Record Linkage Algorithm Research Articles

Privacy Preserving Evaluation of Sensitive Population Record Linkage in the Absence of Ground Truth Data

ObjectivesLinkage of data containing personal information allows extensive studies in the health and social sciences for population level studies. However, real-world linkage applications often lack ground truth (GT) data, generally due to privacy concerns, which hinders linkage quality assessment. We propose a method to assess the linkage of sensitive population data in the absence of GT data. ApproachOur method takes as input a set of sensitive and public population clusters. The sensitive clusters are generated with some record linkage algorithm, whereas the public population dataset, which is considered an approximation for the GT, is assumed to have a group identifier allowing generation of known clusters, such as households or families. We encode clusters as bit vectors and conduct cluster matching across the sensitive and public datasets using vector comparisons. Based on matched clusters, the population linkage quality of sensitive data is estimated. ResultsWe conducted experiments on a sensitive Scottish birth data set containing 17,613 records from 1861 to 1901, and four successive publicly available Scottish census datasets from 1871 to 1901. Using the quality measures precision, recall and the F-measure, we show that our estimations are close to GT quality, with a high correlation in F-measure values ranging from 88% to 99%. ConclusionWe proposed a method to estimate the linkage quality of sensitive population data for situations where public data with similar characteristics are available for approximating the GT. Our experimental evaluation showed that our method was effective in estimating true linkage quality with high accuracy.

Exploring Multimorbidity Through the Use of Deterministic and Probabilistic Record Linkage of Data in Five Health Facilities in Nairobi

IntroductionData collected by Health Demographic Surveillance Systems (HDSSs) at the community level can support very powerful research questions if it was linked to records at health facilities. In particular, understanding multimorbidity and building predictive models for key health outcomes requires linked data collected both in the community and health facilities. Additionally, the linked data can assist health management teams at the national, county, and local levels in making evidence-based choices. Problemthe inability to link population level data with records at health facilities affects the ability to use data to solve key health research questions. lack of common identifiers affects the possibility of integrating clinic records with population-based data. MethodologyWe intend to link data from five health facilities in Nairobi County serving the population in the Nairobi Urban Health Demographic Surveillance System (NUHDSS) where longitudinal data was collected between 2001 and 2018. The longitudinal data is from a database of about 260,000 individuals. This will result in a merged dataset and will also provide an opportunity to assess the accuracy of the deterministic and probabilistic record linkage algorithm compared with other methods. Note: We will have results by the time the conference is happening as groundwork has been put. ConclusionThe growing burden of multimorbidity on the African continent requires creative and effective strategies of addressing the problem through utilization of existing longitudinal data. By merging clinic records with population-based data, this initiative will promote the use of data evidence informed decision making for interventions to address multimorbidity.

A Bayesian Record Linkage Approach that Adjusts for Variables in One File

In many healthcare and social science applications, information about units is dispersed across multiple data files. Linking records across files is necessary to estimate associations between variables exclusive to each of the files. Common record linkage algorithms only rely on similarities between linking variables common to all the files. Moreover, analysis of linked files often ignores errors that may arise from incorrect or missed links. Bayesian record linking methods allow for natural propagation of linkage errors, by jointly sampling the linkage structure and the model parameters. We extend an existing Bayesian record linkage approach to integrate associations between variables exclusive to each file being linked. We show analytically, and using simulations, that the proposed method improves the linkage process, and results in accurate statistical inferences. We apply the proposed method to link Meals on Wheels (MOW) recipients to Medicare Enrollment records, and examine the relationship between activities of daily living and healthcare utilization among MOW recipients.

Developing an integrated HIV clinical and population database in rural south Africa

BackgroundHIV cohort data are crucial for understanding the effects of HIV infection and treatment. Linkage between HIV clinical and population records unlocks potential to understand the cause and consequences of social-economic factors on HIV-care seeking behaviors and mortality outcomes. We created HIV data infrastructure by linking HIV treatment records with routinely collected records. MethodsWe linked HIV cohort data(n=119,434) to Health and Demographic Surveillance(n=256,425), laboratory tests(n=617,314), clinic visits and hospital admissions(n=170,325) from KwaZulu-Natal, an HIV-endemic setting in rural South Africa. We used graph-based probabilistic record linkage algorithm to link the data for HIV patients accessing care at 17 clinics within the Hlabisa sub-district. Africa Health Research Institute (AHRI) has collected health and demographic surveillance data since 2000. AHRI has set up a hospital information system in the district to capture ICD-10 code admissions and discharges. ResultsFrom the 5% sample of 462,524 linkage clusters, the computed F-score was 0.88, sensitivity was 88% and positive predictive value was 91%. 38.5% of the patient ID clusters contained records linked across multiple databases. 36% of the HIV patient ID clusters contained records linked with laboratory tests. 19.5% of the HIV patient ID clusters contained data on demographic surveillance. 12.5% of the HIV patient ID clusters contained records linked with clinic visits and hospital admissions. 4% of patient IDs contained records across all four databases. ConclusionWe have successfully created a cross-sectorial HIV cohort through probabilistic record linkage methods. The cohort provides a platform to answer policy-relevant questions.

