unstructured data into structured and tabulated formats for secondary uses such as decision support, quality assurance, and outcome analysis. However, advances in natural language processing (NLP) approaches have enabled efficient and automated extraction of clinically meaningful medical concepts from unstructured reports. OBJECTIVE: In this study, we aimed to determine the feasibility of using the NLP model for information extraction as an alternative approach to a time-consuming and operationally resource-intensive handcrafted rule-based tool. Therefore, we sought to develop and evaluate a deep learning-based NLP model to derive knowledge and extract information from text-based laboratory reports sourced from a provincial laboratory repository system. METHODS: The NLP model, a hierarchical multilabel classifier, was trained on a corpus of laboratory reports covering testing for 14 different respiratory viruses and viral subtypes. The corpus includes 87,500 unique laboratory reports annotated by 8 subject matter experts (SMEs). The classification task involved assigning the laboratory reports to labels at 2 levels: 24 fine-grained labels in level 1 and 6 coarse-grained labels in level 2. A "label" also refers to the status of a specific virus or strain being tested or detected (eg, influenza A is detected). The model's performance stability and variation were analyzed across all labels in the classification task. Additionally, the model's generalizability was evaluated internally and externally on various test sets. RESULTS: Overall, the NLP model performed well on internal, out-of-time (pre-COVID-19), and external (different laboratories) test sets with microaveraged F1-scores >94% across all classes. Higher precision and recall scores with less variability were observed for the internal and pre-COVID-19 test sets. As expected, the model's performance varied across categories and virus types due to the imbalanced nature of the corpus and sample sizes per class. There were intrinsically fewer classes of viruses being detected than those tested; therefore, the model's performance (lowest F1-score of 57%) was noticeably lower in the detected cases. CONCLUSIONS: We demonstrated that deep learning-based NLP models are promising solutions for information extraction from text-based laboratory reports. These approaches enable scalable, timely, and practical access to high-quality and encoded laboratory data if integrated into laboratory information system repositories.

Describing the linkage between administrative social assistance and health care databases in Ontario, Canada.

Background: The linkage of records across administrative databases has become a powerful tool to increase information available to undertake research and analytics in a privacy protective manner. Objective: The objective of this paper was to describe the data integration strategy used to link the Ontario Ministry of Children, Community and Social Services (MCCSS)-Social Assistance (SA) database with administrative health care data. Methods: Deterministic and probabilistic linkage methods were used to link the MCCSS-SA database (2003-2016) to the Registered Persons Database, a population registry containing data on all individuals issued a health card number in Ontario, Canada. Linkage rates were estimated, and the degree of record linkage and representativeness of the dataset were evaluated by comparing socio-demographic characteristics of linked and unlinked records. Results: There were a total of 2,736,353 unique member IDs in the MCCSS-SA database from the 1st January 2003 to 31st December 2016; 331,238 (12.1%) were unlinked (linkage rate = 87.9%). Despite 16 passes, most record linkages were obtained after 2 deterministic (76.2%) and 14 probabilistic passes (11.7%). Linked and unlinked samples were similar for most socio-demographic characteristics (i.e., sex, age, rural dwelling), except migrant status (non-migrant versus migrant) (standardized difference of 0.52). Linked and unlinked records were also different for SA program-specific characteristics, such as social assistance program, Ontario Works and Ontario Disability Support Program (standardized difference of 0.20 for each), data entry system, Service Delivery Model Technology only and both Service Delivery Model Technology and Social Assistance Management System (standardized difference of 0.53 and 0.52, respectively), and months on social assistance (standardized difference of 0.43). Conclusions: Additional techniques to account for sub-optimal linkage rates may be required to address potential biases resulting from this data linkage. Nonetheless, the linkage between administrative social assistance and health care data will provide important findings on the social determinants of health.

Validating a novel deterministic privacy-preserving record linkage between administrative & clinical data: applications in stroke research.

