Deterministic Matching Research Articles

As our population changes, so must we: how race and ethnicity data merging has evolved over time in the Wisconsin Administrative Data Core

The Institute for Research on Poverty (IRP), centered at the University of Wisconsin-Madison in Wisconsin, USA, began collaboration with various Wisconsin state agencies in 1985 to link data between disparate state agencies. These initial efforts have since evolved into the Wisconsin Administrative Data Core (WADC) which is an individual-level population dataset connecting extensive administrative data spanning decades of Wisconsin social services program history. Since its purpose is to facilitate evidence-driven policymaking and support program evaluation, it is essential that our data linking efforts accurately represent our Wisconsin population. Since its inception, the WADC has adapted to changes in how race and ethnicity are collected and reported. While Wisconsin has a racial makeup that is generally less diverse than the national population, the number of non-white and multi-racial identifying residents has increased over time. Furthermore, varying state agency demographic collection and reporting methods have fueled the need for our linking methods to change, too. IRP initially implemented standard data linkage methods using a mixture of probabilistic and deterministic matching of personal identifying information. But as the racial composition of Wisconsin residents changed over time, our linking assumptions induced implicit biases. In response, we updated our matching algorithm to account for the cultural nuances in how personally identifying information is assigned and reported within various racial and ethnic subgroups. Furthermore, we prioritized values from data sources that were self-reported rather than assigned from a third-party. These methodological modifications have improved the WADC’s reliability and representativeness of the population.

Pregnant People-Infant Linkage for Surveillance

Objectives and ApproachAs maternal mortality and morbidity continue to progress, better data is needed to comprehensively understand the problem, especially in underserved populations who are differentially impacted. Surveillance in safety-net populations is complicated by care fragmentation and interoperability challenges between primary care and hospital settings. . AllianceChicago (AC), a health center-controlled network, supports a common data infrastructure across a set of Federally Qualified Health Centers (FQHCs) in Chicago that is further linked to a distributed query infrastructure connecting Chicago’s major health systems known as CAPriCORN. Leveraging electronic health record (EHR) data collected by the AC and CAPriCORN consortium, the objective of this surveillance effort was to connect infant data with maternal data while protecting privacy. We developed a deterministic matching algorithm that utilized multiple approaches and technologies for Pregnant People-Infant (PPI) pair identification. The linkage process involved a five-step method in which each step is recorded within the linkage database and the confidence interval is calculated. ResultsPreliminary analyses are ongoing. Through each step of the linkage process, the confidence interval increased with the increase of matching nodes, highlighting its accumulative accuracy. The result demonstrates the feasibility of linking maternal and baby healthcare characteristics using a range of clinical and demographic variables captured in pseudonymized hospital and ambulatory data. Conclusions and ImplicationsOur approach improves the capacity of EHR data for surveillance while protecting patient privacy and preserving data confidentiality. Triangulating outcomes recorded in different settings can help improve data quality for research and surveillance.

Longitudinal Follow-up of Medicare Patients After Esophageal Cancer Resection in the STS Database