Generating synthetic identifiers to support development and evaluation of data linkage methods

We aimed to develop a framework for generating synthetic identifier datasets to support development and evaluation of data linkage methods. We evaluated whether replicating associations between attributes and identifiers improved the utility of the synthetic data for assessing linkage error. We determined the steps required to generate synthetic identifiers that replicate the properties of real-world data collection. We generated synthetic versions of a large United Kingdom cohort study (the Avon Longitudinal Study of Parents and Children), according to the quality and completeness of identifiers recorded over several waves of the cohort. We evaluated the utility of the synthetic identifier data in terms of assessing linkage quality (false matches and missed matches). Comparing data from two collection points in ALSPAC, we found within-person disagreement in identifiers (differences in recording due to both natural change and non-valid entries) in 18% of surnames and 12% of forenames. Rates of disagreement varied by maternal age and ethnic group. Synthetic data provided accurate estimates of linkage quality metrics compared with the original data (within 0.13-0.55% for missed matches and 0.00-0.04% for false matches). Incorporating associations between identifier errors and maternal age/ethnicity improved synthetic data utility. Replicating dependencies between attribute values (e.g. ethnicity), values of identifiers (e.g. name), identifier disagreements (e.g. missing values, errors or changes over time), and their patterns and distribution structure enables generation of realistic synthetic data that can be used for robust evaluation of linkage methods. Our framework provides a novel and generalisable mechanism for developing and benchmarking record linkage algorithms.

Building a Historic Population Platform

ObjectivesWith the increasing digital availability of large population databases of historical census or vital event records, the tasks of storing, cleaning, processing, linking and analysing such data become more challenging. Suitable computing platforms and software systems are required to handle such databases, and facilitate the application of complex record linkage algorithms, for example to reconstruct populations that cover a full country over many decades. We present our efforts to achieve these goals on a database of over 20 million vital event records spanning over 120 years to create a “Historic Population Platform” (HiPP). ApproachWe created a graph database using the Neo4J software, where each birth, death and marriage certificate is represented as a node. We then generated actor nodes from these certificates which represent individuals (such as birth babies and their parents, or marriage brides and grooms). Data cleaning steps included the correction and imputation of invalid, corrupted, and missing age and date values using information from related certificates. ResultsOur initial graph database contains over 100 million nodes and nearly 200 million edges, while our data cleaning methods help to substantially increase the number of valid age and date values. ConclusionThe availability of large historical population databases provides exciting opportunities for social science and health research. However, existing methods have limitations in handling data quality and the sizes of such databases. We presented novel methods to deal with these challenges which we hope will be of use for other projects that aim to build a HiPP.

Manual Evaluation of Record Linkage Algorithm Performance in Four Real-World Datasets.

Patient data are fragmented across multiple repositories, yielding suboptimal and costly care. Record linkage algorithms are widely accepted solutions for improving completeness of patient records. However, studies often fail to fully describe their linkage techniques. Further, while many frameworks evaluate record linkage methods, few focus on producing gold standard datasets. This highlights a need to assess these frameworks and their real-world performance. We use real-world datasets and expand upon previous frameworks to evaluate a consistent approach to the manual review of gold standard datasets and measure its impact on algorithm performance. We applied the framework, which includes elements for data description, reviewer training and adjudication, and software and reviewer descriptions, to four datasets. Record pairs were formed and between 15,000 and 16,500 records were randomly sampled from these pairs. After training, two reviewers determined match status for each record pair. If reviewers disagreed, a third reviewer was used for final adjudication. Between the four datasets, the percent discordant rate ranged from 1.8 to 13.6%. While reviewers' discordance rate typically ranged between 1 and 5%, one exhibited a 59% discordance rate, showing the importance of the third reviewer. The original analysis was compared with three sensitivity analyses. The original analysis most often exhibited the highest predictive values compared with the sensitivity analyses. Reviewers vary in their assessment of a gold standard, which can lead to variances in estimates for matching performance. Our analysis demonstrates how a multireviewer process can be applied to create gold standards, identify reviewer discrepancies, and evaluate algorithm performance.

