Recording and analysing ethnicity in public health research: a bibliographical review and focus group discussions with young migrants and refugees in the UK.

BACKGROUND: The ethnicity data gap hinders public health research from addressing ethnic health inequity in the UK, especially for under-served young, migrant populations. We aimed to review how ethnicity was captured, reported, analysed, and theorised within policy-relevant research. METHODS: For this bibliographical review, we reviewed a selection of the 1% most highly cited population health papers reporting UK ethnicity data in MEDLINE and Web of Science databases between Jan 1, 1946, and July 31, 2022, and extracted how ethnicity was recorded and analysed. We included cross-sectional, longitudinal cohort studies, and randomised trials using only UK populations, which were peer-reviewed, were written in English, and reported ethnicity and any health-related outcomes. We held three focus groups with ten participants aged 18-25 years, from Nigeria, Turkistan, Syria, Yemen, and Iran to help us shape and interpret our findings, and asked "How should ethnicity be asked inclusively, and better recorded?" and "Does ethnicity change over time or context? If so, why?". We consolidated feedback from our focus groups into a co-created poster with recommendations for researchers studying ethnicity and health. Written informed consent was obtained for focus group participation. FINDINGS: Of 44 papers included in the review, 19 (43%) used self-reported ethnicity, but the number of ethnic categories provided varied. Of 27 papers that aggregated ethnicity, 13 (48%) provided justification. Only eight (18%) explicitly theorised how ethnicity related to health. The focus groups agreed that (1) ethnicity should not be prescribed by others (individuals could be asked to describe their ethnicity in free-text, which researchers could synthesise to extract relevant dimensions of ethnicity for their research) and (2) Ethnicity changes over time and context according to personal experience, social pressure, and nationality change. The lived experience of ethnicity of migrants and non-migrants is not fully interchangeable, even if they share the same ethnic category. INTERPRETATION: Researchers should communicate clearly how ethnicity is operationalised in their studies, with appropriate justification for clustering and analysis that is meaningfully theorised. Our study was limited by its non-systematic nature. Implementing the recommendation to capture ethnicity via free text remains challenging in administrative data systems. FUNDING: UCL Engagement Beacon Bursary.

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

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

Using linked administrative data to aid the handling of non-response and restore sample representativeness in cohort studies: the 1958 national child development study and hospital episode statistics data.

BACKGROUND: There is growing interest in whether linked administrative data have the potential to aid analyses subject to missing data in cohort studies. METHODS: Using linked 1958 National Child Development Study (NCDS; British cohort born in 1958, n = 18,558) and Hospital Episode Statistics (HES) data, we applied a LASSO variable selection approach to identify HES variables which are predictive of non-response at the age 55 sweep of NCDS. We then included these variables as auxiliary variables in multiple imputation (MI) analyses to explore the extent to which they helped restore sample representativeness of the respondents together with the imputed non-respondents in terms of early life variables (father's social class at birth, cognitive ability at age 7) and relative to external population benchmarks (educational qualifications and marital status at age 55). RESULTS: We identified 10 HES variables that were predictive of non-response at age 55 in NCDS. For example, cohort members who had been treated for adult mental illness had more than 70% greater odds of bring non-respondents (odds ratio 1.73; 95% confidence interval 1.17, 2.51). Inclusion of these HES variables in MI analyses only helped to restore sample representativeness to a limited extent. Furthermore, there was essentially no additional gain in sample representativeness relative to analyses using only previously identified survey predictors of non-response (i.e. NCDS rather than HES variables). CONCLUSIONS: Inclusion of HES variables only aided missing data handling in NCDS to a limited extent. However, these findings may not generalise to other analyses, cohorts or linked administrative datasets. This work provides a demonstration of the use of linked administrative data for the handling of missing cohort data which we hope will act as template for others.

Trends in survival of children with severe congenital heart defects by gestational age at birth: A population-based study using administrative hospital data for England.

