skip to main content

Explore Research Products in the NSF-PAR

Title: Comparing Methods for Record Linkage for Public Health Action: Matching Algorithm Validation Study
Background Many public health departments use record linkage between surveillance data and external data sources to inform public health interventions. However, little guidance is available to inform these activities, and many health departments rely on deterministic algorithms that may miss many true matches. In the context of public health action, these missed matches lead to missed opportunities to deliver interventions and may exacerbate existing health inequities. Objective This study aimed to compare the performance of record linkage algorithms commonly used in public health practice. Methods We compared five deterministic (exact, Stenger, Ocampo 1, Ocampo 2, and Bosh) and two probabilistic record linkage algorithms (fastLink and beta record linkage [BRL]) using simulations and a real-world scenario. We simulated pairs of datasets with varying numbers of errors per record and the number of matching records between the two datasets (ie, overlap). We matched the datasets using each algorithm and calculated their recall (ie, sensitivity, the proportion of true matches identified by the algorithm) and precision (ie, positive predictive value, the proportion of matches identified by the algorithm that were true matches). We estimated the average computation time by performing a match with each algorithm 20 times while varying the size of the datasets being matched. In a real-world scenario, HIV and sexually transmitted disease surveillance data from King County, Washington, were matched to identify people living with HIV who had a syphilis diagnosis in 2017. We calculated the recall and precision of each algorithm compared with a composite standard based on the agreement in matching decisions across all the algorithms and manual review. Results In simulations, BRL and fastLink maintained a high recall at nearly all data quality levels, while being comparable with deterministic algorithms in terms of precision. Deterministic algorithms typically failed to identify matches in scenarios with low data quality. All the deterministic algorithms had a shorter average computation time than the probabilistic algorithms. BRL had the slowest overall computation time (14 min when both datasets contained 2000 records). In the real-world scenario, BRL had the lowest trade-off between recall (309/309, 100.0%) and precision (309/312, 99.0%). Conclusions Probabilistic record linkage algorithms maximize the number of true matches identified, reducing gaps in the coverage of interventions and maximizing the reach of public health action.  more » « less
Award ID(s):
1852841
NSF-PAR ID:
10389231
Author(s) / Creator(s):
; ; ; ; ; ;
Date Published:
Journal Name:
JMIR Public Health and Surveillance
Volume:
6
Issue:
2
ISSN:
2369-2960
Page Range / eLocation ID:
e15917
Format(s):
Medium: X
Sponsoring Org:
National Science Foundation
More Like this
  1. ; ; ; ; ; ; ; ; ; ; et al ( , mSystems)
    Marshall, Christopher W. (Ed.)
    ABSTRACT Identification of genes encoding β-lactamases (BLs) from short-read sequences remains challenging due to the high frequency of shared amino acid functional domains and motifs in proteins encoded by BL genes and related non-BL gene sequences. Divergent BL homologs can be frequently missed during similarity searches, which has important practical consequences for monitoring antibiotic resistance. To address this limitation, we built ROCker models that targeted broad classes (e.g., class A, B, C, and D) and individual families (e.g., TEM) of BLs and challenged them with mock 150-bp- and 250-bp-read data sets of known composition. ROCker identifies most-discriminant bit score thresholds in sliding windows along the sequence of the target protein sequence and hence can account for nondiscriminative domains shared by unrelated proteins. BL ROCker models showed a 0% false-positive rate (FPR), a 0% to 4% false-negative rate (FNR), and an up-to-50-fold-higher F1 score [2 × precision × recall/(precision + recall)] compared to alternative methods, such as similarity searches using BLASTx with various e-value thresholds and BL hidden Markov models, or tools like DeepARG, ShortBRED, and AMRFinder. The ROCker models and the underlying protein sequence reference data sets and phylogenetic trees for read placement are freely available through http://enve-omics.ce.gatech.edu/data/rocker-bla . Application of these BL ROCker models to metagenomics, metatranscriptomics, and high-throughput PCR gene amplicon data should facilitate the reliable detection and quantification of BL variants encoded by environmental or clinical isolates and microbiomes and more accurate assessment of the associated public health risk, compared to the current practice. IMPORTANCE Resistance genes encoding β-lactamases (BLs) confer resistance to the widely prescribed antibiotic class β-lactams. Therefore, it is important to assess the prevalence of BL genes in clinical or environmental samples for monitoring the spreading of these genes into pathogens and estimating public health risk. However, detecting BLs in short-read sequence data is technically challenging. Our ROCker model-based bioinformatics approach showcases the reliable detection and typing of BLs in complex data sets and thus contributes toward solving an important problem in antibiotic resistance surveillance. The ROCker models developed substantially expand the toolbox for monitoring antibiotic resistance in clinical or environmental settings. 