BackgroundUnderstanding characteristics associated with survival after esophagectomy for cancer is critical to preoperative risk stratification. This study sought to define predictors for long-term survival after esophagectomy for cancer in Medicare patients. MethodsThe Society of Thoracic Surgeons General Thoracic Surgery Database was queried for patients aged ≥65 years who underwent esophagectomy for cancer between 2012 and 2020 and linked to Centers for Medicare and Medicaid Services (CMS) data using a deterministic matching algorithm. Patient, hospital, and treatment variables were assessed using a multivariable Cox proportional hazards model to evaluate characteristics associated with long-term mortality and readmission. Kaplan-Meier and cumulative incidence curves were generated and differences evaluated using the log-rank test and Gray’s test, respectively. ResultsAfter CMS linkage, 4798 patients were included. Thirty-day and 90-day mortality in the study group was 3.84% and 7.45%, respectively. In the multivariable model, American Society of Anesthesiologists score >3, body mass index >35, and diabetes were associated with increased mortality <90 days post-surgery, while pN/pT upstaging was associated with increased mortality >90 days post-surgery. Patients upstaged to pN(+) had a 147% increased mortality risk (adjusted hazard ratio [aHR], 2.47; 95% CI, 2.02-3.02) and those that remained pN(+) a 75% increased mortality risk (aHR, 1.75; 95% CI, 1.57-1.95) compared with downstaged patients. Patients who were pT upstaged had a 109% (aHR, 2.09; 95% CI, 1.73-2.53) increased mortality risk compared with pT downstaged patients. Risk for readmission was independent of procedure type or approach and was higher in c stage ≥2, American Society of Anesthesiologists score ≥4, and pN+. ConclusionsMedicare patients undergoing esophagectomy for cancer have identifiable patient-specific predictors for short-term mortality and tumor-specific predictors for long-term mortality and readmission. In the absence of pathologic T and N downstaging, risk for long-term mortality and readmission are increased.

Exploration of Skillification and Its Use in Awarding Credit for Prior Learning

In applications for prior learning credit, learners with professional experience describe competencies gained during their employment history or past study, explaining how these competencies correspond to what is taught in a course sufficiently well to award them credit for prior learning. Comparing skills that are extracted algorithmically from artifacts describing a learner’s prior experience to skills similarly extracted from syllabi offers a potential solution. Current research indicates that algorithmically extracted skills can be used to identify relevance between artifacts and groups of courses but that connecting an artifact to a specific course requires guidance to learners on the artifact creation, careful use of analytical techniques, and faculty involvement. The guidance includes setting expectations for language volume and creating focused text, like an essay, rather than more general descriptive artifacts, like résumés. Effective analytical techniques accommodate observed variation in skill volume between different artifacts, optimizing the potential identification of appropriate relationships. Basing decisions on the degree of overlap in the total artifact-course skill set appears more effective than considering the number of matches to program skills. Deterministic matching is insufficient for identifying course-level correspondence, while approaches that capture information about the covariation of skills among courses offer promise.

Adwords with Unknown Budgets and Beyond

In the classic Adwords problem introduced by [Mehta A, Saberi A, Vazirani U, Vazirani V (2007) Adwords and generalized online matching. J. ACM 54(5):22-es.], we have a bipartite graph between advertisers and queries. Each advertiser has a maximum budget that is known a priori. Queries are unknown a priori and arrive sequentially. When a query arrives, advertisers make bids, and we (immediately and irrevocably) decide which (if any) Ad to display based on the bids and advertiser budgets. The winning advertiser for each query pays their bid up to their remaining budget. Our goal is to maximize total budget used without any foreknowledge of the arrival sequence (which could be adversarial). We consider the setting where the online algorithm does not know the advertisers’ budgets a priori and the budget of an advertiser is revealed to the algorithm only when it is exceeded. A naïve greedy algorithm is 0.5 competitive for this setting, and finding an algorithm with better performance remained an open problem. We show that no deterministic algorithm has competitive ratio better than 0.5 and give the first (randomized) algorithm with strictly better performance guarantee. We show that the competitive ratio of our algorithm is at least 0.522 but also strictly less than [Formula: see text]. We present novel applications of budget oblivious algorithms in search ads and beyond. In particular, we show that our algorithm achieves the best possible performance guarantee for deterministic online matching in the presence of multichannel traffic [Manshadi V, Rodilitz S, Saban D, Suresh A (2022) Online algorithms for matching platforms with multi-channel traffic. Proc. 23rd ACMConf. Econom. Comput. (ACM, NewYork), 986–987.]. This paper was accepted by Omar Besbes, revenue management and market analytics. Funding: NSF Division of Civil, Mechanical, and Manufacturing Innovation [Grant 2340306], and Google Research Scholar Program. Supplemental Material: The online appendices are available at https://doi.org/10.1287/mnsc.2021.03243 .

