Validation of de-identified record linkage to ascertain hospital admissions in a cohort study

Sex differences in risk of cardiovascular events and mortality with sodium glucose co-transporter-2 inhibitors versus glucagon-like peptide 1 receptor agonists in Australians with type 2 diabetes: a population-based cohort study

Background

Sodium glucose co-transporter-2 inhibitors (SGLT2i) and glucagon-like peptide 1 receptor agonists (GLP-1RAs) reduce major adverse cardiovascular events (MACE) in people with type 2 diabetes (T2D). Despite known sex differences in diabetes-induced cardiovascular disease (CVD), pharmacological treatment recommendations are independent of sex. Our objective was to investigate possible sex differences in rates of MACE with SGLT2i vs. GLP-1RA use.

Methods

This population-based cohort study included men and women with T2D (≥30 years), discharged from a Victorian hospital between 1st July 2013 and 1st July 2017, and dispensed an SGLT2i or GLP-1RA within 60 days of discharge. Using Cox proportional hazards regression with competing risks, subdistribution hazard ratios (sHR) with 95% confidence intervals (CI) were estimated for MACE in a follow-up to 30th June 2018. Analyses were conducted for men and women, and subgroups based on age, baseline heart failure (HF), and atherosclerotic CVD (ASCVD) status.

Findings

From a total of 8026 people (44.3% women, median follow-up time = 756 days), SGLT2i (n = 4231), compared to GLP-1RAs (n = 3795), reduced MACE rates in men (sHR 0.78; 95%CI 0.66-0.93), but not women. SGLT2i reduced MACE rates in men (sHR 0.72; 95%CI 0.54-0.98) and women (sHR 0.52; 95%CI 0.31-0.86) ≥65 years; in men with baseline HF (sHR 0.45; 95%CI 0.28-0.73); and in women with ASCVD (sHR 0.36; 95%CI 0.18-0.71).

Interpretations

SGLT2i, relative to GLP-1RAs, demonstrate favourable effects for MACE reductions among older Australian men and women with T2D. Analogous benefits were also observed in men with HF and women with ASCVD.

Funding

Dementia Australia Yulgilbar Innovation Award.

Assessing the impact of privacy-preserving record linkage on record overlap and patient demographic and clinical characteristics in PCORnet®, the National Patient-Centered Clinical Research Network

OBJECTIVE

This article describes the implementation of a privacy-preserving record linkage (PPRL) solution across PCORnet®, the National Patient-Centered Clinical Research Network.

MATERIAL AND METHODS

Using a PPRL solution from Datavant, we quantified the degree of patient overlap across the network and report a de-duplicated analysis of the demographic and clinical characteristics of the PCORnet population.

RESULTS

There were ∼170M patient records across the responding Network Partners, with ∼138M (81%) of those corresponding to a unique patient. 82.1% of patients were found in a single partner and 14.7% were in 2. The percentage overlap between Partners ranged between 0% and 80% with a median of 0%. Linking patients' electronic health records with claims increased disease prevalence in every clinical characteristic, ranging between 63% and 173%.

DISCUSSION

The overlap between Partners was variable and depended on timeframe. However, patient data linkage changed the prevalence profile of the PCORnet patient population.

CONCLUSIONS

This project was one of the largest linkage efforts of its kind and demonstrates the potential value of record linkage. Linkage between Partners may be most useful in cases where there is geographic proximity between Partners, an expectation that potential linkage Partners will be able to fill gaps in data, or a longer study timeframe.

Lessons in Linkage: combining administrative data using deterministic linkage for surveillance of sports and recreation injuries in Florida, United States

Objectives

Previous and ongoing epidemiological surveillance of sports and recreation injuries (SRI) has been cross-sectional in nature, utilised a subset of injuries based on athletic trainer availability, or focused on elite and professional athletes. In the United States, surveillance is often prohibitively expensive and not well funded by national organisations or agencies, except for the case of some professional and elite sports. This paper details the methodology, barriers, and successes of using deterministic linkage to combine emergency department and hospitalisation data with a single identifier for use in surveilling sports injuries for persons aged 5 to 18 years.

Design

Data linkage of a population cohort.

Methods

We performed deterministic linkage of administrative emergency department and hospitalisation data from the state of Florida in the US. Data was acquired from the Florida Agency for Health Care Administration. With limited identifiers available due to privacy, we combined data across multiple years using a near universal identifier. We identified sport and recreation injuries using a modified External Cause of Injury Morbidity Matrix and ICD codes across all possible diagnoses. Finally, we obtained descriptive statistics of records that were successfully linked and those that were not to assess similarities between the groups.