Variable selection for latent class analysis in the presence of missing data with application to record linkage.

The Fellegi-Sunter model is a latent class model widely used in probabilistic linkage to identify records that belong to the same entity. Record linkage practitioners typically employ all available matching fields in the model with the premise that more fields convey greater information about the true match status and hence result in improved match performance. In the context of model-based clustering, it is well known that such a premise is incorrect and the inclusion of noisy variables could compromise the clustering. Variable selection procedures have therefore been developed to remove noisy variables. Although these procedures have the potential to improve record matching, they cannot be applied directly due to the ubiquity of the missing data in record linkage applications. In this paper, we modify the stepwise variable selection procedure proposed by Fop, Smart, and Murphy and extend it to account for missing data common in record linkage. Through simulation studies, our proposed method is shown to select the correct set of matching fields across various settings, leading to better-performing algorithms. The improved match performance is also seen in a real-world application. We therefore recommend the use of our proposed selection procedure to identify informative matching fields for probabilistic record linkage algorithms.

Overdose and mortality risk following a non-fatal opioid overdose treated by Emergency Medical Services in King County, Washington

BackgroundEmergency Medical Services (EMS) agencies respond to hundreds of thousands of acute overdose events each year. We conducted a retrospective cohort study of EMS patients who survived a prior opioid overdose in 2019–2021 in King County, Washington. MethodsA novel record linkage algorithm was applied to EMS electronic health records and the state vital statistics registry to identify repeat overdoses and deaths that occurred up to 3 years following the index opioid overdose. We measured overdose incidence rates and applied survival analysis techniques to assess all-cause and overdose-specific mortality risks. ResultsIn the year following the index opioid overdose, the overdose (fatal or non-fatal) incidence rate was 23.3 per 100 person-year, overdose mortality rate was 2.7 per 100 person-year, and all-cause mortality rate was 5.2 per 100 person-year in this cohort of overdose survivors (n=4234). Overdose incidence was highest in the first 30 days following the index overdose (43 opioid overdoses and 4 fatal overdoses per 1000 person-months), declined precipitously, and then plateaued from the third month onwards (10–15 opioid overdoses and 1–2 fatal overdoses per 1000 person-months). Overdose incidence rates, measured at 30 days, were highest among overdose survivors who were young, male, and experienced a low severity index opioid overdose, but these differences diminished when measured at 12 months. ConclusionsAmong EMS patients who survived an opioid overdose, the risk of subsequent overdose is high, especially in the weeks following the index opioid overdose. Non-fatal overdose may represent a pivotal time to connect patients with harm-reduction, treatment, and other support services.

Comparison of record linkage software for deduplicating patient identities in California's Prescription Drug Monitoring Program.

To help prevent overdose deaths involving prescription drugs, accurate linkage of prescription drug monitoring program (PDMP) records for individual patients is essential. To compare the accuracy of the linkage program used by California's PDMP against various record linkage programs with respect to accuracy in deduplicating patient identities in the PDMP, with implications for identifying high-risk opioid use and outlier behaviors. We evaluated California's program, Link Plus, LinkSolv, and The Link King on 557 861 PDMP identity records with addresses in two 3-digit zip code areas for patients who filled a controlled substance prescription in 2013. Manual review was performed on a stratified sample of 720 paired records identified as matches by at least one program. We estimated sensitivity and positive predictive value, and computed PDMP patient alerts for the patient entities identified by each program. Sensitivity was 95% for LinkSolv and The Link King, 84% for Link Plus, and 73% for California's program; positive predictive value was ≥93% for all programs. The number of patient entities prompting a PDMP alert was similar among the programs for all alerts except multiple provider episodes (obtaining prescriptions from ≥6 prescribers or ≥6 pharmacies in the last 6 months), which were 10.9%, 26.6%, and 16.9% greater using The Link King, Link Plus, and LinkSolv, respectively, compared to California's program. PDMPs should assess the accuracy of record linkage algorithms and the impacts of these algorithms on patient safety alerts and develop national best practices for PDMP record linkage.