Pediatric Primary Care Clinicians' Perspectives on Telemedicine Use, 2020 Versus 2021.

Objective: We examined the change in pediatric primary care clinician attitudes and perceptions about telemedicine after one year of telemedicine use. Methods: We administered a survey to pediatric primary care clinicians across 50 primary care practices in Pennsylvania in 2020 and 2021. Surveys were linked using a combination of deterministic and probabilistic matching. We used McNemar's test to compare change in responses from 2020 to 2021. Results: Among pediatric primary care clinicians surveyed in 2020 and 2021 (n = 101), clinicians agreed that telemedicine could always or usually deliver high-quality care for mental health (80% in 2020 and 78% in 2021), care coordination (77% in 2020 and 70% in 2021), acute care (33% in 2020 and 34% in 2021), or preventive care (25% in 2020 and 18% in 2021) and this did not significantly change. Clinician perceptions of usability, while high, declined over time with fewer endorsing ease of use (93% in 2020 and 80% in 2021) and reliability (14% in 2020 and 0% in 2021) over time. Despite this, 62% of clinicians agreed that they were satisfied with their use of telemedicine at both time points. Respondents anticipated positive impact on equity and timeliness of care from telemedicine use but did not anticipate positive impact across child health, health care delivery, or clinician experience. Perceptions across these domains did not change over time. Conclusions: With one year of telemedicine experience, primary care clinicians maintained beliefs that telemedicine could deliver high-quality care for specific clinical needs but had worsening perceptions of usability over time.

Improving our understanding of the social determinants of mental health: a data linkage study of mental health records and the 2011 UK census

ObjectivesTo address the lack of individual-level socioeconomic information in electronic healthcare records, we linked the 2011 census of England and Wales to patient records from a large mental healthcare provider....

Linking Trauma Registry Patients With Insurance Claims: Creating a Longitudinal Patient Record

IntroductionTrauma registries and their quality improvement programs only collect data from the acute hospital admission, and no additional information is captured once the patient is discharged. This lack of long-term data limits these programs’ ability to affect change. The goal of this study was to create a longitudinal patient record by linking trauma registry data with third party payer claims data to allow the tracking of these patients after discharge. MethodsTrauma quality collaborative data (2018-2019) was utilized. Inclusion criteria were patients age ≥18, ISS ≥5 and a length of stay ≥1 d. In-hospital deaths were excluded. A deterministic match was performed with insurance claims records based on the hospital name, date of birth, sex, and dates of service (±1 d). The effect of payer type, ZIP code, International Classification of Diseases, Tenth Revision, Clinical Modification diagnosis specificity and exact dates of service on the match rate was analyzed. ResultsThe overall match rate between these two patient record sources was 27.5%. There was a significantly higher match rate (42.8% versus 6.1%, P < 0.001) for patients with a payer that was contained in the insurance collaborative. In a subanalysis, exact dates of service did not substantially affect this match rate; however, specific International Classification of Diseases, Tenth Revision, Clinical Modification codes (i.e., all 7 characters) reduced this rate by almost half. ConclusionsWe demonstrated the successful linkage of patient records in a trauma registry with their insurance claims. This will allow us to the collect longitudinal information so that we can follow these patients’ long-term outcomes and subsequently improve their care.

De-identified Bayesian personal identity matching for privacy-preserving record linkage despite errors: development and validation

