What is counted counts: An innovative linkage of police, hospital, and spatial data for transportation injury prevention

INTRODUCTION

Growing research indicates transportation injury surveillance using police collision reporting alone underrepresents injury to vulnerable groups, including pedestrians, cyclists, and people of color. This reflects differing reporting patterns and non-clinicians' challenge in accurately evaluating injury severity. To our knowledge, San Francisco is the first U.S. city to link and map hospital and police injury data. Analysis of linked data injury patterns informs interventions supporting traffic fatality and injury prevention goals.

METHODS

Injury and fatality records 2013-2015 were collected from San Francisco Police, Emergency Medical Services (EMS), Medical Examiner, and Zuckerberg San Francisco General Hospital (ZSFG). Probabilistic linkage was conducted using LinkSolv9.0 on match variables collision/admission time, name, birthdate, sex, travel mode, and geographic collision location.

RESULTS

From 2013-2015, this study identified 17,000+ transportation-related injuries on public roadways in San Francisco. Twenty-six percent (n = 4,415) appeared in both police and ZSFG sources. Linked injury records represent 39% of police records (N = 11,403) and 43% of hospital records (N = 10,223). Among hospital records, 34% of cyclist, 38% of motor vehicle occupant, 61% of pedestrian, and 54% of motorcyclist records linked with a police record. Linkage rate varied by travel mode even after controlling for injury severity. Transportation-injured ZSFG-treated patients lacking police reports were more often cyclists, male, Hispanic or Black, and less often occupants of motor vehicles compared to those with injuries captured only in police reports.

CONCLUSIONS

Incorporating hospital and EMS spatial data into injury surveillance systems historically reliant on police reports offers trifold benefits. First, linkage captures injuries absent in police data, adding data on populations empirically vulnerable to injury. Second, it improves injury severity assessment. Finally, linked data better informs and targets interventions serving injury-burdened populations and road users, advancing transportation injury prevention.

PRACTICAL APPLICATIONS

Linkage closes data gaps, improving ability to quantify injury and develop evidence-based interventions for vulnerable groups.

A new hybrid record linkage process to make epidemiological databases interoperable: application to the GEMO and GENEPSO studies involving BRCA1 and BRCA2 mutation carriers

BACKGROUND

Linking independent sources of data describing the same individuals enable innovative epidemiological and health studies but require a robust record linkage approach. We describe a hybrid record linkage process to link databases from two independent ongoing French national studies, GEMO (Genetic Modifiers of BRCA1 and BRCA2), which focuses on the identification of genetic factors modifying cancer risk of BRCA1 and BRCA2 mutation carriers, and GENEPSO (prospective cohort of BRCAx mutation carriers), which focuses on environmental and lifestyle risk factors.

METHODS

To identify as many as possible of the individuals participating in the two studies but not registered by a shared identifier, we combined probabilistic record linkage (PRL) and supervised machine learning (ML). This approach (named "PRL + ML") combined together the candidate matches identified by both approaches. We built the ML model using the gold standard on a first version of the two databases as a training dataset. This gold standard was obtained from PRL-derived matches verified by an exhaustive manual review. Results The Random Forest (RF) algorithm showed a highest recall (0.985) among six widely used ML algorithms: RF, Bagged trees, AdaBoost, Support Vector Machine, Neural Network. Therefore, RF was selected to build the ML model since our goal was to identify the maximum number of true matches. Our combined linkage PRL + ML showed a higher recall (range 0.988-0.992) than either PRL (range 0.916-0.991) or ML (0.981) alone. It identified 1995 individuals participating in both GEMO (6375 participants) and GENEPSO (4925 participants).

CONCLUSIONS

Our hybrid linkage process represents an efficient tool for linking GEMO and GENEPSO. It may be generalizable to other epidemiological studies involving other databases and registries.

Probabilistic linkage without personal information successfully linked national clinical datasets

Background

Probabilistic linkage can link patients from different clinical databases without the need for personal information. If accurate linkage can be achieved, it would accelerate the use of linked datasets to address important clinical and public health questions.

Objective

We developed a step-by-step process for probabilistic linkage of national clinical and administrative datasets without personal information, and validated it against deterministic linkage using patient identifiers.

Study Design and Setting

We used electronic health records from the National Bowel Cancer Audit and Hospital Episode Statistics databases for 10,566 bowel cancer patients undergoing emergency surgery in the English National Health Service.