Bayesian record linkage with variables in one file.

In many healthcare and social science applications, information about units is dispersed across multiple data files. Linking records across files is necessary to estimate the associations of interest. Common record linkage algorithms only rely on similarities between linking variables that appear in all the files. Moreover, analysis of linked files often ignores errors that may arise from incorrect or missed links. Bayesian record linking methods allow for natural propagation of linkage error, by jointly sampling the linkage structure and the model parameters. We extend an existing Bayesian record linkage method to integrate associations between variables exclusive to each file being linked. We show analytically, and using simulations, that the proposed method can improve the linking process, and can result in accurate inferences. We apply the method to link Meals on Wheels recipients to Medicare enrollment records.

Optimizing Patient Record Linkage in a Master Patient Index Using Machine Learning: Algorithm Development and Validation

Background To provide quality care, modern health care systems must match and link data about the same patient from multiple sources, a function often served by master patient index (MPI) software. Record linkage in the MPI is typically performed manually by health care providers, guided by automated matching algorithms. These matching algorithms must be configured in advance, such as by setting the weights of patient attributes, usually by someone with knowledge of both the matching algorithm and the patient population being served. Objective We aimed to develop and evaluate a machine learning–based software tool, which automatically configures a patient matching algorithm by learning from pairs of patient records previously linked by humans already present in the database. Methods We built a free and open-source software tool to optimize record linkage algorithm parameters based on historical record linkages. The tool uses Bayesian optimization to identify the set of configuration parameters that lead to optimal matching performance in a given patient population, by learning from prior record linkages by humans. The tool is written assuming only the existence of a minimal HTTP application programming interface (API), and so is agnostic to the choice of MPI software, record linkage algorithm, and patient population. As a proof of concept, we integrated our tool with SantéMPI, an open-source MPI. We validated the tool using several synthetic patient populations in SantéMPI by comparing the performance of the optimized configuration in held-out data to SantéMPI’s default matching configuration using sensitivity and specificity. Results The machine learning–optimized configurations correctly detect over 90% of true record linkages as definite matches in all data sets, with 100% specificity and positive predictive value in all data sets, whereas the baseline detects none. In the largest data set examined, the baseline matching configuration detects possible record linkages with a sensitivity of 90.2% (95% CI 88.4%-92.0%) and specificity of 100%. By comparison, the machine learning–optimized matching configuration attains a sensitivity of 100%, with a decreased specificity of 95.9% (95% CI 95.9%-96.0%). We report significant gains in sensitivity in all data sets examined, at the cost of only marginally decreased specificity. The configuration optimization tool, data, and data set generator have been made freely available. Conclusions Our machine learning software tool can be used to significantly improve the performance of existing record linkage algorithms, without knowledge of the algorithm being used or specific details of the patient population being served.

No ground truth? No problem: Improving administrative data linking using active learning and a little bit of guile.

While linking records across large administrative datasets ["big data"] has the potential to revolutionize empirical social science research, many administrative data files do not have common identifiers and are thus not designed to be linked to others. To address this problem, researchers have developed probabilistic record linkage algorithms which use statistical patterns in identifying characteristics to perform linking tasks. Naturally, the accuracy of a candidate linking algorithm can be substantially improved when an algorithm has access to "ground-truth" examples-matches which can be validated using institutional knowledge or auxiliary data. Unfortunately, the cost of obtaining these examples is typically high, often requiring a researcher to manually review pairs of records in order to make an informed judgement about whether they are a match. When a pool of ground-truth information is unavailable, researchers can use "active learning" algorithms for linking, which ask the user to provide ground-truth information for select candidate pairs. In this paper, we investigate the value of providing ground-truth examples via active learning for linking performance. We confirm popular intuition that data linking can be dramatically improved with the availability of ground truth examples. But critically, in many real-world applications, only a relatively small number of tactically-selected ground-truth examples are needed to obtain most of the achievable gains. With a modest investment in ground truth, researchers can approximate the performance of a supervised learning algorithm that has access to a large database of ground truth examples using a readily available off-the-shelf tool.

