Impact of linkage quality on inferences drawn from analyses using data with high rates of linkage errors in rural Tanzania

BACKGROUND

Studies based on high-quality linked data in developed countries show that even minor linkage errors, which occur when records of two different individuals are erroneously linked or when records belonging to the same individual are not linked, can impact bias and precision of subsequent analyses. We evaluated the impact of linkage quality on inferences drawn from analyses using data with substantial linkage errors in rural Tanzania.

METHODS

Semi-automatic point-of-contact interactive record linkage was used to establish gold standard links between community-based HIV surveillance data and medical records at clinics serving the surveillance population. Automated probabilistic record linkage was used to create analytic datasets at minimum, low, medium, and high match score thresholds. Cox proportional hazards regression models were used to compare HIV care registration rates by testing modality (sero-survey vs. clinic) in each analytic dataset. We assessed linkage quality using three approaches: quantifying linkage errors, comparing characteristics between linked and unlinked data, and evaluating bias and precision of regression estimates.

RESULTS

Between 2014 and 2017, 405 individuals with gold standard links were newly diagnosed with HIV in sero-surveys (n = 263) and clinics (n = 142). Automated probabilistic linkage correctly identified 233 individuals (positive predictive value [PPV] = 65%) at the low threshold and 95 individuals (PPV = 90%) at the high threshold. Significant differences were found between linked and unlinked records in primary exposure and outcome variables and for adjusting covariates at every threshold. As expected, differences attenuated with increasing threshold. Testing modality was significantly associated with time to registration in the gold standard data (adjusted hazard ratio [HR] 4.98 for clinic-based testing, 95% confidence interval [CI] 3.34, 7.42). Increasing false matches weakened the association (HR 2.76 at minimum match score threshold, 95% CI 1.73, 4.41). Increasing missed matches (i.e., increasing match score threshold and positive predictive value of the linkage algorithm) was strongly correlated with a reduction in the precision of coefficient estimate (R2 = 0.97; p = 0.03).

CONCLUSIONS

Similar to studies with more negligible levels of linkage errors, false matches in this setting reduced the magnitude of the association; missed matches reduced precision. Adjusting for these biases could provide more robust results using data with considerable linkage errors.

A systematic review of the clinical application of data-driven population segmentation analysis

BACKGROUND

Data-driven population segmentation analysis utilizes data analytics to divide a heterogeneous population into parsimonious and relatively homogenous groups with similar healthcare characteristics. It is a promising patient-centric analysis that enables effective integrated healthcare interventions specific for each segment. Although widely applied, there is no systematic review on the clinical application of data-driven population segmentation analysis.

METHODS

We carried out a systematic literature search using PubMed, Embase and Web of Science following PRISMA criteria. We included English peer-reviewed articles that applied data-driven population segmentation analysis on empirical health data. We summarized the clinical settings in which segmentation analysis was applied, compared and contrasted strengths, limitations, and practical considerations of different segmentation methods, and assessed the segmentation outcome of all included studies. The studies were assessed by two independent reviewers.

RESULTS

We retrieved 14,514 articles and included 216 articles. Data-driven population segmentation analysis was widely used in different clinical contexts. 163 studies examined the general population while 53 focused on specific population with certain diseases or conditions, including psychological, oncological, respiratory, cardiovascular, and gastrointestinal conditions. Variables used for segmentation in the studies are heterogeneous. Most studies (n = 170) utilized secondary data in community settings (n = 185). The most common segmentation method was latent class/profile/transition/growth analysis (n = 96) followed by K-means cluster analysis (n = 60) and hierarchical analysis (n = 50), each having its advantages, disadvantages, and practical considerations. We also identified key criteria to evaluate a segmentation framework: internal validity, external validity, identifiability/interpretability, substantiality, stability, actionability/accessibility, and parsimony.

CONCLUSIONS

Data-driven population segmentation has been widely applied and holds great potential in managing population health. The evaluations of segmentation outcome require the interplay of data analytics and subject matter expertise. The optimal framework for segmentation requires further research.

The Forteo Patient Registry linkage to multiple state cancer registries: study design and results from the first 8 years

The Forteo Patient Registry (FPR) aims to estimate the incidence of osteosarcoma in US patients treated with teriparatide. Enrollment began in 2009 and will continue through 2019, with linkage planned through 2024. To date, no incident cases of osteosarcoma have been identified among patients registered in the FPR.