BackgroundEpidemiological research may require linkage of information from multiple organizations. This can bring two problems: (1) the information governance desirability of linkage without sharing direct identifiers, and (2) a requirement to link databases without a common person-unique identifier.MethodsWe develop a Bayesian matching technique to solve both. We provide an open-source software implementation capable of de-identified probabilistic matching despite discrepancies, via fuzzy representations and complete mismatches, plus de-identified deterministic matching if required. We validate the technique by testing linkage between multiple medical records systems in a UK National Health Service Trust, examining the effects of decision thresholds on linkage accuracy. We report demographic factors associated with correct linkage.ResultsThe system supports dates of birth (DOBs), forenames, surnames, three-state gender, and UK postcodes. Fuzzy representations are supported for all except gender, and there is support for additional transformations, such as accent misrepresentation, variation for multi-part surnames, and name re-ordering. Calculated log odds predicted a proband’s presence in the sample database with an area under the receiver operating curve of 0.997–0.999 for non-self database comparisons. Log odds were converted to a decision via a consideration threshold θ and a leader advantage threshold δ. Defaults were chosen to penalize misidentification 20-fold versus linkage failure. By default, complete DOB mismatches were disallowed for computational efficiency. At these settings, for non-self database comparisons, the mean probability of a proband being correctly declared to be in the sample was 0.965 (range 0.931–0.994), and the misidentification rate was 0.00249 (range 0.00123–0.00429). Correct linkage was positively associated with male gender, Black or mixed ethnicity, and the presence of diagnostic codes for severe mental illnesses or other mental disorders, and negatively associated with birth year, unknown ethnicity, residential area deprivation, and presence of a pseudopostcode (e.g. indicating homelessness). Accuracy rates would be improved further if person-unique identifiers were also used, as supported by the software. Our two largest databases were linked in 44 min via an interpreted programming language.ConclusionsFully de-identified matching with high accuracy is feasible without a person-unique identifier and appropriate software is freely available.

Determinants of Successful Human Immunodeficiency Virus Treatment Outcomes: A Linkage of National Data Sources in Malaysia.

Concerted efforts have been undertaken to reduce the human immunodeficiency virus (HIV) infection by the year 2030 in Malaysia. A situational analysis of the performance of successful HIV treatment and its determinants is vital; however, this information remains scarce. This study aimed to identify the determinants of undetectable viral load among people living with HIV (PLHIV). Newly diagnosed HIV cases (n = 493) registered under the Malaysia HIV/AIDS-related national databases from June 2018 to December 2019 were studied. The deterministic matching method was applied to link the records in two national databases (at Kuala Lumpur and Putrajaya Federal Territories Health Department, JKWPKLP HIV line-listing database and National AIDS Registry). Successful HIV treatment, an outcome variable, was measured by the undetectable viral load < 200 copies/mL after 1 year of antiretroviral therapy initiation. Logistic regression analysis was applied in the current study. Results showed that 454/493 (92.2%; 95% confidence interval [CI]: 89.8%, 94.6%) PLHIV had successful HIV treatment. Study participants had a mean (SD) age of 30 (8.10) years old, predominantly male (96.1%) and sexually transmission (99.9%). The multiple logistic regression analysis revealed two significant determinants including the timing of ART initiation (AOR = 3.94; 95% CI: 1.32, 11.70; P = 0.014) and establishment of Sexually Transmitted Infection Friendly Clinic (STIFC) (AOR = 3.40; 95% CI: 1.47, 7.85; P = 0.004). Non-significant variables included gender, education level, HIV risk exposure, and co-infections of tuberculosis and Hepatitis C. JKWPKLP is on the right track to achieving universal treatment as a prevention strategy. Reinforcement of early ART initiation and establishment of STIFC are recommended.

Linking Biomedical Data Warehouse Records With the National Mortality Database in France: Large-scale Matching Algorithm.