Introduction: Research data combined with administrative data provides a robust resource capable of answering unique research questions. However, in cases where personal health data are encrypted, due to ethics requirements or institutional restrictions, traditional methods of deterministic and probabilistic record linkages are not feasible. Instead, privacy-preserving record linkages must be used to protect patients' personal data during data linkage. Objectives: To determine the feasibility and validity of a deterministic privacy preserving data linkage protocol using homomorphically encrypted data. Methods: Feasibility was measured by the number of records that successfully matched via direct identifiers. Validity was measured by the number of records that matched with multiple indirect identifiers. The threshold for feasibility and validity were both set at 95%. The datasets shared a single, direct identifier (health card number) and multiple indirect identifiers (sex and date of birth). Direct identifiers were encrypted in both datasets and then transferred to a third-party server capable of linking the encrypted identifiers without decrypting individual records. Once linked, the study team used indirect identifiers to verify the accuracy of the linkage in the final dataset. Results: With a combination of manual and automated data transfer in a sample of 8,128 individuals, the privacy-preserving data linkage took 36 days to match to a population sample of over 3.2 million records. 99.9% of the records were successfully matched with direct identifiers, and 99.8% successfully matched with multiple indirect identifiers. We deemed the linkage both feasible and valid. Conclusions: As combining administrative and research data becomes increasingly common, it is imperative to understand options for linking data when direct linkage is not feasible. The current linkage process ensured the privacy and security of patient data and improved data quality. While the initial implementations required significant computational and human resources, increased automation keeps the requirements within feasible bounds.

Population coverage of the Canadian Chronic Disease Surveillance System: a survey of the contents of health insurance registries across Canada. / Couverture de la population dans le Système canadien de surveillance des maladies chroniques : enquête sur le contenu des registres d'assurance-maladie au Canada.

INTRODUCTION: Health insurance registries, which capture insurance coverage and demographic information for entire populations, are a critical component of population health surveillance and research when using administrative data. Lack of standardization of registry information across Canada's provinces and territories could affect the comparability of surveillance measures. We assessed the contents of health insurance registries across Canada to describe the populations covered and document registry similarities and differences. METHODS: A survey about the data and population identifiers in health insurance registries was developed by the study team and representatives from the Public Health Agency of Canada. The survey was completed by key informants from most provinces and territories and then descriptively analyzed. RESULTS: Responses were received from all provinces; partial responses were received from the Northwest Territories. Demographic information in health insurance registries, such as primary address, date of birth and sex, were captured in all jurisdictions. Data captured on familial relationships, ethnicity and socioeconomic status varied among jurisdictions, as did start and end dates of coverage and frequency of registry updates. Identifiers for specific populations, such as First Nations individuals, were captured in some, but not all jurisdictions. CONCLUSION: Health insurance registries are a rich source of information about the insured populations of the provinces and territories. However, data heterogeneity may affect who is included and excluded in population surveillance estimates produced using administrative health data. Development of a harmonized data framework could support timely and comparable population health research and surveillance results from multi-jurisdiction studies.

Linking national immigration data to provincial repositories: The case of Canada.

BACKGROUND: Canadian health data repositories link datasets at the provincial level, based on their residents' registrations to provincial health insurance plans. Linking national datasets with provincial health care registries poses several challenges that may result in misclassification and impact the estimation of linkage rates. A recent linkage of a federal immigration database in the province of Manitoba illustrates these challenges. OBJECTIVES: a) To describe the linkage of the federal Immigration, Refugees and Citizenship Canada Permanent Resident (IRCC-PR) database with the Manitoba healthcare registry and b) compare data linkage methods and rates between four Canadian provinces accounting for interprovincial mobility of immigrants. METHODS: We compared linkage rates by immigrant's province of intended destination (province vs. rest of Canada). We used external nationwide immigrant tax filing records to approximate actual settlement and obtain linkage rates corrected for interprovincial mobility. RESULTS: The immigrant linkage rates in Manitoba before and after accounting for interprovincial mobility were 84.8% and 96.1, respectively. Linkage rates did not substantially differ according to immigrants' characteristics, with a few exceptions. Observed linkage rates across the four provinces ranged from 74.0% to 86.7%. After correction for interprovincial mobility, the estimated linkage rates increased > 10 percentage points for the provinces that stratified by intended destination (British Columbia and Manitoba) and decreased up to 18 percentage points for provinces that could not use immigration records of those who did not intend to settle in the province (New Brunswick and Ontario). CONCLUSIONS: Despite variations in methodology, provincial linkage rates were relatively high. The use of a national immigration dataset for linkage to provincial repositories allows a more comprehensive linkage than that of province-specific subsets. Observed linkage rates can be biased downwards by interprovincial migration, and methods that use external data sources can contribute to assessing potential selection bias and misclassification.