BACKGROUND: Children with congenital heart defects (CHD) are twice as likely as their peers to be born preterm (<37 weeks' gestation), yet descriptions of recent trends in long-term survival by gestational age at birth (GA) are lacking. OBJECTIVES: To quantify changes in survival to age 5 years of children in England with severe CHD by GA. METHODS: We estimated changes in survival to age five of children with severe CHD and all other children born in England between April 2004 and March 2016, overall and by GA-group using linked hospital and mortality records. RESULTS: Of 5,953,598 livebirths, 5.7% (339,080 of 5,953,598) were born preterm, 0.35% (20,648 of 5,953,598) died before age five and 3.6 per 1000 (21,291 of 5,953,598) had severe CHD. Adjusting for GA, under-five mortality rates fell at a similar rate between 2004-2008 and 2012-2016 for children with severe CHD (adjusted hazard ratio [HR] 0.79, 95% CI 0.71, 0.88) and all other children (HR 0.78, 95% CI 0.76, 0.81). For children with severe CHD, overall survival to age five increased from 87.5% (95% CI 86.6, 88.4) in 2004-2008 to 89.6% (95% CI 88.9, 90.3) in 2012-2016. There was strong evidence for better survival in the ≥39-week group (90.2%, 95% CI 89.1, 91.2 to 93%, 95% CI 92.4, 93.9), weaker evidence at 24-31 and 37-38 weeks and no evidence at 32-36 weeks. We estimate that 51 deaths (95% CI 24, 77) per year in children with severe CHD were averted in 2012-2016 compared to what would have been the case had 2004-2008 mortality rates persisted. CONCLUSIONS: Nine out of 10 children with severe CHD in 2012-2016 survived to age five. The small improvement in survival over the study period was driven by increased survival in term children. Most children with severe CHD are reaching school age and may require additional support by schools and healthcare services.

How is ethnicity reported, described, and analysed in health research in the UK? A bibliographical review and focus group discussions with young refugees.

BACKGROUND: The ethnicity data gap pertains to 3 major challenges to address ethnic health inequality: 1) Under-representation of ethnic minorities in research; 2) Poor data quality on ethnicity; 3) Ethnicity data not being meaningfully analysed. These challenges are especially relevant for research involving under-served migrant populations in the UK. We aimed to review how ethnicity is captured, reported, analysed and theorised within policy-relevant research on ethnic health inequities. METHODS: We reviewed a selection of the 1% most highly cited population health papers that reported UK data on ethnicity, and extracted how ethnicity was recorded and analysed in relation to health outcomes. We focused on how ethnicity was obtained (i.e. self reported or not), how ethnic groups were categorised, whether justification was provided for any categorisation, and how ethnicity was theorised to be related to health. We held three 1-h-long guided focus groups with 10 young people from Nigeria, Turkistan, Syria, Yemen and Iran. This engagement helped us shape and interpret our findings, and reflect on. 1) How should ethnicity be asked inclusively, and better recorded? 2) Does self-defined ethnicity change over time or context? If so, why? RESULTS: Of the 44 included papers, most (19; 43%) used self-reported ethnicity, categorised in a variety of ways. Of the 27 papers that aggregated ethnicity, 13 (48%) provided justification. Only 8 of 33 papers explicitly theorised how ethnicity related to health. The focus groups agreed that 1) Ethnicity should not be prescribed by others; individuals could be asked to describe their ethnicity in free-text which researchers could synthesise to extract relevant dimensions of ethnicity for their research; 2) Ethnicity changes over time and context according to personal experience, social pressure, and nationality change; 3) Migrants and non-migrants' lived experience of ethnicity is not fully inter-changeable, even if they share the same ethnic category. CONCLUSIONS: Ethnicity is a multi-dimensional construct, but this is not currently reflected in UK health research studies, where ethnicity is often aggregated and analysed without justification. Researchers should communicate clearly how ethnicity is operationalised for their study, with appropriate justification for clustering and analysis that is meaningfully theorised. We can only start to tackle ethnic health inequity by treating ethnicity as rigorously as any other variables in our research.