    more » « less
  2. ; ; ; ; ; ( , Journal of Medical Internet Research)
    Background Social networks such as Twitter offer the clinical research community a novel opportunity for engaging potential study participants based on user activity data. However, the availability of public social media data has led to new ethical challenges about respecting user privacy and the appropriateness of monitoring social media for clinical trial recruitment. Researchers have voiced the need for involving users’ perspectives in the development of ethical norms and regulations. Objective This study examined the attitudes and level of concern among Twitter users and nonusers about using Twitter for monitoring social media users and their conversations to recruit potential clinical trial participants. Methods We used two online methods for recruiting study participants: the open survey was (1) advertised on Twitter between May 23 and June 8, 2017, and (2) deployed on TurkPrime, a crowdsourcing data acquisition platform, between May 23 and June 8, 2017. Eligible participants were adults, 18 years of age or older, who lived in the United States. People with and without Twitter accounts were included in the study. Results While nearly half the respondents—on Twitter (94/603, 15.6%) and on TurkPrime (509/603, 84.4%)—indicated agreement that social media monitoring constitutes a form of eavesdropping that invades their privacy, over one-third disagreed and nearly 1 in 5 had no opinion. A chi-square test revealed a positive relationship between respondents’ general privacy concern and their average concern about Internet research (P<.005). We found associations between respondents’ Twitter literacy and their concerns about the ability for researchers to monitor their Twitter activity for clinical trial recruitment (P=.001) and whether they consider Twitter monitoring for clinical trial recruitment as eavesdropping (P<.001) and an invasion of privacy (P=.003). As Twitter literacy increased, so did people’s concerns about researchers monitoring Twitter activity. Our data support the previously suggested use of the nonexceptionalist methodology for assessing social media in research, insofar as social media-based recruitment does not need to be considered exceptional and, for most, it is considered preferable to traditional in-person interventions at physical clinics. The expressed attitudes were highly contextual, depending on factors such as the type of disease or health topic (eg, HIV/AIDS vs obesity vs smoking), the entity or person monitoring users on Twitter, and the monitored information. Conclusions The data and findings from this study contribute to the critical dialogue with the public about the use of social media in clinical research. The findings suggest that most users do not think that monitoring Twitter for clinical trial recruitment constitutes inappropriate surveillance or a violation of privacy. However, researchers should remain mindful that some participants might find social media monitoring problematic when connected with certain conditions or health topics. Further research should isolate factors that influence the level of concern among social media users across platforms and populations and inform the development of more clear and consistent guidelines. 
    more » « less
  3. ; ; ; ( , JMIR Public Health and Surveillance)
    null (Ed.)