A fast privacy-preserving patient record linkage of time series data

Recent advances in technology have led to an explosion of data in virtually all domains of our lives. Modern biomedical devices can acquire a large number of physical readings from patients. Often, these readings are stored in the form of time series data. Such time series data can form the basis for important research to advance healthcare and well being. Due to several considerations including data size, patient privacy, etc., the original, full data may not be available to secondary parties or researchers. Instead, suppose that a subset of the data is made available. A fast and reliable record linkage algorithm enables us to accurately match patient records in the original and subset databases while maintaining privacy. The problem of record linkage when the attributes include time series has not been studied much in the literature. We introduce two main contributions in this paper. First, we propose a novel, very efficient, and scalable record linkage algorithm that is employed on time series data. This algorithm is 400× faster than the previous work. Second, we introduce a privacy preserving framework that enables health institutions to safely release their raw time series records to researchers with bare minimum amount of identifying information.

Piloting and validating the linkage of routine ART and laboratory records in an HIV endemic setting.

ObjectivesIntegrating clinical and laboratory data systems offers opportunities to understand disease aetiology and improve patient outcomes. We developed and piloted a data linkage algorithm to link HIV patient treatment records with routine laboratory service records in an HIV endemic setting in rural South Africa. ApproachWe used data from the South Africa’s National Health Laboratory Service (NHLS) and Three Integrated Electronic Registers (TIER.Net) databases for 17 primary healthcare clinics from uMkhanyakude district, South Africa. The two databases contained data on key demographic variables such as patient first name, surname, date of birth, sex, and health facility, and national ID numbers for a subset of the data. We adapted a probabilistic record linkage algorithm we previously developed for linkage of laboratory results. We trained and validated our algorithm using two approaches: 1) firstly, we constructed a quasi-gold standard based on manual review of potential matches for 1,061 randomly selected patients; 2) secondly, we constructed a quasi-gold standard based on records that contained national ID numbers. We calculated the probabilities that the algorithm could correctly identify a true match (sensitivity) and the probabilities that the match identified by the algorithm was truly a match (positive predictive value) using the two approaches. ResultsBetween 2015 and 2020, 55,077 patients were recorded in the HIV treatment database and 386,577 laboratory tests were recorded in the laboratory service database. The sensitivity and positive predictive value in the manually reviewed data were estimated to be 94.4% and 81.2% respectively. The sensitivity and positive predictive value in the matched individual identifiers were estimated to be 99.6% and 99.7% respectively. ConclusionsRecords can be linked successfully, but estimated performance of record linkage depends on the validation set used. Manually reviewed data contain noise and may underestimate performance, while national ID numbers may overestimate performance due to non-random patterns of missingness.

Which Conference Is That? A Case Study in Computer Science

Conferences play a major role in some disciplines such as computer science and are often used in research quality evaluation exercises. Differently from journals and books, for which ISSN and ISBN codes provide unambiguous keys, recognizing the conference series in which a paper was published is a rather complex endeavor: There is no unique code assigned to conferences, and the way their names are written may greatly vary across years and catalogs. In this article, we propose a technique for the entity resolution of conferences based on the analysis of different semantic parts of their names. We present the results of an investigation of our technique on a dataset of 42,395 distinct computer science conference names excerpted from the DBLP computer science repository, 1 which we automatically link to different authority files. With suitable data cleaning, the precision of our record linkage algorithm can be as high as 94%. A comparison with results obtainable using state-of-the-art general-purpose record linkage algorithms rounds off the article, showing that our ad hoc solution largely outperforms them in terms of the quality of the results.

FIRLA: a Fast Incremental Record Linkage Algorithm

Record linkage is an important problem studied widely in many domains including biomedical informatics. A standard version of this problem is to cluster records from several datasets, such that each cluster has records pertinent to just one individual. Typically, datasets are huge in size. Hence, existing record linkage algorithms take a very long time. It is thus essential to develop novel fast algorithms for record linkage. The incremental version of this problem is to link previously clustered records with new records added to the input datasets.A novel algorithm has been created to efficiently perform standard and incremental record linkage. This algorithm leverages a set of efficient techniques that significantly restrict the number of record pair comparisons and distance computations. Our algorithm shows an average speed-up of 2.4x (up to 4x) for the standard linkage problem as compared to the state-of-the-art, without any drop in linkage performance at all. On average, our algorithm can incrementally link records in just 33% of the time required for linking them from scratch.Our algorithms achieve comparable or superior linkage performance and outperform the state-of-the-art in terms of linking time in all cases where the number of comparison attributes is greater than two. In practice, more than two comparison attributes are quite common. The proposed algorithm is very efficient and could be used in practice for record linkage applications especially when records are being added over time and linkage output needs to be updated frequently.