Biases arising from linked administrative data for epidemiological research: a conceptual framework from registration to analyses.

Linked administrative data offer a rich source of information that can be harnessed to describe patterns of disease, understand their causes and evaluate interventions. However, administrative data are primarily collected for operational reasons such as recording vital events for legal purposes, and planning, provision and monitoring of services. The processes involved in generating and linking administrative datasets may generate sources of bias that are often not adequately considered by researchers. We provide a framework describing these biases, drawing on our experiences of using the 100 Million Brazilian Cohort (100MCohort) which contains records of more than 131 million people whose families applied for social assistance between 2001 and 2018. Datasets for epidemiological research were derived by linking the 100MCohort to health-related databases such as the Mortality Information System and the Hospital Information System. Using the framework, we demonstrate how selection and misclassification biases may be introduced in three different stages: registering and recording of people's life events and use of services, linkage across administrative databases, and cleaning and coding of variables from derived datasets. Finally, we suggest eight recommendations which may reduce biases when analysing data from administrative sources.

Preterm birth, unplanned hospital contact, and mortality in infants born to teenage mothers in five countries: An administrative data cohort study.

BACKGROUND: Young maternal age is associated with lower birthweight and higher rates of preterm birth and childhood hospitalisations. Internationally, teen pregnancy rates vary widely, reflecting differences in social, welfare, and health care factors in different cultural contexts. OBJECTIVES: To determine whether the increased risk of adverse infant outcomes among teenage mothers varies by country, reflecting different national teenage birth rates and country-specific social/welfare policies, in Scotland (higher teenage pregnancy rates), England, New South Wales (NSW; Australia), Ontario (Canada), and Sweden (lower rates). METHODS: We used administrative hospital data capturing 3 002 749 singleton births surviving to postnatal discharge between 2010 and 2014 (2008-2012 for Sweden). We compared preterm birth (24-36 weeks' gestation), mortality within 12 months of postnatal discharge, unplanned hospital admissions, and emergency department visits within 12 months of postnatal discharge, for infants born to mothers aged 15-19, 20-24, 25-29, and 30-34 years. RESULTS: Compared to births to women aged 30-34 years, risks of adverse outcomes among teenage mothers were higher in all countries, but the magnitude of effects was not related to country-specific rates of teenage births. Teenage mothers had between 1.2% (95% confidence interval [CI] 0.7, 1.7, Sweden) and 2.0% (95% CI 1.4, 2.5, NSW) more preterm births, and between 9.8 (95% CI 7.2, 12.4, England) and 19.7 (95% CI 8.7, 30.6, Scotland) more deaths per 10 000 infants, compared with mothers aged 30-34. Between 6.4% (95% CI 5.5, 7.4, NSW) and 25.4% (95% CI 24.7, 26.1, Ontario), more infants born to teenage mothers had unplanned hospital contacts compared with those born to mothers aged 30-34. CONCLUSIONS: Regardless of country, infants born to teenage mothers had universally worse outcomes than those born to older mothers. This excess risk did not vary by national rates of livebirths to teenage mothers. Current mechanisms to support teenage mothers have not eliminated maternal age-related disparities in infant outcomes; further strategies to mitigate excess risk in all countries are needed.

Assessing data linkage quality in cohort studies.

Background: Linkage of administrative data sources provides an efficient means of collecting detailed data on how individuals interact with cross-sectoral services, society, and the environment. These data can be used to supplement conventional cohort studies, or to create population-level electronic cohorts generated solely from administrative data. However, errors occurring during linkage (false matches/missed matches) can lead to bias in results from linked data.Aim: This paper provides guidance on evaluating linkage quality in cohort studies.Methods: We provide an overview of methods for linkage, describe mechanisms by which linkage error can introduce bias, and draw on real-world examples to demonstrate methods for evaluating linkage quality.Results: Methods for evaluating linkage quality described in this paper provide guidance on (i) estimating linkage error rates, (ii) understanding the mechanisms by which linkage error might bias results, and (iii) information that should be shared between data providers, linkers and users, so that approaches to handling linkage error in analysis can be implemented.Conclusion: Linked administrative data can enhance conventional cohorts and offers the ability to answer questions that require large sample sizes or hard-to-reach populations. Care needs to be taken to evaluate linkage quality in order to provide robust results.