INTRODUCTION

The Forteo Patient Registry (FPR) was established in 2009 to estimate the incidence of osteosarcoma in US patients treated with teriparatide. The objective of this paper is to describe study methods, challenges encountered, and progress to date.

METHODS

The FPR is a prospective US registry designed to link data from participants annually with state cancer registries. Patient enrollment is planned for 10 years (2009-2019) and annual linkage with US state cancer registries for 15 years (2010-2024). All US state cancer registries and DC were invited to participate. Patients are recruited using pre-enrollment materials included in teriparatide device packaging, kits, and brochures distributed by health-care providers; a toll-free number; and a study website. A linkage algorithm is used to match data from enrolled participants with cancer registry data.

RESULTS

For the eighth annual linkage in 2017, information necessary for linkage with 63,270 patients in the FPR was submitted to each of the 42 participating registries. These patients contributed approximately 242,782 person-years of follow-up. A total of 5268 adult osteosarcoma cases diagnosed since January 1, 2009, were available for linkage from participating state cancer registries. To date, no incident cases of osteosarcoma have been identified among patients registered in the FPR.

CONCLUSIONS

Based on the estimated 242,782 person-years of observation as of the eighth annual linkage and projecting current enrollment rate to study end in 2024, it is anticipated that the completed study will be able to detect a fourfold increase in the risk of osteosarcoma if one exists.

Timing of singleton births by onset of labour and mode of birth in NHS maternity units in England, 2005-2014: A study of linked birth registration, birth notification, and hospital episode data

BACKGROUND

Maternity care has to be available 24 hours a day, seven days a week. It is known that obstetric intervention can influence the time of birth, but no previous analysis at a national level in England has yet investigated in detail the ways in which the day and time of birth varies by onset of labour and mode of giving birth.

METHOD

We linked data from birth registration, birth notification, and Maternity Hospital Episode Statistics and analysed 5,093,615 singleton births in NHS maternity units in England from 2005 to 2014. We used descriptive statistics and negative binomial regression models with harmonic terms to establish how patterns of timing of birth vary by onset of labour, mode of giving birth and gestational age.

RESULTS

The timing of birth by time of day and day of the week varies considerably by onset of labour and mode of birth. Spontaneous births after spontaneous onset are more likely to occur between midnight and 6am than at other times of day, and are also slightly more likely on weekdays than at weekends and on public holidays. Elective caesarean births are concentrated onto weekday mornings. Births after induced labours are more likely to occur at hours around midnight on Tuesdays to Saturdays and on days before a public holiday period, than on Sundays, Mondays and during or just after a public holiday.

CONCLUSION

The timing of births varies by onset of labour and mode of birth and these patterns have implications for midwifery and medical staffing. Further research is needed to understand the processes behind these findings.

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.

National Health IT Services in Finland

Objectives: In 2002 a decision was reached to set up a nation-wide electronic health record system in Finland. The legal framework of actors with the necessary mandate was approved in the parliament in December 2006. A set of standards and norms have been selected that all health care actors need to follow. Functional specifications of the services were completed in 2006. Setting up the centralized health IT services begins in 2007.

Centralization of patient record data allows the reorganization of health service providers to take place at local and regional levels according to need. The services allow users to access patient records securely from anywhere with the provision that they have the right to access private patient data.

Methods: The functionality of the services and the necessary infrastructure has been agreed to in projects and working groups involving users, experts, key stakeholders and vendors.

Results: The legal framework was approved in the parliament in December 2006. The functional specifications of thecentralized health IT services were finalized in 2006.

Conclusions: The implementation of the services will start in 2007.

Errors in Survival Rates Caused by Routinely Used Deterministic Record Linkage Methods

Objective: It was the objective of this study to assess the impact of applying various record linkage methods to one of the most important outcome measures in oncological epidemiology, namely survival rates.