Pivoting data and analytic capacity to support Ontario's COVID-19 response.

Introduction: Health care systems have faced unprecedented challenges due to the COVID-19 pandemic. Access to timely population-based data has been vital to informing public health policy and practice. Methods: We describe how ICES, an independent not-for-profit research and analytic institute in Ontario, Canada, pivoted existing research infrastructure and engaged health system stakeholders to provide near real-time population-based data and analytics to support Ontario's COVID-19 pandemic response. Results: Since April 2020, ICES provided the Ontario COVID-19 Provincial Command Table and public health partners with regular and ad hoc reports on SARS-CoV-2 testing and COVID-19 vaccine coverage. These reports: 1) helped identify congregate care/shared living settings that needed testing and prevention efforts early in the pandemic; 2) provided early indications of inequities in testing and infection in marginalized neighbourhoods, including areas with higher proportions of immigrants and visible minorities; 3) identified areas with high test positivity, which helped Public Health Units target and evaluate prevention efforts; and 4) contributed to altering the province's COVID-19 vaccine roll-out strategy to target high-risk neighbourhoods and helping Public Health Units and community organizations plan local vaccination programs. In addition, ICES is a key component of the Ontario Health Data Platform, which provides scientists with data access to conduct COVID-19 research and analyses. Discussion and Conclusion: ICES was well-positioned to provide rapid analyses for decision-makers to respond to the evolving public health emergency, and continues to contribute to Ontario's pandemic response by providing timely, relevant reports to health system stakeholders and facilitating data access for externally-funded COVID-19 research.

Validation study of health administrative data algorithms to identify individuals experiencing homelessness and estimate population prevalence of homelessness in Ontario, Canada.

OBJECTIVES: To validate case ascertainment algorithms for identifying individuals experiencing homelessness in health administrative databases between 2007 and 2014; and to estimate homelessness prevalence trends in Ontario, Canada, between 2007 and 2016. DESIGN: A population-based retrospective validation study. SETTING: Ontario, Canada, from 2007 to 2014 (validation) and 2007 to 2016 (estimation). PARTICIPANTS: Our reference standard was the known housing status of a longitudinal cohort of housed (n=137 200) and homeless or vulnerably housed (n=686) individuals. Two reference standard definitions of homelessness were adopted: the housing episode and the annual housing experience (any homelessness within a calendar year). MAIN OUTCOME MEASURES: Sensitivity, specificity, positive and negative predictive values and positive likelihood ratios of 30 case ascertainment algorithms for detecting homelessness using up to eight health service databases. RESULTS: Sensitivity estimates ranged from 10.8% to 28.9% (housing episode definition) and 18.5% to 35.6% (annual housing experience definition). Specificities exceeded 99% and positive likelihood ratios were high using both definitions. The most optimal algorithm estimates that 59 974 (95% CI 55 231 to 65 208) Ontarians (0.53% of the adult population) experienced homelessness in 2016, a 67.3% increase from 2007. CONCLUSIONS: In Ontario, case ascertainment algorithms for identifying homelessness had low sensitivity but very high specificity and positive likelihood ratio. The use of health administrative databases may offer opportunities to track individuals experiencing homelessness over time and inform efforts to improve housing and health status in this vulnerable population.

Assessing the quality of administrative data for research: a framework from the Manitoba Centre for Health Policy.

The growth of administrative data repositories worldwide has spurred the development and application of data quality frameworks to ensure that research analyses based on these data can be used to draw meaningful conclusions. However, the research literature on administrative data quality is sparse, and there is little consensus regarding which dimensions of data quality should be measured. Here we present the core dimensions of the data quality framework developed at the Manitoba Centre for Health Policy, a world leader in the use of administrative data for research purposes, and provide examples and context for the application of these dimensions to conducting data quality evaluations. In sharing this framework, our ultimate aim is to promote best practices in rigorous data quality assessment among users of administrative data for research.

Challenges in administrative data linkage for research.