Fairness and Cost Constrained Privacy-Aware Record Linkage

Record linkage algorithms match and link records from different databases that refer to the same real-world entity based on direct and/or quasi-identifiers, such as name, address, age, and gender, available in the records. Since these identifiers generally contain personal identifiable information (PII) about the entities, record linkage algorithms need to be developed with privacy constraints. Known as privacy-preserving record linkage (PPRL), many research studies have been conducted to perform the linkage on encoded and/or encrypted identifiers. Differential privacy (DP) combined with computationally efficient encoding methods, e.g. Bloom filter encoding, has been used to develop PPRL with provable privacy guarantees. The standard DP notion does not however address other constraints, among which the most important ones are fairness-bias and cost of linkage in terms of number of record pairs to be compared. In this work, we propose new notions of fairness-constrained DP and fairness and cost-constrained DP for PPRL and develop a framework for PPRL with these new notions of DP combined with Bloom filter encoding. We provide theoretical proofs for the new DP notions for fairness and cost-constrained PPRL and experimentally evaluate them on two datasets containing person-specific data. Our experimental results show that with these new notions of DP, PPRL with better performance (compared to the standard DP notion for PPRL) can be achieved with regard to privacy, cost and fairness constraints.

Factors Associated with Unsuccessful Admissions to Inpatient Rehabilitation Facilities Among Older Adults With Traumatic Brain Injury

Research Objectives Older adults comprise a growing proportion of patients with TBI in inpatient rehabilitation facilities (IRF). However, concerns about their tolerance of rehabilitation-intensive IRF care remain. Centers for Medicare & Medicaid defined ‘short stays’ (Length of stay (LOS) < 3 days) and ‘short-stay transfers’ (transfers to an institutional setting with a shorter than expected LOS) suggest a limited benefit from an IRF admission. Our objective is to examine the relationship between age, chronic conditions, functional status, and the odds of an ‘unsuccessful IRF admission’ among older adults with TBI. Design In this retrospective cohort study we used Medicare administrative data linked to the National Trauma Data Bank via a Bayesian record linkage algorithm to examine the relationship between age, chronic conditions, functional status, and an ‘unsuccessful IRF admission.’ We used a two-level hierarchical logistic regression model, adjusting for demographics, injury characteristics, and facility effects. Setting U.S. Inpatient Rehabilitation Facilities. Participants Medicare fee-for-service beneficiaries hospitalized with TBI between 2011 and 2015. Interventions N/A. Main Outcome Measures Unsuccessful IRF admission (‘short stays’ or ‘short-stay transfers’). Results Cognitive function in the lowest quartile (OR:2.5 (1.0, 6.3), or motor function in the lowest two quartiles (OR:3.1(1.4, 7.1), OR:6.2 (2.4, 15.6)) were associated with higher odds of an ‘unsuccessful IRF admission’ compared to the highest quartile of each domain. The model differentiated ‘unsuccessful IRF admissions’ (AUC:0.74 (0.69, 0.80), however, 46% of those in cognitive quartile 1, and 52% of those in motor quartiles 1 and 2 were able to discharge home. Conclusions Among older adults with TBI, significant impairments in cognitive and motor function are associated with increased risk of an ‘unsuccessful IRF admission’. Factors unaccounted for in our models, potentially acute management characteristics, may help to differentiate those who are unlikely to benefit from IRF care. Author(s) Disclosures Co-Author Roee Gutman has served as an expert witness for Johnson & Johnson. The authors have no other conflicts of interest to report. Older adults comprise a growing proportion of patients with TBI in inpatient rehabilitation facilities (IRF). However, concerns about their tolerance of rehabilitation-intensive IRF care remain. Centers for Medicare & Medicaid defined ‘short stays’ (Length of stay (LOS) < 3 days) and ‘short-stay transfers’ (transfers to an institutional setting with a shorter than expected LOS) suggest a limited benefit from an IRF admission. Our objective is to examine the relationship between age, chronic conditions, functional status, and the odds of an ‘unsuccessful IRF admission’ among older adults with TBI. In this retrospective cohort study we used Medicare administrative data linked to the National Trauma Data Bank via a Bayesian record linkage algorithm to examine the relationship between age, chronic conditions, functional status, and an ‘unsuccessful IRF admission.’ We used a two-level hierarchical logistic regression model, adjusting for demographics, injury characteristics, and facility effects. U.S. Inpatient Rehabilitation Facilities. Medicare fee-for-service beneficiaries hospitalized with TBI between 2011 and 2015. N/A. Unsuccessful IRF admission (‘short stays’ or ‘short-stay transfers’). Cognitive function in the lowest quartile (OR:2.5 (1.0, 6.3), or motor function in the lowest two quartiles (OR:3.1(1.4, 7.1), OR:6.2 (2.4, 15.6)) were associated with higher odds of an ‘unsuccessful IRF admission’ compared to the highest quartile of each domain. The model differentiated ‘unsuccessful IRF admissions’ (AUC:0.74 (0.69, 0.80), however, 46% of those in cognitive quartile 1, and 52% of those in motor quartiles 1 and 2 were able to discharge home. Among older adults with TBI, significant impairments in cognitive and motor function are associated with increased risk of an ‘unsuccessful IRF admission’. Factors unaccounted for in our models, potentially acute management characteristics, may help to differentiate those who are unlikely to benefit from IRF care.