Methods: To assess the life status of patients, incidence data published by the Cancer Registry of Tyrol were analyzed with three routinely used methods of record linkage for incidence and mortality data. Of these methods, two were deterministic and the third a probabilistic method developed by the Cancer Registry of Tyrol. We studied the impact of record linkage methods on a simple measure (mortality rate) and a more complex measure (relative survival rate). The analysis was based on the published incidence data for Tyrol for the years 1992 to 1996. Results of deterministic record linkage methodswere simulated.

Results: The error rates for simple mortality rate and relative survival rate are considerable. For the first deterministic record linkage method, relative differences in mortality rate range from 11.9% to 14.8% (men) and 24.5% to 28.2% (women) and relative differences in relative five-year survival from 11.4% to 16.3% (men) and from 19.3% to 26.4% (women). For the second deterministic record linkage method, relative differences in mortality rate range from 4.8% to 5.9% (men) and from 4.9% to 7.4% (women), while relative differences in relative five-year survival range from 5.1% to 7.0% (men) and from 4.4% to 6.1% (women).

Conclusions: Our study shows that in order to calculate valid mortality and survival rates a probabilistic method of record linkage must be applied.

Using Web Services for Linking Genomic Data to Medical Information Systems

Objectives: To develop a new perspective for biomedical information systems, regarding the introduction of ideas, methods and tools related to the new scenario of genomic medicine.

Methods: Technological aspects related to the analysis and integration of heterogeneous clinical and genomic data include mapping clinical and genetic concepts, potential future standards or the development of integrated biomedical ontologies. In this clinicomics scenario, we describe the use of Web services technologies to improve access to and integrate different information sources. We give a concrete example of the use of Web services technologies: the Onto Fusion project.

Results: Web services provide new biomedical informatics (BMI) approaches related to genomic medicine. Customized workflowswill aid research tasks by linking heterogeneous Web services. Two significant examples of these European Commission-funded efforts are the INFOBIOMED Network of Excellence and the Advancing Clinico-Genomic Trials on Cancer (ACGT) integrated project.

Conclusions: Supplying medical researchers and practitioners with omicsdata and biologists with clinical datasets can help to develop genomic medicine. BMI is contributing by providing the informatics methods and technological infrastructure needed for these collaborative efforts.

SS-MIX: A Ministry Project to Promote Standardized Healthcare Information Exchange

Objectives: To promote healthcare information exchange between providers and to allow hospital information systems (HIS) export information in standardized format (HL7 and DICOM) in an environment of widespread legacy systems, which only can export data in proprietary format.

Methods: Through the Shizuoka prefecture EMR project in 2004–2005, followed by the ministry’s SS-MIX project, many software products have been provided, which consist of 1) a standardized storage to receive HL7 v2.5 mes sages of patient demographics, prescription orders, laboratory results, and diagnostic disease in ICD-10, 2) a referral letter creation system, 3) a formatted document creation system, 4) a progress note/nursing record system, and 5) an archive/viewer to incorporate incoming healthcare data CD and allow users to view on HIS terminal. Meanwhile, other useful applications have been produced, such as adverse event reporting and clinical information retrieval. To achieve the above-mentioned objectives, these software products were created and propagated, because users can use these software products, provided that their HIS can export the above information to the standardized storage in HL7 v2.5 format.

Results: In 20 hospitals of Japan, the standardized storage has been installed and some applications have been used. As major HIS vendors are shipping HIS with HL7 export function since 2007, HIS of 594 hospitals in Japan became capable of exporting data in HL7 v2.5 format (as of March 2010).

Conclusions: In high CPOE installation rate (85% in 400+ bed hospitals), though most of them only capable of exporting data in proprietary format, prefecture and ministry projects were effective to promote healthcare information exchange between providers. The standardized storage became an infrastructure for many useful applications, and many hospitals started using them. Ministry designation of proposed healthcare standards was effective so as to allow vendors to conform their products, and users to install them.

Can routine data be used to support cancer clinical trials? A historical baseline on which to build: retrospective linkage of data from the TACT (CRUK 01/001) breast cancer trial and the National Cancer Data Repository

BACKGROUND

Randomised clinical trials (RCTs) are the gold standard for evaluating new cancer treatments. They are, however, expensive to conduct, particularly where long-term follow-up of participants is required. Tracking participants via routine datasets could provide a cost-effective alternative for ascertaining follow-up information required to evaluate disease outcomes. This project explores the potential for routine data to inform cancer trials, using, the historical National Cancer Data Repository (NCDR) for English NHS sites and, for validation, mature data available from the TACT trial.