Results

We found 384,731 visits for SRI over a seven-year period. We were able to link approximately 70% of the records using a single identifier. There were statistically significant differences by age, sex, payer, and race/ethnicity for the records that were linked compared to the records that were not linked.

Conclusions

This study is significant because while similar methods have been used to examine other conditions (e.g. asthma), few have linked multiple types of administrative data especially with nearly no identifiers to examine sports and recreation injuries. This method was found useful to identify injuries over time for the same individuals seeking care in emergency departments, or in hospital inpatient settings, though future work will need to address the limitations of this method. If we expect to move health surveillance forward as budgets for it become even more limited, we must develop and improve methods to do it with fewer resources, including using data that has great limitations.

Investigating first-year graduate paramedics' reason for current work location: A cross-sectional, data linkage study

OBJECTIVE

This study aims to describe the demographic and employment characteristics of first-year graduates from a Victorian-based paramedicine course and investigate factors that influenced their choice in place of practice.

DESIGN

Cross-sectional study using data from the Nursing and Allied Health Graduate Outcomes Tracking study.

SETTING

Victoria, Australia.

PARTICIPANTS

First-year graduates (2019) from the Monash University range of paramedicine programs.

MAIN OUTCOME MEASURES

Variables of interest included principal place of practice and the reasons for working in the current location.

RESULTS

Over half of the 2018 paramedicine course graduates responded to the 2019 Graduate Outcomes Survey. Nearly all were registered as paramedics (including double registrants as nurses), and over a fifth were from a rural background; however, less than that were working in a rural area. Of those with complete data, the most cited reasons for current work location were 'spouse/partner's employment or career', 'opportunity for career advancement' and 'scope of practice within the role'.

CONCLUSION

This study provides important insight into the factors associated with rural practice location amongst paramedicine graduates, specifically rural origin or personal, lifestyle and professional influences. The study adds to the sparse literature about paramedic practice location decision-making and highlights the need for further systematic longitudinal research examining the 'where' and 'why'.

Caustic ingestions over 10 years in Victoria, Australia: High rates in migrants and women

OBJECTIVES

Caustic ingestion is relatively common in developing countries and can result in life-threatening sequelae. There is limited understanding of the epidemiology and incidence in Australia. This statewide 10 year audit aims to investigate this further.

METHOD

A retrospective review was conducted over ten years (2007-2016), including all admissions to hospitals in Victoria. This includes a population of 5.9 million people and 22 hospitals.

RESULTS

384 cases of caustic ingestion were admitted to hospital between January 2007 and December 2016. The overall incidence was 7 cases/million/year. This cohort included 217 females (56.5%), 193 overseas born patients (50.2%) and 196 people (51%) with a history of mental illness. The countries of birth with the highest incidence of caustic ingestion were Ethiopia, 11 patients (227 cases/million/year, RR 31.7, p < 0.0001), Sudan, 11 patients (161 cases/million/year, RR 22.6, p < 0.0001), India, 38 patients (27 cases/million/year, RR 3.9, p < 0.0001). All had a significantly higher incidence than the Australian-born population of only 6.5 cases/million/year (RR 0.4, p < 0.0001). Of those born in India, Sudan and Ethiopia, rates of females were considerably higher than males (72% females). The overall mortality rate in this cohort was 2.3%.

CONCLUSIONS

Caustic ingestion remains a significant cause of morbidity and health expenditure in Victoria, particularly amongst vulnerable groups such as recent female migrants from areas in Africa and India. The high frequency of events seen in migrant populations highlights the significant need for awareness of risks in these groups for development of possible prevention strategies which are required. This article is protected by copyright. All rights reserved.

Validating linkage of multiple population-based administrative databases in Brazil

Background

Linking routinely-collected data provides an opportunity to measure the effects of exposures that occur before birth on maternal, fetal and infant outcomes. High quality linkage is a prerequisite for producing reliable results, and there are specific challenges in mother-baby linkage. Using population-based administrative databases from Brazil, this study aimed to estimate the accuracy of linkage between maternal deaths and birth outcomes and dengue notifications, and to identify potential sources of bias when assessing the risk of maternal death due to dengue in pregnancy.

Methods

We identified women with dengue during pregnancy in a previously linked dataset of dengue notifications in women who had experienced a live birth or stillbirth during 2007–2012. We then linked this dataset with maternal death records probabilistically using maternal name, age and municipality. We estimated the accuracy of the linkage, and examined the characteristics of false-matches and missed-matches to identify any sources of bias.