Results

Probabilistic linkage linked 81.4% of National Bowel Cancer Audit records to Hospital Episode Statistics, vs. 82.8% using deterministic linkage. No systematic differences were seen between patients that were and were not linked, and regression models for mortality and length of hospital stay according to patient and tumour characteristics were not sensitive to the linkage approach.

Conclusion

Probabilistic linkage was successful in linking national clinical and administrative datasets for patients undergoing a major surgical procedure. It allows analysts outside highly secure data environments to undertake linkage while minimizing costs and delays, protecting data security, and maintaining linkage quality.

Probabilistic Linkage of Randomized Controlled Trial Data to Administrative Claims: A Case Study of Patients from Baricitinib Clinical Trials

Introduction

The aim of this work is to assess the feasibility of probabilistically linking randomized controlled trial (RCT) data to claims data in a real-world setting to inform future rheumatoid arthritis (RA) research.

Methods

This retrospective cohort study utilized IQVIA’s Patient Centric Medical Claims (Dx) Database, IQVIA’s Longitudinal Prescription Claims (LRx) Database, and Lilly’s baricitinib RCT data from a sample of patients that consented to the linkage of their de-identified insurance claims to their de-identified RCT data. Patients were initially matched on age, gender, and three-digit ZIP code of the provider and further matched according to a point scoring system using additional clinical variables.

Results

A total of 245 patients from 49 US clinical trial sites were eligible for the study and 78 (31.8%) of these patients consented to participate. Of the 78 consented patients, 69 (88%) were successfully matched on age, gender, and three-digit ZIP code of the provider. Of the 69 patients successfully matched on age, gender, and three-digit ZIP code of the provider, 44 (63.8%) had at least one sufficient match using the point scoring system. Of these 44, 23 (52.3%) patients matched at a ratio of one RCT patient to one Dx/LRx patient, 11 (25.0%) at a ratio of 1:2, 7 (15.9%) at a ratio of 1:3 and three (6.8%) at a ratio of 1:4 or greater. To further improve match ratios, a variable hierarchy was applied to the 18 RCT patients with 2–3 matches. Overall, 38 of the 78 (48.7%) consented RCT patients were successfully matched 1:1 to claims database patients.

Conclusions

This probabilistic linkage methodology demonstrates the feasibility, at a moderate linkage rate, of linking patients from RCTs to real-world data, which can provide a means to assess additional information not usually collected within or following a clinical trial.

Linking Datasets to Characterize Injury and Illness in Alaska's Fishing Industry

Limited research has characterized nonfatal injury/illness in Alaska's hazardous fishing industry. This study aimed to determine (a) the utility of linking datasets to conduct surveillance, and (b) injury/illness patterns during 2012-2016. Data were obtained from the Alaska Trauma Registry (ATR), Fishermen's Fund (FF), and US Coast Guard (USCG). Datasets were coded to identify patterns in injury/illness characteristics and circumstances. Probabilistic linkage methods were utilized to identify unique incidents that appeared in more than one dataset. After linking datasets, 3,014 unique injury/illness cases were identified. By dataset, 2,365 cases appeared only in FF, 486 only in USCG, 110 only in ATR, 25 in ATR and FF, 15 in ATR and USCG, 10 in USCG and FF, and 3 in all datasets. FF mainly captured claims submitted by small, independently-owned vessels in Southcentral and Southeastern Alaska. In contrast, USCG mainly captured reports from large, company-owned vessels in Western Alaska. By nature, cases were most frequently sprains, strain, and tears (27%), cuts (15%), and fractures (11%). Across fleets, injuries/illnesses most frequently resulted from contact with objects and equipment (41%), overexertion and bodily reaction (27%), and slips, trips, and falls (20%). Work processes associated with traumatic injuries were most frequently hauling gear (18%) and walking, climbing, and descending (18%). Half of all injuries were of moderate severity (53%). Linking datasets, which capture different segments of Alaska's fishing industry, provides the most comprehensive understanding of nonfatal injury/illness to date. These results, stratified by fleet and severity, will inform prevention strategies.

A comparison of accuracy and computational feasibility of two record linkage algorithms in retrieving vital status information from HIV/AIDS patients registered in Brazilian public databases

BACKGROUND AND OBJECTIVE

While cross-referencing information from people living with HIV/AIDS (PLWHA) to the official mortality database is a critical step in monitoring the HIV/AIDS epidemic in Brazil, the accuracy of the linkage routine may compromise the validity of the final database, yielding to biased epidemiological estimates. We compared the accuracy and the total runtime of two linkage algorithms applied to retrieve vital status information from PLWHA in Brazilian public databases.