METHODS

Datasets were matched using patients' NHS number, date of birth (dob) and name/initials. Demographics, clinical characteristics and outcomes were assessed for agreement and completeness. Overall survival was compared between NCDR and TACT.

RESULTS

A total of 3151 patients underwent linkage; 3047 (96.7%) of which had matched records. Extensive cleaning was required for some registry data fields, e.g. cause of death, whilst others had large amounts of missing data, e.g. tumour size (22.1%). Other data had high levels of matching such as dob (99.6%) and date of death (89.6%). There was no evidence of differential survival rates (8-year survival: TACT = 75% (95% CI 73, 76); NCDR = 76% (95% CI 74, 77)).

CONCLUSIONS

Data quality and completeness requires improvement before routine data could be used for RCTs. Introduction of new routine datasets, including COSD, is welcomed although reporting of disease-recurrence events remains a concern. Prospective validation of such datasets is required before RCTs can confidently switch patient follow-up to utilise routinely collected NHS-based data.

TACT TRIAL REGISTRATION

Clinicaltrials.gov NCT00033683 , registered on 9 April 2002; ISRCTN79718493 , registered on 1 July 2001.

An observational study of the relationship between meaningful use-based electronic health information exchange, interoperability, and medication reconciliation capabilities

Stagnation in hospitals' adoption of data integration functionalities coupled with reduction in the number of operational health information exchanges could become a significant impediment to hospitals' adoption of 3 critical capabilities: electronic health information exchange, interoperability, and medication reconciliation, in which electronic systems are used to assist with resolving medication discrepancies and improving patient safety. Against this backdrop, we assessed the relationships between the 3 capabilities.We conducted an observational study applying partial least squares-structural equation modeling technique to 27 variables obtained from the 2013 American Hospital Association annual survey Information Technology (IT) supplement, which describes health IT capabilities.We included 1330 hospitals. In confirmatory factor analysis, out of the 27 variables, 15 achieved loading values greater than 0.548 at P < .001, as such were validated as the building blocks of the 3 capabilities. Subsequent path analysis showed a significant, positive, and cyclic relationship between the capabilities, in that decreases in the hospitals' adoption of one would lead to decreases in the adoption of the others.These results show that capability for high quality medication reconciliation may be impeded by lagging adoption of interoperability and health information exchange capabilities. Policies focused on improving one or more of these capabilities may have ancillary benefits.

A scaling approach to record linkage

With increasing availability of large datasets derived from administrative and other sources, there is an increasing demand for the successful linking of these to provide rich sources of data for further analysis. Variation in the quality of identifiers used to carry out linkage means that existing approaches are often based upon 'probabilistic' models, which are based on a number of assumptions, and can make heavy computational demands. In this paper, we suggest a new approach to classifying record pairs in linkage, based upon weights (scores) derived using a scaling algorithm. The proposed method does not rely on training data, is computationally fast, requires only moderate amounts of storage and has intuitive appeal. Copyright © 2017 John Wiley & Sons, Ltd.

[Extending traceability of malignant testicular tumours using hospital discharge records: an experience in Veneto Region (Northern Italy)]

OBJECTIVES

validation of codes of hospital discharge records (SDO) for identification of new cases of malignant testicular tumour in the Veneto Region (Northern Italy).

DESIGN

record linkage between the regional archive of SDO and the archive of the Veneto Tumour Registry (VTR).

SETTING AND PARTICIPANTS

extraction of cases from SDO source with ICD-9-CM 186 code for diagnosis and 62.3-62.4 codes for surgical procedure, and from VTR database using ICD-O-3 C62 code for site and 9060-9062, 9064-9066, 9070, 9071, 9080-9083, 9085, 9100, 9101 codes for morphology, with 5th digit behaviour code equal to "/3". Comparison of the two sources in a classification table using VTR data as gold standard.

MAIN OUTCOME MEASURES

positive predictive value and sensitivity of SDO, with 95% confidence interval (95%CI) based on binomial distribution.