Often missing from or uncertain in a biomedical data warehouse (BDW), vital status after discharge is central to the value of a BDW in medical research. The French National Mortality Database (FNMD) offers open-source nominative records of every death. Matching large-scale BDWs records with the FNMD combines multiple challenges: absence of unique common identifiers between the 2 databases, names changing over life, clerical errors, and the exponential growth of the number of comparisons to compute. We aimed to develop a new algorithm for matching BDW records to the FNMD and evaluated its performance. We developed a deterministic algorithm based on advanced data cleaning and knowledge of the naming system and the Damerau-Levenshtein distance (DLD). The algorithm's performance was independently assessed using BDW data of 3 university hospitals: Lille, Nantes, and Rennes. Specificity was evaluated with living patients on January 1, 2016 (ie, patients with at least 1 hospital encounter before and after this date). Sensitivity was evaluated with patients recorded as deceased between January 1, 2001, and December 31, 2020. The DLD-based algorithm was compared to a direct matching algorithm with minimal data cleaning as a reference. All centers combined, sensitivity was 11% higher for the DLD-based algorithm (93.3%, 95% CI 92.8-93.9) than for the direct algorithm (82.7%, 95% CI 81.8-83.6; P<.001). Sensitivity was superior for men at 2 centers (Nantes: 87%, 95% CI 85.1-89 vs 83.6%, 95% CI 81.4-85.8; P=.006; Rennes: 98.6%, 95% CI 98.1-99.2 vs 96%, 95% CI 94.9-97.1; P<.001) and for patients born in France at all centers (Nantes: 85.8%, 95% CI 84.3-87.3 vs 74.9%, 95% CI 72.8-77.0; P<.001). The DLD-based algorithm revealed significant differences in sensitivity among centers (Nantes, 85.3% vs Lille and Rennes, 97.3%, P<.001). Specificity was >98% in all subgroups. Our algorithm matched tens of millions of death records from BDWs, with parallel computing capabilities and low RAM requirements. We used the Inseehop open-source R script for this measurement. Overall, sensitivity/recall was 11% higher using the DLD-based algorithm than that using the direct algorithm. This shows the importance of advanced data cleaning and knowledge of a naming system through DLD use. Statistically significant differences in sensitivity between groups could be found and must be considered when performing an analysis to avoid differential biases. Our algorithm, originally conceived for linking a BDW with the FNMD, can be used to match any large-scale databases. While matching operations using names are considered sensitive computational operations, the Inseehop package released here is easy to run on premises, thereby facilitating compliance with cybersecurity local framework. The use of an advanced deterministic matching algorithm such as the DLD-based algorithm is an insightful example of combining open-source external data to improve the usage value of BDWs.

COVID-19 vaccination uptake for half a million non-EU migrants and refugees in England: a linked retrospective population-based cohort study

BackgroundCOVID-19 has highlighted severe health inequities experienced by certain migrants. Despite evidence suggesting that migrants are at risk of under-immunisation, data are limited for migrants' COVID-19 vaccine uptake in England. MethodsWe did a retrospective population-based cohort study on COVID-19 vaccination uptake in England. We linked the Million Migrant cohort (which includes non-EU migrants and resettled refugees) to the national COVID-19 vaccination dataset, using a stepwise deterministic matching procedure adapted from NHS Digital, and compared migrants with the general population. For migrants who linked to at least one vaccination record, we estimate temporal trends in first dose uptake and differences in second and third dose uptake and consequent delays between Dec 8, 2020, and April 20, 2022, by age, visa type, and ethnicity. FindingsOf the 465 470 migrants who linked to one or more vaccination record, 427 073 (91·8%) received a second dose and 238 721 (51·3%) received a third. Migrants (>30 years) reached 75% first-dose coverage between 1 and 2 weeks after the general population in England, with the gap widening to 6 weeks for younger migrants (16–29 years). Refugees specifically had a higher risk of a delayed second dose (odds ratio 1·75 [95 CI% 1·62–1·88]) and third dose (1·41 [1·31–1·53]). Older migrants (>65 years) were at least four times more likely to have not received their second or third dose compared with those of the same age in England. InterpretationUptake of the first dose was slower across all age groups for migrants compared with the general population. Refugees and older migrants were more likely to have delayed uptake of COVID-19 vaccines and to not have received their second or third dose. Policymakers, researchers, and practitioners should consider how to best drive uptake of COVID-19 and other routine vaccine doses and understand and address personal and structural barriers to vaccination systems for diverse migrant populations. FundingWellcome Trust and UK Research and Innovation.