    Background Significant uncertainty has existed about the safety of reopening college and university campuses before the COVID-19 pandemic is better controlled. Moreover, little is known about the effects that on-campus students may have on local higher-risk communities. Objective We aimed to estimate the range of potential community and campus COVID-19 exposures, infections, and mortality under various university reopening plans and uncertainties. Methods We developed campus-only, community-only, and campus × community epidemic differential equations and agent-based models, with inputs estimated via published and grey literature, expert opinion, and parameter search algorithms. Campus opening plans (spanning fully open, hybrid, and fully virtual approaches) were identified from websites and publications. Additional student and community exposures, infections, and mortality over 16-week semesters were estimated under each scenario, with 10% trimmed medians, standard deviations, and probability intervals computed to omit extreme outliers. Sensitivity analyses were conducted to inform potential effective interventions. Results Predicted 16-week campus and additional community exposures, infections, and mortality for the base case with no precautions (or negligible compliance) varied significantly from their medians (4- to 10-fold). Over 5% of on-campus students were infected after a mean of 76 (SD 17) days, with the greatest increase (first inflection point) occurring on average on day 84 (SD 10.2 days) of the semester and with total additional community exposures, infections, and mortality ranging from 1-187, 13-820, and 1-21 per 10,000 residents, respectively. Reopening precautions reduced infections by 24%-26% and mortality by 36%-50% in both populations. Beyond campus and community reproductive numbers, sensitivity analysis indicated no dominant factors that interventions could primarily target to reduce the magnitude and variability in outcomes, suggesting the importance of comprehensive public health measures and surveillance. Conclusions Community and campus COVID-19 exposures, infections, and mortality resulting from reopening campuses are highly unpredictable regardless of precautions. Public health implications include the need for effective surveillance and flexible campus operations. 
    more » « less
  4. ; ; ; ; ; ; ( , International Conference on Privacy in Statistical Databases)
    The COVID-19 pandemic highlights the need for broad dissemination of case surveillance data. Local and global public health agencies have initiated efforts to do so, but there remains limited data available, due in part to concerns over privacy. As a result, current COVID-19 case surveillance data sharing policies are based on strong adversarial assumptions, such as the expectation that an attacker can readily re-identify individuals based on their distinguishability in a dataset. There are various re-identification risk measures to account for adversarial capabilities; however, the current array insufficiently accounts for real world data challenges - particularly issues of missing records in resources of identifiable records that adversaries may rely upon to execute attacks (e.g., 10 50-year-old male in the de-identified dataset vs. 5 50-year-old male in the identified dataset). In this paper, we introduce several approaches to amend such risk measures and assess re-identification risk in light of how an attacker's capabilities relate to missing records. We demonstrate the potential for these measures through a record linkage attack using COVID-19 case surveillance data and voter registration records in the state of Florida. Our findings demonstrate that adversarial assumptions, as realized in a risk measure, can dramatically affect re-identification risk estimation. Notably, we show that the re-identification risk is likely to be substantially smaller than the typical risk thresholds, which suggests that more detailed data could be shared publicly than is currently the case. 
    more » « less
  5. ; ; ; ; ; ; ; ( , Infection Control & Hospital Epidemiology)
    null (Ed.)
    Background: In spring of 2019, 2 positive sputum cases of Pseudomonas aeruginosa in the cardiac critical care unit (CCU) were reported to the UFHJ infection prevention (IP) department. The initial 2 cases, detected within 3 days of each other, were followed shortly by a third case. Epidemiological evidence was initially consistent with a hospital-acquired infection (HAI): 2 of the 3 patients roomed next to each other, and all 3 patients were ventilated, 2 of whom shared the same respiratory therapist. However, no other changes in routine or equipment were noted. The samples were cultured and processed using Illumina NGS technology, generating 1–2 million short (ie, 250-bp) reads across the P. aeruginosa genome. As an additional positive control, 8 P . aeruginosa NGS data sets, previously shown to be from a single outbreak in a UK facility, were included. Reads were mapped back to a reference sequence, and single-nucleotide polymorphisms (SNPs) between each sample and the reference were extracted. Genetic distances (ie, the number of unshared SNPs) between all UFHJ and UK samples were calculated. Genetic linkage was determined using hierarchical clustering, based on a commonly used threshold of 40 SNPs. All UFHJ patient samples were separated by >18,000 SNPs, indicating genetically distinct samples from separate sources. In contrast, UK samples were separated from each other by <16 SNPs, consistent with genetic linkage and a single outbreak. Furthermore, the UFHJ samples were separated from the UK samples by >17,000 SNPs, indicating a lack of geographical distinction of the UFHJ samples (Fig. 1). These results demonstrated that while the initial epidemiological evidence pointed towards a single HAI, the high-precision and relatively inexpensive ( more » « less
  • Citation Formats
  • MLA
  • APA
  • Chicago
  • BibTeX
  • Save / Share this Record
  • Send to Email