RESULTS

from 2006 to 2008, in areas covered by the registry, SDO and VTR identified, respectively, 221 and 216 cases of testicular cancer. SDO procedure showed a sensitivity of 92% (95%CI 87%- 95%) and a positive predictive value of 90% (95%CI 85%-93%).

CONCLUSIONS

the SDO procedure can be considered an acceptable proxy for testis cancer incidence, thus allowing a wider spatiotemporal observation of the epidemiological trends.

The Building Blocks of Interoperability. A Multisite Analysis of Patient Demographic Attributes Available for Matching

BACKGROUND

Patient matching is a key barrier to achieving interoperability. Patient demographic elements must be consistently collected over time and region to be valuable elements for patient matching.

OBJECTIVES

We sought to determine what patient demographic attributes are collected at multiple institutions in the United States and see how their availability changes over time and across clinical sites.

METHODS

We compiled a list of 36 demographic elements that stakeholders previously identified as essential patient demographic attributes that should be collected for the purpose of linking patient records. We studied a convenience sample of 9 health care systems from geographically distinct sites around the country. We identified changes in the availability of individual patient demographic attributes over time and across clinical sites.

RESULTS

Several attributes were consistently available over the study period (2005-2014) including last name (99.96%), first name (99.95%), date of birth (98.82%), gender/sex (99.73%), postal code (94.71%), and full street address (94.65%). Other attributes changed significantly from 2005-2014: Social security number (SSN) availability declined from 83.3% to 50.44% (p<0.0001). Email address availability increased from 8.94% up to 54% availability (p<0.0001). Work phone number increased from 20.61% to 52.33% (p<0.0001).

CONCLUSIONS

Overall, first name, last name, date of birth, gender/sex and address were widely collected across institutional sites and over time. Availability of emerging attributes such as email and phone numbers are increasing while SSN use is declining. Understanding the relative availability of patient attributes can inform strategies for optimal matching in healthcare.

[Improving the visibility of rare diseases in health care systems by specific routine coding]

The evaluation of healthcare providers' routine data is an important basis for the analysis, planning and evaluation of measures in public health. The representation of rare diseases in the classifications that are used to record health data is not adequate. Coding rare diseases in a specific way is a challenge all around the world. There is still no general international solution for the routine coding of rare diseases.The double coding of rare diseases with ICD-10 Codes and Orphacodes is a short-term and low-cost alternative solution. Furthermore, this double coding enables international comparability. The specific encoding of rare diseases through this double coding can improve their capturing for statistical analysis and thus their visibility in healthcare systems. Nevertheless, the provision of a new classification is not enough to gather valid data. Some measures have already been adopted in Germany (and at the European level) in order to support the implementation of this double coding. Subsequently it would be possible to adopt more specific public health measures, based on better data, in order to provide better care to the more than four million people in Germany affected by rare diseases.

Linking medical and dental health record data: a partnership with the Rochester Epidemiology Project

PURPOSE

The purpose of this project was to expand the Rochester Epidemiology Project (REP) medical records linkage infrastructure to include data from oral healthcare providers. The goal of this linkage is to facilitate research studies examining the role of oral health in overall health and quality of life.

PARTICIPANTS

Eight dental practices joined the REP between 2011 and 2015. The REP study team has linked oral healthcare information with medical record information from local healthcare providers for 31 750 participants who have resided in Olmsted County, Minnesota. Overall, 17 718 (56%) participants are women, 14 318 (45%) are 40 years of age or older and 26 090 (82%) are white.

FINDINGS TO DATE

A first study using this new information was recently completed. This resource was used to determine whether the 2007 guidelines from the American Heart Association affected prescription rates of antibiotics to patients with moderate-risk cardiac conditions prior to dental procedures. The REP infrastructure was used to identify a series of patients diagnosed with moderate-risk cardiac conditions by the local healthcare providers (n=1351), and to abstract antibiotic prescriptions from dental records both pre-2007 and post-2007. Antibiotic prescriptions prior to dental procedures declined from 62% to 7% following the change in guidelines.

FUTURE PLANS

Dental data from participating practitioners will be updated on an annual basis, and new dental data will be linked to patient medical records. In addition, we will continue to invite new dental practices to participate in the REP. Finally, we will continue to use this research infrastructure to investigate associations between oral and medical health, and will present findings at conferences and in the scientific literature.