Long-term mortality in mothers of infants with neonatal abstinence syndrome: A population-based parallel-cohort study in England and Ontario, Canada.

BACKGROUND: Opioid addiction is a major public health threat to healthy life expectancy; however, little is known of long-term mortality for mothers with opioid use in pregnancy. Pregnancy and delivery care are opportunities to improve access to addiction and supportive services. Treating neonatal abstinence syndrome (NAS) as a marker of opioid use during pregnancy, this study reports long-term maternal mortality among mothers with a birth affected by NAS in relation to that of mothers without a NAS-affected birth in 2 high-prevalence jurisdictions, England and Ontario, Canada. METHODS AND FINDINGS: We conducted a population-based study using linked administrative health data to develop parallel cohorts of mother-infant dyads in England and Ontario between 2002 and 2012. The study population comprised 13,577 and 4,966 mothers of infants with NAS and 4,205,675 and 929,985 control mothers in England and Ontario, respectively. Death records captured all-cause maternal mortality after delivery through March 31, 2016, and cause-specific maternal mortality to December 31, 2014. The primary exposure was a live birth of an infant with NAS, and the main outcome was all deaths among mothers following their date of delivery. We modelled the association between NAS and all-cause maternal mortality using Cox regression, and the cumulative incidence of cause-specific mortality within a competing risks framework. All-cause mortality rates, 10-year cumulative incidence risk of death, and crude and age-adjusted hazard ratios were calculated. Estimated crude 10-year mortality based on Kaplan-Meier curves in mothers of infants with NAS was 5.1% (95% CI 4.7%-5.6%) in England and 4.6% (95% CI 3.8%-5.5%) in Ontario versus 0.4% (95% CI 0.41%-0.42%) in England and 0.4% (95% CI 0.38%-0.41%) in Ontario for controls (p < 0.001 for all comparisons). Survival curves showed no clear inflection point or period of heightened risk. The crude hazard ratio for all-cause mortality was 12.1 (95% 11.1-13.2; p < 0.001) in England and 11.4 (9.7-13.4; p < 0.001) in Ontario; age adjustment did not reduce the hazard ratios. The cumulative incidence of death was higher among NAS mothers than controls for almost all causes of death. The majority of deaths were by avoidable causes, defined as those that are preventable, amenable to care, or both. Limitations included lack of direct measures of maternal opioid use, other substance misuse, and treatments or supports received. CONCLUSIONS: In this study, we found that approximately 1 in 20 mothers of infants with NAS died within 10 years of delivery in both England and Canada-a mortality risk 11-12 times higher than for control mothers. Risk of death was not limited to the early postpartum period targeted by most public health programs. Policy responses to the current opioid epidemic require effective strategies for long-term support to improve the health and welfare of opioid-using mothers and their children.

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.

GUILD: GUidance for Information about Linking Data sets.

Record linkage of administrative and survey data is increasingly used to generate evidence to inform policy and services. Although a powerful and efficient way of generating new information from existing data sets, errors related to data processing before, during and after linkage can bias results. However, researchers and users of linked data rarely have access to information that can be used to assess these biases or take them into account in analyses. As linked administrative data are increasingly used to provide evidence to guide policy and services, linkage error, which disproportionately affects disadvantaged groups, can undermine evidence for public health. We convened a group of researchers and experts from government data providers to develop guidance about the information that needs to be made available about the data linkage process, by data providers, data linkers, analysts and the researchers who write reports. The guidance goes beyond recommendations for information to be included in research reports. Our aim is to raise awareness of information that may be required at each step of the linkage pathway to improve the transparency, reproducibility, and accuracy of linkage processes, and the validity of analyses and interpretation of results.