Linkage of population-based administrative data is a valuable tool for combining detailed individual-level information from different sources for research. While not a substitute for classical studies based on primary data collection, analyses of linked administrative data can answer questions that require large sample sizes or detailed data on hard-to-reach populations, and generate evidence with a high level of external validity and applicability for policy making. There are unique challenges in the appropriate research use of linked administrative data, for example with respect to bias from linkage errors where records cannot be linked or are linked together incorrectly. For confidentiality and other reasons, the separation of data linkage processes and analysis of linked data is generally regarded as best practice. However, the 'black box' of data linkage can make it difficult for researchers to judge the reliability of the resulting linked data for their required purposes. This article aims to provide an overview of challenges in linking administrative data for research. We aim to increase understanding of the implications of (i) the data linkage environment and privacy preservation; (ii) the linkage process itself (including data preparation, and deterministic and probabilistic linkage methods) and (iii) linkage quality and potential bias in linked data. We draw on examples from a number of countries to illustrate a range of approaches for data linkage in different contexts.

Describing the linkages of the immigration, refugees and citizenship Canada permanent resident data and vital statistics death registry to Ontario's administrative health database.

BACKGROUND: Ontario, the most populous province in Canada, has a universal healthcare system that routinely collects health administrative data on its 13 million legal residents that is used for health research. Record linkage has become a vital tool for this research by enriching this data with the Immigration, Refugees and Citizenship Canada Permanent Resident (IRCC-PR) database and the Office of the Registrar General's Vital Statistics-Death (ORG-VSD) registry. Our objectives were to estimate linkage rates and compare characteristics of individuals in the linked versus unlinked files. METHODS: We used both deterministic and probabilistic linkage methods to link the IRCC-PR database (1985-2012) and ORG-VSD registry (1990-2012) to the Ontario's Registered Persons Database. Linkage rates were estimated and standardized differences were used to assess differences in socio-demographic and other characteristics between the linked and unlinked records. RESULTS: The overall linkage rates for the IRCC-PR database and ORG-VSD registry were 86.4 and 96.2 %, respectively. The majority (68.2 %) of the record linkages in IRCC-PR were achieved after three deterministic passes, 18.2 % were linked probabilistically, and 13.6 % were unlinked. Similarly the majority (79.8 %) of the record linkages in the ORG-VSD were linked using deterministic record linkage, 16.3 % were linked after probabilistic and manual review, and 3.9 % were unlinked. Unlinked and linked files were similar for most characteristics, such as age and marital status for IRCC-PR and sex and most causes of death for ORG-VSD. However, lower linkage rates were observed among people born in East Asia (78 %) in the IRCC-PR database and certain causes of death in the ORG-VSD registry, namely perinatal conditions (61.3 %) and congenital anomalies (81.3 %). CONCLUSIONS: The linkages of immigration and vital statistics data to existing population-based healthcare data in Ontario, Canada will enable many novel cross-sectional and longitudinal studies to be conducted. Analytic techniques to account for sub-optimal linkage rates may be required in studies of certain ethnic groups or certain causes of death among children and infants.

Healthcare utilization in patients with Parkinson's disease: a population-based analysis.

PURPOSE: To investigate factors associated with healthcare utilization and prescription drug use for Parkinson's disease (PD) patients and matched controls. METHODS: A retrospective matched-group design was adopted using administrative data from Manitoba, Canada. PD cases (N = 1469) were identified from diagnoses in hospital records and physician billing claims and matched to controls (N = 2938) on age, sex, and region of residence. Sixteen measures of healthcare utilization were examined over a six-year period using generalized linear models. RESULTS: PD cases had greater healthcare utilization than controls for almost all investigated services, with the exception of visits to non-neurological specialists and hospital use for non-mental disorder diagnoses. For controls, utilization of all forms of healthcare increased with age; for PD cases the relationship was weak, except for specialist visits, where an inverse relationship was observed. A rural region of residence was associated with a lower rate of seeing a specialist or any medical doctor, with a higher rate of hospitalization than for urban cases or controls. Comorbidity was strongly associated with healthcare use for both groups. Over the six-year study period significant differences in the trend were observed for mental disorder hospitalizations, hospital days, and physician visits. CONCLUSIONS: Factors associated with healthcare utilization in PD patients differ from those without PD. This information may help to identify and optimize healthcare services and associated costs for PD patients.

Osteoporosis-related fracture case definitions for population-based administrative data.