Building partnerships, capacity, and knowledge through a use of newly linked child development and education datasets in Ontario, Canada.

ObjectivesThe objective of this study was to establish a partnership between a university and a jurisdictional education body (Education Quality and Assessment Organization, EQAO) which would allow creation of a linked dataset from kindergarten to later grades in order to examine educational trajectory in mathematics in Ontario. ApproachBuilding on mutual goals of improving the understanding of children’s learning trajectories, we developed a project with an investigator team that included university researchers and representatives of the provincial educational assessment body, to link a database of child development status in kindergarten (Early Development Instrument/EDI data, including neighbourhood socioeconomic/SES index) with academic assessment EQAO data, and received research funding. A deterministic matching process was employed to match the datasets. We examined differences between the unmatched and fully matched cases and constructed a growth mixture model of math scores in grades 3, 6 and 9, with key EDI/SES variables as covariates. ResultsDespite lacking a common identifier, we successfully matched approximately 50% of the EDI cases from 2002-2014 (n=183,771). Effect sizes indicated negligible differences between matched and unmatched, except for SES and child development status, which were poorer for unmatched group. A 3-class solution was the best fit for a 20,000-person subsample of math trajectories based on AIC, BIC, ICL, and entropy values as well as sufficiently high proportions of posterior probabilities, which indicate confidence in class membership. 61% of sample showed steady moderate-high achievement; 9% started high, but declined, and 30% deteriorated then improved. Males, children in low SES, and those with adequate kindergarten EDI outcomes had better math achievement trajectories than females, children in high SES, and those with poor kindergarten outcomes. ConclusionGiven the two datasets were collected without explicit linkage plan, the matching was only 50%, nevertheless resulting in a large database that allows study of early development antecedents of students’ educational trajectories. The partnership between university and EQAO ensures a wide dissemination of results in both academia and policy worlds.

A population-based descriptive study of neonatal abstinence syndrome using hospital discharge and birth certificate data

ABSTRACT Background Neonatal abstinence syndrome (NAS), largely a consequence of prenatal opioid exposure, results in substantial morbidity. Population-based studies of NAS going beyond Medicaid populations and hospital discharge data (HDD) alone are limited. Using statewide Tennessee (TN) HDD and birth certificate (BC) data, we examined trends and evaluated maternal and infant factors associated with NAS. Methods We conducted a population-based descriptive study during 2013–2017 in TN. NAS infants were identified with International Classification of Diseases (ICD)-9-Clinical Modification (CM) and ICD-10-CM codes in HDD and linked to BC data using iterative deterministic matching algorithms. Descriptive analyses were conducted for infant and maternal factors (exposures) by NAS (outcome). Multivariable logistic regression models were used to estimate adjusted ORs and 95% CIs. Results NAS incidence increased from 13.4 to 15.4 per 1,000 live births between 2013 and 2017 (15% increase; ptrend <0.001), but remained stable in 2017. In adjusted models, maternal factors associated with reduced odds of NAS included breastfeeding (OR:0.55, 95%CI:0.52–0.59) and prenatal care (OR:0.36, 95%CI:0.32–0.41). Smoking, preterm birth, and lower birthweight were associated with increased odds of NAS. Conclusions This study highlights the value of utilizing surveillance data to monitor trends and correlates of NAS to inform prevention efforts and targeting of public health resources.

Comparing age model techniques for the last glacial cycle: A case study of ten Iberian Margin sediment cores