Agreement Between Injury Severity Scores Calculation Methods Among Older Adults with Traumatic Brain Injury

Research Objectives Medicare administrative data is a valuable resource for understanding post-acute care outcomes of older adults. However, these records lack information about injury severity that is useful for understanding post-acute care among those with TBI. The Injury Severity Score (ISS) reflects anatomic injury severity and is calculated by trauma registries, but can also be calculated from ICD-9 codes available in administrative data. Our objective is to examine the agreement between ICD-9 derived ISS scores and those sourced from trauma registries. Design Design: We linked Medicare administrative data to the National Trauma Data Bank (NTDB) using a Bayesian record linkage algorithm. We then examined the agreement between trauma registry- ISS scores and those calculated from acute and post-acute ICD-9 codes using Injury Categorization Software (ICDPIC) software. Setting U.S. Inpatient Rehabilitation Facilities (IRF) and Skilled Nursing Facilities (SNF). Participants Medicare beneficiaries who discharged to a SNF or an IRF following hospitalization for TBI between 2011 and 2015. Interventions N/A. Main Outcome Measures Injury Severity Score. Results For scores derived from acute claims, the absolute agreement with trauma registry scores was 11.2% (8.3, 14.2), with 41.8% (37.4, 46.2) of scores being within +/- 5 points. Trauma registry sourced ISS scores were more often higher than the ISS from the trauma registry. In the majority of cases ISS scores could not be calculated from post-acute claims, due to a lack of injury specific ICD-9 codes. Conclusions There is limited agreement between trauma registry ISS scores and those calculated from the ICD-9 codes of acute claims. The scores derived from acute claims may have utility for adjustment, but may underestimate anatomic injury severity of older adults with TBI. Post-acute claims lack the detailed injury codes required for calculation of ISS scores. Author(s) Disclosures Co-author Roee Gutman has served as an expert witness for Johnson & Johnson. The authors have no other conflicts of interest to report. Medicare administrative data is a valuable resource for understanding post-acute care outcomes of older adults. However, these records lack information about injury severity that is useful for understanding post-acute care among those with TBI. The Injury Severity Score (ISS) reflects anatomic injury severity and is calculated by trauma registries, but can also be calculated from ICD-9 codes available in administrative data. Our objective is to examine the agreement between ICD-9 derived ISS scores and those sourced from trauma registries. Design: We linked Medicare administrative data to the National Trauma Data Bank (NTDB) using a Bayesian record linkage algorithm. We then examined the agreement between trauma registry- ISS scores and those calculated from acute and post-acute ICD-9 codes using Injury Categorization Software (ICDPIC) software. U.S. Inpatient Rehabilitation Facilities (IRF) and Skilled Nursing Facilities (SNF). Medicare beneficiaries who discharged to a SNF or an IRF following hospitalization for TBI between 2011 and 2015. N/A. Injury Severity Score. For scores derived from acute claims, the absolute agreement with trauma registry scores was 11.2% (8.3, 14.2), with 41.8% (37.4, 46.2) of scores being within +/- 5 points. Trauma registry sourced ISS scores were more often higher than the ISS from the trauma registry. In the majority of cases ISS scores could not be calculated from post-acute claims, due to a lack of injury specific ICD-9 codes. There is limited agreement between trauma registry ISS scores and those calculated from the ICD-9 codes of acute claims. The scores derived from acute claims may have utility for adjustment, but may underestimate anatomic injury severity of older adults with TBI. Post-acute claims lack the detailed injury codes required for calculation of ISS scores.