BACKGROUND: Population-based administrative data have been used to study osteoporosis-related fracture risk factors and outcomes, but there has been limited research about the validity of these data for ascertaining fracture cases. The objectives of this study were to: (a) compare fracture incidence estimates from administrative data with estimates from population-based clinically-validated data, and (b) test for differences in incidence estimates from multiple administrative data case definitions. METHODS: Thirty-five case definitions for incident fractures of the hip, wrist, humerus, and clinical vertebrae were constructed using diagnosis codes in hospital data and diagnosis and service codes in physician billing data from Manitoba, Canada. Clinically-validated fractures were identified from the Canadian Multicentre Osteoporosis Study (CaMos). Generalized linear models were used to test for differences in incidence estimates. RESULTS: For hip fracture, sex-specific differences were observed in the magnitude of under- and over-ascertainment of administrative data case definitions when compared with CaMos data. The length of the fracture-free period to ascertain incident cases had a variable effect on over-ascertainment across fracture sites, as did the use of imaging, fixation, or repair service codes. Case definitions based on hospital data resulted in under-ascertainment of incident clinical vertebral fractures. There were no significant differences in trend estimates for wrist, humerus, and clinical vertebral case definitions. CONCLUSIONS: The validity of administrative data for estimating fracture incidence depends on the site and features of the case definition.

Direct costs of fractures in Canada and trends 1996-2006: a population-based cost-of-illness analysis.

Cost-of-illness (COI) analysis is used to evaluate the economic burden of illness in terms of health care resource (HCR) consumption. We used the Population Health Research Data Repository for Manitoba, Canada, to identify HCR costs associated with 33,887 fracture cases (22,953 women and 10,934 men) aged 50 years and older that occurred over a 10-year period (1996-2006) and 101,661 matched control individuals (68,859 women and 32,802 men). Costs (in 2006 Canadian dollars) were estimated for the year before and after fracture, and the change (incremental cost) was modeled using quantile regression analysis to adjust for baseline covariates and to study temporal trends. The greatest total incremental costs were associated with hip fractures (median $16,171 in women and $13,111 for men), followed by spine fractures ($8,345 in women and $6,267 in men). The lowest costs were associated with wrist fractures ($663 in women and $764 in men). Costs for all fracture types were greater in older individuals (p < 0.001). Similar results were obtained with regression-based adjustment for baseline factors. Some costs showed a slight increase over the 10 years. The largest temporal increase in women was for hip fracture ($13 per year, 95% CI $6-$21, p < 0.001) and in men was for humerus fracture ($11 per year, 95% CI $3-$19, p = 0.007). At the population level, hip fractures were responsible for the largest proportion of the costs after age 80, but the other fractures were more important prior to age 80. We found that there are large incremental health care costs associated with incident fractures in Canada. Identifying COI from HCR use offers a cost baseline for measuring the effects of evidence-based guidelines implementation.

Socioeconomic variations in the prevalence and incidence of Parkinson's disease: a population-based analysis.

BACKGROUND: For many chronic conditions, lower socioeconomic status is associated with higher rates of disease. Previous research has not investigated whether this inverse relationship exists for Parkinson's disease (PD). The purpose was to investigate the association between socioeconomic status and prevalence and incidence of PD. METHODS: The study was conducted using population-based administrative data from Manitoba, Canada. PD cases were identified from diagnoses in hospital and physician records. Area-level census data on average household income and residential postal codes in health insurance registration files were used to assign PD cases to urban and rural income quintiles. Annual adjusted prevalence and incidence estimates were calculated for fiscal years 1987/88-2006/07. Hypotheses about differences between quintiles in average estimates and average rates of change were tested using generalised linear models with generalised estimating equations. RESULTS: The estimated prevalence of PD increased over the 20-year-period but incidence remained unchanged. In urban regions, average prevalence and incidence estimates were significantly higher for the lowest income quintile than the highest quintile. In rural regions, average prevalence estimates were significantly higher for the lowest quintile than for the highest quintile but incidence estimates were not significantly different. The annual rate of increase in the PD prevalence was significantly different for the lowest urban and rural income quintiles. CONCLUSIONS: There is a greater burden of PD in low-income areas. An understanding of socioeconomic inequalities is useful when formulating hypotheses about factors associated with disease onset and developing equity-oriented policies about access to healthcare resources.