Conducting Privacy-Preserving Multivariable Propensity Score Analysis When Patient Covariate Information Is Stored in Separate Locations

Distributed networks of health-care data sources are increasingly being utilized to conduct pharmacoepidemiologic database studies. Such networks may contain data that are not physically pooled but instead are distributed horizontally (separate patients within each data source) or vertically (separate measures within each data source) in order to preserve patient privacy. While multivariable methods for the analysis of horizontally distributed data are frequently employed, few practical approaches have been put forth to deal with vertically distributed health-care databases. In this paper, we propose 2 propensity score-based approaches to vertically distributed data analysis and test their performance using 5 example studies. We found that these approaches produced point estimates close to what could be achieved without partitioning. We further found a performance benefit (i.e., lower mean squared error) for sequentially passing a propensity score through each data domain (called the "sequential approach") as compared with fitting separate domain-specific propensity scores (called the "parallel approach"). These results were validated in a small simulation study. This proof-of-concept study suggests a new multivariable analysis approach to vertically distributed health-care databases that is practical, preserves patient privacy, and warrants further investigation for use in clinical research applications that rely on health-care databases.

Requesting a unique personal identifier or providing a souvenir incentive did not affect overall consent to health record linkage: evidence from an RCT nested within a cohort

OBJECTIVE

It is unclear if unique personal identifiers should be requested from participants for health record linkage: this permits high-quality data linkage but at the potential cost of lower consent rates due to privacy concerns.

STUDY DESIGN AND SETTING

Drawing from a sampling frame based on the FAMILY Cohort, using a 2 × 2 factorial design, we randomly assigned 1,200 participants to (1) request for Hong Kong Identity Card number (HKID) or no request and (2) receiving a souvenir incentive (valued at USD4) or no incentive. The primary outcome was consent to health record linkage. We also investigated associations between demographics, health status, and postal reminders with consent.

RESULTS

Overall, we received signed consent forms from 33.3% (95% confidence interval [CI] 30.6-36.0%) of respondents. We did not find an overall effect of requesting HKID (-4.3%, 95% CI -9.8% to 1.2%) or offering souvenir incentives (2.4%, 95% CI -3.1% to 7.9%) on consent to linkage. In subgroup analyses, requesting HKID significantly reduced consent among adults aged 18-44 years (odds ratio [OR] 0.53, 95% CI 0.30-0.94, compared to no request). Souvenir incentives increased consent among women (OR 1.55, 95% CI 1.13-2.11, compared to no souvenirs).

CONCLUSIONS

Requesting a unique personal identifier or providing a souvenir incentive did not affect overall consent to health record linkage.

Combining Different Privacy-Preserving Record Linkage Methods for Hospital Admission Data

Record linkage (RL) is the process of identifying pairs of records that correspond to the same entity, for example the same patient. The basic approach assigns to each pair of records a similarity weight, and then determines a certain threshold, above which the two records are considered to be a match. Three different RL methods were applied under privacy-preserving conditions on hospital admission data: deterministic RL (DRL), probabilistic RL (PRL), and Bloom filters. The patient characteristics like names were one-way encrypted (DRL, PRL) or transformed to a cryptographic longterm key (Bloom filters). Based on one year of hospital admissions, the data set was split randomly in 30 thousand new and 1,5 million known patients. With the combination of the three RL-methods, a positive predictive value of 83 % (95 %-confidence interval 65 %-94 %) was attained. Thus, the application of the presented combination of RL-methods seem to be suited for other applications of population-based research.

US hospital engagement in core domains of interoperability

OBJECTIVES

To assess US hospital engagement in the 4 core domains of interoperability (find, send, receive, integrate) and whether engaging in these domains is associated with electronic availability of clinical data from outside providers.

STUDY DESIGN

Retrospective analysis of survey data.

METHODS

Analysis of the American Hospital Association (AHA) Annual Survey of Hospitals and the American Hospital Association (AHA) Annual Survey of Hospitals - IT Supplement datasets for 2014. Respondents included 3307 US hospitals to the AHA Annual Survey - IT Supplement. We created measures of hospital engagement in 4 core domains of interoperability, as well as access to electronic clinical data from outside providers. Regression analysis was to identify hospital characteristics associated with each measure.