Perinatal mortality associated with induction of labour versus expectant management in nulliparous women aged 35 years or over: An English national cohort study.

BACKGROUND: A recent randomised controlled trial (RCT) demonstrated that induction of labour at 39 weeks of gestational age has no short-term adverse effect on the mother or infant among nulliparous women aged ≥35 years. However, the trial was underpowered to address the effect of routine induction of labour on the risk of perinatal death. We aimed to determine the association between induction of labour at ≥39 weeks and the risk of perinatal mortality among nulliparous women aged ≥35 years. METHODS AND FINDINGS: We used English Hospital Episode Statistics (HES) data collected between April 2009 and March 2014 to compare perinatal mortality between induction of labour at 39, 40, and 41 weeks of gestation and expectant management (continuation of pregnancy to either spontaneous labour, induction of labour, or caesarean section at a later gestation). Analysis was by multivariable Poisson regression with adjustment for maternal characteristics and pregnancy-related conditions. Among the cohort of 77,327 nulliparous women aged 35 to 50 years delivering a singleton infant, 33.1% had labour induced: these women tended to be older and more likely to have medical complications of pregnancy, and the infants were more likely to be small for gestational age. Induction of labour at 40 weeks (compared with expectant management) was associated with a lower risk of in-hospital perinatal death (0.08% versus 0.26%; adjusted risk ratio [adjRR] 0.33; 95% CI 0.13-0.80, P = 0.015) and meconium aspiration syndrome (0.44% versus 0.86%; adjRR 0.52; 95% CI 0.35-0.78, P = 0.002). Induction at 40 weeks was also associated with a slightly increased risk of instrumental vaginal delivery (adjRR 1.06; 95% CI 1.01-1.11, P = 0.020) and emergency caesarean section (adjRR 1.05; 95% CI 1.01-1.09, P = 0.019). The number needed to treat (NNT) analysis indicated that 562 (95% CI 366-1,210) inductions of labour at 40 weeks would be required to prevent 1 perinatal death. Limitations of the study include the reliance on observational data in which gestational age is recorded in weeks rather than days. There is also the potential for unmeasured confounders and under-recording of induction of labour or perinatal death in the dataset. CONCLUSIONS: Bringing forward the routine offer of induction of labour from the current recommendation of 41-42 weeks to 40 weeks of gestation in nulliparous women aged ≥35 years may reduce overall rates of perinatal death.

Impregnated central venous catheters for prevention of bloodstream infection in children (the CATCH trial): a randomised controlled trial.