Results

Of the 10,259 maternal deaths recorded in 2007–2012, 6717 were linked: 5444 to a live birth record, 1306 to a stillbirth record, and 33 to both a live and stillbirth record. After identifying 2620 missed-matches and 124 false-matches, our estimated sensitivity was 72%, specificity was 88%, and positive predictive value was 98%. Linkage errors were associated with maternal education and self-identified race; women with more than 7 years of education or who self-declared as Caucasian were more likely to link. Dengue status was not associated with linkage error.

Conclusion

Despite not having unique identifiers to link mothers and birth outcomes, we demonstrated a high standard of linkage, with sensitivity and specificity values comparable to previous literature. Although there were no differences in the characteristics of dengue cases missed or included in our linked dataset, linkage error occurred disproportionally by some social-demographic characteristics, which should be taken into account in future analyses.

Prevalence of multiple sclerosis in Sardinia: A systematic cross-sectional multi-source survey

Background:

Partial surveys in sub-regions of Sardinia have suggested a high prevalence of multiple sclerosis (MS) on the island, relative to other Mediterranean populations. We assessed the island-wide prevalence of MS and its detailed distribution in Sardinia.

Methods:

The study population consisted of 5677 MS patients, 1735 men and 3942 women, living in Sardinia. Neurologists retrospectively examined electronic and paper-based records of patients with a diagnosis of MS. The data were then linked to the administrative health information systems. Crude, age-, and sex-specific prevalence estimates of disease were calculated.

Results:

The overall age-adjusted MS prevalence was 330 per 100,000 (95% confidence interval (CI) 321–338) in individuals older than 15 years, 447 in women (95% CI 433–461), and 205 in men (95% CI 195–214). The prevalence was highest in the Ogliastra and Nuoro districts, respectively, 425 (95% CI 372–478) and 419 (95% CI 387–451), and lowest in the Olbia-Tempio district, 217 (95% CI 195–239). Most cases had relapsing–remitting MS (79.3%), 16.3% had secondary-progressive MS, and 4.4% had primary-progressive MS.

Conclusion:

These prevalence are among the highest reported so far worldwide. They provide estimates for comparative analyses in other populations and are essential for public health interventions.

A before-after study of hospital use in two frail populations receiving different home-based services over the same time in Vancouver, Canada

Background

As individuals age, they are more likely to experience increasing frailty and more frequent use of hospital services. First, we explored whether initiating home-based primary care in a frail homebound cohort, influenced hospital use. Second, we explored whether initiating regular home care support for personal care with usual primary care, in a second somewhat less frail cohort, influenced hospital use.

Methods

This was a before-after retrospective cohort study of two frail populations in Vancouver, Canada using administrative data to assess the influence of two different services started in two different cohorts over the same time period. The participants were 246 recipients of integrated home-based primary care and 492 recipients of home care followed between July 1st, 2008 and June 30th, 2013 before and after starting their respective services. Individuals in each group were linked to their hospital emergency department visit and discharge abstract records. The main outcome measures were mean emergency department visit and hospital admission rates per 1000 patient days for 21 months before versus the period after receipt of services, and the adjusted incidence rate ratios (IRRs) on these outcomes post receipt of service.

Results

Before versus after starting integrated home-based primary care, emergency department visit rates per 1000 patient days (95% confidence intervals) were 4.1 (3.8, 4.4) versus 3.7 (3.3, 4.1), and hospital admissions rates were 2.3 (2.1, 2.5) versus 2.2 (1.9, 2.5). Before versus after starting home care, emergency department visit rates per 1000 patient days (95% confidence intervals) were 3.0 (2.8, 3.2) versus 4.0 (3.7, 4.3) visits and hospital admissions rates were 1.3 (1.2, 1.4) versus 1.9 (1.7, 2.1). Home-based primary care IRRs were 0.91 (0.72, 1.15) and 0.99 (0.76, 1.27) and home care IRRs were 1.34 (1.15, 1.56) and 1.46 (1.22, 1.74) for emergency department visits and hospital admissions respectively.

Conclusions

After enrollment in integrated home-based primary care, emergency department visit and hospital admission rates stabilized. After starting home care with usual primary care, emergency department visit and hospital admission rates continued to rise.

Following Up Crack Users after Hospital Discharge Using Record Linkage Methodology: An Alternative to Find Hidden Populations

This paper presents the probabilistic record linkage (PRL) methodology as an alternative way to find or follow up hard-to-reach population as crack users. PRL was based on secondary data from public health information systems and the strategy used from standardization; phonetic encoding and the rounds of matching data were described. A total of 293 patient records from medical database and two administrative datasets obtained from Ministry of Health Information Systems were used. Patient information from the medical database was the identifiers to the administrative datasets containing data on outpatient treatment and hospital admissions. 40% of patient records were found in the hospital database and 12% were found in the outpatient database; 95% of the patients were hospitalized up to 5 times, and only 10 out of them had outpatient information. The record linkage methodology by linking government databases may help to address research questions about the path of patients in the care network without spending time and financial resources with primary data collection.