RESULTS

Twenty-one percent of US hospitals engaged in all 4 interoperability domains, and 25% engaged in none. Hospitals engaged in all 4 domains were more likely to have a "basic" (odds ratio [OR], 3.53; P < .01) or "comprehensive" (OR, 5.04; P < .01) electronic health record (EHR) in comparison to a less than "basic" EHR, participate in a Regional Health Information Organization (OR, 4.29; P < .01), use a single EHR vendor (OR, 2.15; P < .01), and have a third-party health information exchange vendor (OR, 2.32; P < .01). They also differed by non-IT characteristics, such as medical home participation (OR, 1.77; P < .01). Hospitals that find (OR, 5.51; P < .01), receive (OR, 2.56; P < .01), or integrate (OR, 2.53; P < .01) information were more likely to report routine clinical information availability from outside providers.

CONCLUSIONS

The one-fifth of US hospitals engaged in key domains of interoperability were more likely to have certain information technology infrastructure and participate in delivery reform. Encouragingly, interoperability engagement was associated with routine clinical information availability. Our results point to the need for ongoing efforts to expand interoperability, with the potential benefit of better information availability for clinicians and better care.

Education as a predictor of antidepressant and anxiolytic medication use after bereavement: a population-based record linkage study

Purpose

Educational attainment has been shown to be positively associated with mental health and a potential buffer to stressful events. One stressful life event likely to affect everyone in their lifetime is bereavement. This paper assesses the effect of educational attainment on mental health post-bereavement.

Methods

By utilising large administrative datasets, linking Census returns to death records and prescribed medication data, we analysed the bereavement exposure of 208,332 individuals aged 25–74 years. Two-level multi-level logistic regression models were constructed to determine the likelihood of antidepressant medication use (a proxy of mental ill health) post-bereavement given level of educational attainment.

Results

Individuals who are bereaved have greater antidepressant use than those who are not bereaved, with over a quarter (26.5 %) of those bereaved by suicide in receipt of antidepressant medication compared to just 12.4 % of those not bereaved. Within individuals bereaved by a sudden death, those with a university degree or higher qualifications are 73 % less likely to be in receipt of antidepressant medication compared to those with no qualifications, after full adjustment for demographic, socio-economic and area factors (OR 0.27, 95 % CI 0.09,0.75). Higher educational attainment and no qualifications have an equivalent effect for those bereaved by suicide.

Conclusions

Education may protect against poor mental health, as measured by the use of antidepressant medication, post-bereavement, except in those bereaved by suicide. This is likely due to the improved cognitive, personal and psychological skills gained from time spent in education.

Electronic supplementary material

The online version of this article (doi:10.1007/s11136-016-1440-1) contains supplementary material, which is available to authorized users.

Linking Data for Mothers and Babies in De-Identified Electronic Health Data

OBJECTIVE

Linkage of longitudinal administrative data for mothers and babies supports research and service evaluation in several populations around the world. We established a linked mother-baby cohort using pseudonymised, population-level data for England.

DESIGN AND SETTING

Retrospective linkage study using electronic hospital records of mothers and babies admitted to NHS hospitals in England, captured in Hospital Episode Statistics between April 2001 and March 2013.

RESULTS

Of 672,955 baby records in 2012/13, 280,470 (42%) linked deterministically to a maternal record using hospital, GP practice, maternal age, birthweight, gestation, birth order and sex. A further 380,164 (56%) records linked using probabilistic methods incorporating additional variables that could differ between mother/baby records (admission dates, ethnicity, 3/4-character postcode district) or that include missing values (delivery variables). The false-match rate was estimated at 0.15% using synthetic data. Data quality improved over time: for 2001/02, 91% of baby records were linked (holding the estimated false-match rate at 0.15%). The linked cohort was representative of national distributions of gender, gestation, birth weight and maternal age, and captured approximately 97% of births in England.

CONCLUSION

Probabilistic linkage of maternal and baby healthcare characteristics offers an efficient way to enrich maternity data, improve data quality, and create longitudinal cohorts for research and service evaluation. This approach could be extended to linkage of other datasets that have non-disclosive characteristics in common.