BACKGROUND: Impregnated central venous catheters are recommended for adults to reduce bloodstream infections but not for children because there is not enough evidence to prove they are effective. We aimed to assess the effectiveness of any type of impregnation (antibiotic or heparin) compared with standard central venous catheters to prevent bloodstream infections in children needing intensive care. METHODS: We did a randomised controlled trial of children admitted to 14 English paediatric intensive care units. Children younger than 16 years were eligible if they were admitted or being prepared for admission to a participating paediatric intensive care unit and were expected to need a central venous catheter for 3 or more days. Children were randomly assigned (1:1:1) to receive a central venous catheter impregnated with antibiotics, a central venous catheter impregnated with heparin, or a standard central venous catheter with computer generated randomisation in blocks of three and six, stratified by method of consent, site, and envelope storage location within the site. The clinician responsible for inserting the central venous catheter was not masked to allocation, but allocation was concealed from patients, their parents, and the paediatric intensive care unit personnel responsible for their care. The primary outcome was time to first bloodstream infection between 48 h after randomisation and 48 h after central venous catheter removal with impregnated (antibiotic or heparin) versus standard central venous catheters, assessed in the intention-to-treat population. Safety analyses compared central venous catheter-related adverse events in the subset of children for whom central venous catheter insertion was attempted (per-protocol population). This trial is registered with ISRCTN number, ISRCTN34884569. FINDINGS: Between Nov 25, 2010, and Nov 30, 2012, 1485 children were recruited to this study. We randomly assigned 502 children to receive standard central venous catheters, 486 to receive antibiotic-impregnated catheters, and 497 to receive heparin-impregnated catheters. Bloodstream infection occurred in 18 (4%) of those in the standard catheters group, 7 (1%) in the antibiotic-impregnated group, and 17 (3%) assigned to heparin-impregnated catheters. Primary analyses showed no effect of impregnated (antibiotic or heparin) catheters compared with standard central venous catheters (hazard ratio [HR] for time to first bloodstream infection 0.71, 95% CI 0.37-1.34). Secondary analyses showed that antibiotic central venous catheters were better than standard central venous catheters (HR 0.43, 0.20-0.96) and heparin central venous catheters (HR 0.42, 0.19-0.93), but heparin did not differ from standard central venous catheters (HR 1.04, 0.53-2.03). Clinically important and statistically significant absolute risk differences were identified only for antibiotic-impregnated catheters versus standard catheters (-2.15%, 95% CI -4.09 to -0.20; number needed to treat [NNT] 47, 95% CI 25-500) and antibiotic-impregnated catheters versus heparin-impregnated catheters (-1.98%, -3.90 to -0.06, NNT 51, 26-1667). Nine children (2%) in the standard central venous catheter group, 14 (3%) in the antibiotic-impregnated group, and 8 (2%) in the heparin-impregnated group had catheter-related adverse events. 45 (8%) in the standard group, 35 (8%) antibiotic-impregnated group, and 29 (6%) in the heparin-impregnated group died during the study. INTERPRETATION: Antibiotic-impregnated central venous catheters significantly reduced the risk of bloodstream infections compared with standard and heparin central venous catheters. Widespread use of antibiotic-impregnated central venous catheters could help prevent bloodstream infections in paediatric intensive care units. FUNDING: National Institute for Health Research, UK.

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.

Utilising identifier error variation in linkage of large administrative data sources.

BACKGROUND: Linkage of administrative data sources often relies on probabilistic methods using a set of common identifiers (e.g. sex, date of birth, postcode). Variation in data quality on an individual or organisational level (e.g. by hospital) can result in clustering of identifier errors, violating the assumption of independence between identifiers required for traditional probabilistic match weight estimation. This potentially introduces selection bias to the resulting linked dataset. We aimed to measure variation in identifier error rates in a large English administrative data source (Hospital Episode Statistics; HES) and to incorporate this information into match weight calculation. METHODS: We used 30,000 randomly selected HES hospital admissions records of patients aged 0-1, 5-6 and 18-19 years, for 2011/2012, linked via NHS number with data from the Personal Demographic Service (PDS; our gold-standard). We calculated identifier error rates for sex, date of birth and postcode and used multi-level logistic regression to investigate associations with individual-level attributes (age, ethnicity, and gender) and organisational variation. We then derived: i) weights incorporating dependence between identifiers; ii) attribute-specific weights (varying by age, ethnicity and gender); and iii) organisation-specific weights (by hospital). Results were compared with traditional match weights using a simulation study. RESULTS: Identifier errors (where values disagreed in linked HES-PDS records) or missing values were found in 0.11% of records for sex and date of birth and in 53% of records for postcode. Identifier error rates differed significantly by age, ethnicity and sex (p < 0.0005). Errors were less frequent in males, in 5-6 year olds and 18-19 year olds compared with infants, and were lowest for the Asian ethic group. A simulation study demonstrated that substantial bias was introduced into estimated readmission rates in the presence of identifier errors. Attribute- and organisational-specific weights reduced this bias compared with weights estimated using traditional probabilistic matching algorithms. CONCLUSIONS: We provide empirical evidence on variation in rates of identifier error in a widely-used administrative data source and propose a new method for deriving match weights that incorporates additional data attributes. Our results demonstrate that incorporating information on variation by individual-level characteristics can help to reduce bias due to linkage error.