METHODS

Nominally identified records from PLWHA were obtained from three distinct government databases. Linkage routines included an algorithm in Python language (PLA) and Reclink software (RlS), a probabilistic software largely utilized in Brazil. Records from PLWHA¹ known to be alive were added to those from patients reported as deceased. Data were then searched into the mortality system. Scenarios where 5% and 50% of patients actually dead were simulated, considering both complete cases and 20% missing maternal names.

RESULTS

When complete information was available both algorithms had comparable accuracies. In the scenario of 20% missing maternal names, PLA² and RlS³ had sensitivities of 94.5% and 94.6% (p > 0.5), respectively; after manual reviewing, PLA sensitivity increased to 98.4% (96.6-100.0) exceeding that for RlS (p < 0.01). PLA had higher positive predictive value in 5% death proportion. Manual reviewing was intrinsically required by RlS in up to 14% register for people actually dead, whereas the corresponding proportion ranged from 1.5% to 2% for PLA. The lack of manual inspection did not alter PLA sensitivity when complete information was available. When incomplete data was available PLA sensitivity increased from 94.5% to 98.4%, thus exceeding that presented by RlS (94.6%, p < 0.05). RlS spanned considerably less processing time compared to PLA.

CONCLUSION

Both linkage algorithms presented interchangeable accuracies in retrieving vital status data from PLWHA. RlS had a considerably lesser runtime but intrinsically required manually reviewing a fastidious proportion of the matched registries. On the other hand, PLA spent quite more runtime but spared manual reviewing at no expense of accuracy.

Motorcycle helmet effectiveness in reducing head, face and brain injuries by state and helmet law

Background

Despite evidence that motorcycle helmets reduce morbidity and mortality, helmet laws and rates of helmet use vary by state in the U.S.

Methods

We pooled data from eleven states: five with universal laws requiring all motorcyclists to wear a helmet, and six with partial laws requiring only a subset of motorcyclists to wear a helmet. Data were combined in the Crash Outcome Data Evaluation System’s General Use Model and included motorcycle crash records probabilistically linked to emergency department and inpatient discharges for years 2005-2008. Medical outcomes were compared between partial and universal helmet law settings. We estimated adjusted relative risks (RR) and 95 % confidence intervals (CIs) for head, facial, traumatic brain, and moderate to severe head/facial injuries associated with helmet use within each helmet law setting using generalized log-binomial regression.

Results

Reported helmet use was higher in universal law states (88 % vs. 42 %). Median charges, adjusted for inflation and differences in state-incomes, were higher in partial law states (emergency department $1987 vs. $1443; inpatient $31,506 vs. $25,949). Injuries to the head and face, including traumatic brain injuries, were more common in partial law states. Effectiveness estimates of helmet use were higher in partial law states (adjusted-RR (CI) of head injury: 2.1 (1.9-2.2) partial law single vehicle; 1.4 (1.2, 1.6) universal law single vehicle; 1.8 (1.6-2.0) partial law multi-vehicle; 1.2 (1.1-1.4) universal law multi-vehicle).

Conclusions

Medical charges and rates of head, facial, and brain injuries among motorcyclists were lower in universal law states. Helmets were effective in reducing injury in both helmet law settings; lower effectiveness estimates were observed in universal law states.

The utility of imputed matched sets. Analyzing probabilistically linked databases in a low information setting

OBJECTIVE

To compare results from high probability matched sets versus imputed matched sets across differing levels of linkage information.

METHODS

A series of linkages with varying amounts of available information were performed on two simulated datasets derived from multiyear motor vehicle crash (MVC) and hospital databases, where true matches were known. Distributions of high probability and imputed matched sets were compared against the true match population for occupant age, MVC county, and MVC hour. Regression models were fit to simulated log hospital charges and hospitalization status.

RESULTS

High probability and imputed matched sets were not significantly different from occupant age, MVC county, and MVC hour in high information settings (p > 0.999). In low information settings, high probability matched sets were significantly different from occupant age and MVC county (p < 0.002), but imputed matched sets were not (p > 0.493). High information settings saw no significant differences in inference of simulated log hospital charges and hospitalization status between the two methods. High probability and imputed matched sets were significantly different from the outcomes in low information settings; however, imputed matched sets were more robust.

CONCLUSIONS

The level of information available to a linkage is an important consideration. High probability matched sets are suitable for high to moderate information settings and for situations involving case-specific analysis. Conversely, imputed matched sets are preferable for low information settings when conducting population-based analyses.