Because interpretations of paleoclimate records from ocean sediment cores rely on age models to estimate age as a function of core depth, age model uncertainty can affect the conclusions drawn from paleoclimate data. To assess the age model uncertainty associated with three methods, we compare the age models they generate over the last glacial cycle for high-resolution data from ten western and southern Iberian Margin cores. The first method uses radiocarbon data and the Bayesian statistical software Bacon [Blaauw and Christen, 2011]. The second method aligns benthic δ18O to a target core using the probabilistic alignment algorithm HMM-Match [Lin et al., 2014]. The third method is alignment of planktonic δ18O or sea surface temperature (SST) using the deterministic alignment software Match [Lisiecki and Lisiecki, 2002]. Where three or more age model types overlap in a core, the average range of age estimates is 1.7 kyr, and 90% of age ranges are less than 3 kyr. For the two probabilistic methods, HMM-Match and Bacon, age estimates agree to within uncertainty, and the highest resolution records yield 95% confidence interval widths of 1–2 kyr. However, HMM-Match appears to underestimate alignment uncertainty near gaps in benthic δ18O data in one core, as indicated by disagreement with the core's planktonic δ18O and SST alignments. Planktonic δ18O appears to provide the most precise age models for Iberian Margin cores due to strong millennial-scale variability and good correlation between aligned cores. Probabilistic alignment algorithms should be developed for planktonic δ18O and SST proxies to improve estimates of the age uncertainty for these alignments.

Matching Special Olympics registration data with administrative health databases: Feasibility and health status differences in children and youth with IDD

BackgroundChildren and youth with intellectual and developmental disabilities (IDD) experience health disparities. What is unknown is if data collected from children and youth with IDD who participate in Special Olympics is representative of children and youth with IDD who do not. ObjectivesAim 1: determine the feasibility of matching a database of registrants from Special Olympics Ontario (SOO), with population-based health services databases in Ontario, Canada housed at ICES. Aim 2: evaluate the differences between the database sources with regards to demographic variables and clinical status. MethodsUsing deterministic and probabilistic matching, registration data from SOO were matched to administrative health databases. Established algorithms were used to determine the prevalence of asthma, diabetes, and mental disorder in addition to demographic variables. ResultsThe matching rate was over 90%; 8404 were attributed to children and youth between the ages of 0–19 years. When comparing SOO participants with IDD to non-SOO participants with IDD, children and youth who participate in SOO were, on average, older with no further differences between groups on clinical or demographic variables. When comparing those previously not identified in the health services databases (from SOO) to those with IDD identified by ICES, the SOO participants appear to use the health system less, possibly indicating a better health status. ConclusionsResearch conducted on child and youth who participate in Special Olympics Ontario can be generalized to the broader population of children and youth with IDD in Canada when adjusted for age; however, care should be taken when comparing levels of overall morbidity.

Setting up statistical registers of individuals and dwellings in France: Approach and first steps

In order to rationalize the massive use of administrative data in the production of demographic and social statistics, the French national statistical institute (INSEE) has decided to launch a program to create a system of interconnected registers of individuals, dwellings and households. This program, called RESIL, is based on the mobilization of external data, particularly administrative data, in strict compliance with the conditions for the protection of individual data. Thus, through these registers, INSEE will have a reference universe that will allow: (i) to constitute the sampling frames as well as the calibration margins of household surveys; (ii) to measure the quality of the coverage of sources; and (iii) to match different datasets: surveys with administrative data, administrative data with administrative data in order to provide richer information. To set up this information system, INSEE will draw on international experiences, as it plans to use methodologies already implemented in several national statistical institutes, including: • deterministic and probabilistic matching methods for identifying and comparing different administrative sources. These methods are essential insofar as France does not have a unique and shared identifier; • the sign of life method to define the reference population from the presence of individuals in several input data; • the Dual System Estimation methodology to measure the quality of coverage of the registers using annual population census data (it is a real opportunity to have these annual points of comparison to assess the quality of the registers, especially during the set-up period). This paper will therefore present the stakes and the context, in particular the legal context, of the implementation of these registers, and will then describe the main methodological principles and the proofs of concept planned for its implementation.

Development of a homelessness risk screening tool for emergency department patients.