Examining the quality of name code record linkage: what is the impact on death and cancer risk estimates? A validation study

OBJECTIVE

To examine the validity and impact of record linkage using name code compared to full name records.

METHODS

A registry of 45,419 opioid substitution clients (1985-2007) was linked with national population-based death and cancer registries using registrant's name, date of birth, sex, state, postcode and date of death. Records were linked using full name and then using the first two letters of the given and surname (2×2 name code). Sensitivity and specificity were computed and regression analysis used to identify factors related to linkage accuracy. Standardised mortality ratios (SMR) and standardised cancer incidence ratios (SIR) were estimated.

RESULTS

The sensitivity and specificity of name code compared to full name linkage were 65.31% and 99.91% for death records and 76.81% and 99.89% for cancer records. Registrants' age and sex and accuracy of the registries were associated with risk of false linkages. Death and cancer risks (SMR 6.98, 95%CI 6.77-7.19; SIR 1.16, 95%CI 1.08-1.24) were significantly under-estimated using name code linkage (SMR 4.39, 95%CI 4.23-4.56; SIR 0.92, 95%CI 0.85-0.99).

CONCLUSION

Record linkage using 2×2 name code has low sensitivity but high specificity, resulting in conservative estimates of death and cancer risk. This may translate to meaningful differences in outcomes.

The promise of record linkage for assessing the uptake of health services in resource constrained settings: a pilot study from South Africa

Background

Health and Demographic Surveillance Systems (HDSS) have been instrumental in advancing population and health research in low- and middle- income countries where vital registration systems are often weak. However, the utility of HDSS would be enhanced if their databases could be linked with those of local health facilities. We assess the feasibility of record linkage in rural South Africa using data from the Agincourt HDSS and a local health facility.

Methods

Using a gold standard dataset of 623 record pairs matched by means of fingerprints, we evaluate twenty record linkage scenarios (involving different identifiers, string comparison techniques and with and without clerical review) based on the Fellegi-Sunter probabilistic record linkage model. Matching rates and quality are measured by their sensitivity and positive predictive value (PPV). Background characteristics of matched and unmatched cases are compared to assess systematic bias in the resulting record-linked dataset.

Results

A hybrid approach of deterministic followed by probabilistic record linkage, and scenarios that use an extended set of identifiers including another household member’s first name yield the best results. The best fully automated record linkage scenario has a sensitivity of 83.6% and PPV of 95.1%. The sensitivity and PPV increase to 84.3% and 96.9%, respectively, when clerical review is undertaken on 10% of the record pairs. The likelihood of being linked is significantly lower for females, non-South Africans and the elderly.

Conclusion

Using records matched by means of fingerprints as the gold standard, we have demonstrated the feasibility of fully automated probabilistic record linkage using identifiers that are routinely collected in health facilities in South Africa. Our study also shows that matching statistics can be improved if other identifiers (e.g., another household member’s first name) are added to the set of matching variables, and, to a lesser extent, with clerical review. Matching success is, however, correlated with background characteristics that are indicative of the instability of personal attributes over time (e.g., surname in the case of women) or with misreporting (e.g., age).

The false security of blind dates: chrononymization's lack of impact on data privacy of laboratory data

BACKGROUND

The reuse of clinical data for research purposes requires methods for the protection of personal privacy. One general approach is the removal of personal identifiers from the data. A frequent part of this anonymization process is the removal of times and dates, which we refer to as "chrononymization." While this step can make the association with identified data (such as public information or a small sample of patient information) more difficult, it comes at a cost to the usefulness of the data for research.

OBJECTIVES

We sought to determine whether removal of dates from common laboratory test panels offers any advantage in protecting such data from re-identification.

METHODS

We obtained a set of results for 5.9 million laboratory panels from the National Institutes of Health's (NIH) Biomedical Translational Research Information System (BTRIS), selected a random set of 20,000 panels from the larger source sets, and then identified all matches between the sets.

RESULTS

We found that while removal of dates could hinder the re-identification of a single test result, such removal had almost no effect when entire panels were used.

CONCLUSIONS

Our results suggest that reliance on chrononymization provides a false sense of security for the protection of laboratory test results. As a result of this study, the NIH has chosen to rely on policy solutions, such as strong data use agreements, rather than removal of dates when reusing clinical data for research purposes.