To develop a screening tool to identify emergency department (ED) patients at risk of entering a homeless shelter, which could inform targeting of interventions to prevent future homelessness episodes. Linked data from (1) ED patient baseline questionnaires and (2) citywide administrative homeless shelter database. Stakeholder-informed predictive modeling utilizing ED patient questionnaires linked with prospective shelter administrative data. The outcome was shelter entry documented in administrative data within 6months following the baseline ED visit. Exposures were responses to questions on homelessness risk factors from baseline questionnaires. Research assistants completed questionnaires with randomly sampled ED patients who were medically stable, not in police/prison custody, and spoke English or Spanish. Questionnaires were linked to administrative data using deterministic and probabilistic matching. Of 1993 ED patients who were not homeless at baseline, 5.6% entered a shelter in the next 6months. A screening tool consisting of two measures of past shelter use and one of past criminal justice involvement had 83.0% sensitivity and 20.4% positive predictive value for future shelter entry. Our study demonstrates the potential of using cross-sector data to improve hospital initiatives to address patients' social needs.

Improving Longitudinal Outcomes, Efficiency, and Equity in the Care of Patients With Congenital Heart Disease

BackgroundLongitudinal follow-up, resource utilization, and health disparities are top congenital heart research and care priorities. Medicaid claims include longitudinal data on inpatient, outpatient, emergency, pharmacy, rehabilitation, home health utilization, and social determinants of health—including mother-infant pairs. ObjectivesThe New York Congenital Heart Surgeons Collaborative for Longitudinal Outcomes and Utilization of Resources linked robust clinical details from locally held state and national registries from 10 of 11 New York congenital heart centers to Medicaid claims, building a novel, statewide mechanism for longitudinal assessment of outcomes, expenditures, and health inequities. MethodsThe authors included all children <18 years of age undergoing cardiac surgery in The Society of Thoracic Surgeons Congenital Heart Surgery Database or the New York State Pediatric Congenital Cardiac Surgery Registry from 10 of 11 New York centers, 2006 to 2019. Data were linked via iterative, ranked deterministic matching on direct identifiers. Match rates were calculated and compared. Proportions of the linked cohort trackable over 3, 5, and 10 years were described. ResultsOf 14,097 registry cases, 59% (n = 8,322) reported Medicaid use. Of these, 7,414 were linked to New York claims, at an 89% match rate. Of matched cases, the authors tracked 79%, 74%, and 65% of children over 3, 5, and 10 years when requiring near-continuous Medicaid enrollment. Allowing more lenient enrollment criteria, the authors tracked 86%, 82%, and 76%, respectively. Mortality over this time was 7.7%, 8.4%, and 10.0%, respectively. Manual validation revealed ∼100% true matches. ConclusionsThis establishes a novel statewide data resource for assessment of longitudinal outcome, health expenditure, and disparities for children with congenital heart disease.

Relaxing door-to-door matching reduces passenger waiting times: A workflow for the analysis of driver GPS traces in a stochastic carpooling service

Carpooling has the potential to transform itself into a mass transportation mode by abandoning its adherence to deterministic passenger-driver matching for door-to-door journeys, and by adopting instead stochastic matching on a network of fixed meeting points. Stochastic matching is where a passenger sends out a carpooling request at a meeting point, and then waits for the arrival of a self-selected driver who is already travelling to the requested meeting point. Crucially there is no centrally dispatched driver. Moreover, the carpooling is assured only between the meeting points, so the onus is on the passengers to travel to/from them by their own means. Thus the success of a stochastic carpooling service relies on the convergence, with minimal perturbation to their existing travel patterns, to the meeting points which are highly frequented by both passengers and drivers. Due to the innovative nature of stochastic carpooling, existing off-the-shelf workflows are largely insufficient for this purpose. To fill the gap in the market, we introduce a novel workflow, comprising of a combination of data science and GIS (Geographic Information Systems), to analyse driver GPS traces. We implement it for an operational stochastic carpooling service in south-eastern France, and we demonstrate that relaxing door-to-door matching reduces passenger waiting times. Our workflow provides additional key operational indicators, namely the driver flow maps, the driver flow temporal profiles and the driver participation rates.