Newborn Length of Stay and Risk of Readmission.

BACKGROUND: Evidence on the association between newborn length of hospital stay (LOS) and risk of readmission is conflicting. We compared methods for modelling this relationship, by gestational age, using population-level hospital data on births in England between 2005-14. METHODS: The association between LOS and unplanned readmission within 30 days of postnatal discharge was explored using four approaches: (i) modelling hospital-level LOS and readmission rates; (ii) comparing trends over time in LOS and readmission; (iii) modelling individual LOS and adjusted risk of readmission; and (iv) instrumental variable analyses (hospital-level mean LOS and number of births on the same day). RESULTS: Of 4 667 827 babies, 5.2% were readmitted within 30 days. Aggregated data showed hospitals with longer mean LOS were not associated with lower readmission rates for vaginal (adjusted risk ratio (aRR) 0.87, 95% confidence interval (CI) 0.66, 1.13), or caesarean (aRR 0.89, 95% CI 0.72, 1.12) births. LOS fell by an average 2.0% per year for vaginal births and 3.4% for caesarean births, while readmission rates increased by 4.4 and 5.1% per year respectively. Approaches (iii) and (iv) indicated that longer LOS was associated with a reduced risk of readmission, but only for late preterm, vaginal births (34-36 completed weeks' gestation). CONCLUSIONS: Longer newborn LOS may benefit late preterm babies, possibly due to increased medical or psychosocial support for those at greater risk of potentially preventable readmissions after birth. Research based on observational data to evaluate relationships between LOS and readmission should use methods to reduce the impact of unmeasured confounding.

Evaluation of record linkage of two large administrative databases in a middle income country: stillbirths and notifications of dengue during pregnancy in Brazil.

BACKGROUND: Due to the increasing availability of individual-level information across different electronic datasets, record linkage has become an efficient and important research tool. High quality linkage is essential for producing robust results. The objective of this study was to describe the process of preparing and linking national Brazilian datasets, and to compare the accuracy of different linkage methods for assessing the risk of stillbirth due to dengue in pregnancy. METHODS: We linked mothers and stillbirths in two routinely collected datasets from Brazil for 2009-2010: for dengue in pregnancy, notifications of infectious diseases (SINAN); for stillbirths, mortality (SIM). Since there was no unique identifier, we used probabilistic linkage based on maternal name, age and municipality. We compared two probabilistic approaches, each with two thresholds: 1) a bespoke linkage algorithm; 2) a standard linkage software widely used in Brazil (ReclinkIII), and used manual review to identify further links. Sensitivity and positive predictive value (PPV) were estimated using a subset of gold-standard data created through manual review. We examined the characteristics of false-matches and missed-matches to identify any sources of bias. RESULTS: From records of 678,999 dengue cases and 62,373 stillbirths, the gold-standard linkage identified 191 cases. The bespoke linkage algorithm with a conservative threshold produced 131 links, with sensitivity = 64.4% (68 missed-matches) and PPV = 92.5% (8 false-matches). Manual review of uncertain links identified an additional 37 links, increasing sensitivity to 83.7%. The bespoke algorithm with a relaxed threshold identified 132 true matches (sensitivity = 69.1%), but introduced 61 false-matches (PPV = 68.4%). ReclinkIII produced lower sensitivity and PPV than the bespoke linkage algorithm. Linkage error was not associated with any recorded study variables. CONCLUSION: Despite a lack of unique identifiers for linking mothers and stillbirths, we demonstrate a high standard of linkage of large routine databases from a middle income country. Probabilistic linkage and manual review were essential for accurately identifying cases for a case-control study, but this approach may not be feasible for larger databases or for linkage of more common outcomes.

Technical feasibility and validation of a coronary artery calcium scoring system using CT coronary angiography images.

OBJECTIVES: We validate a novel CT coronary angiography (CCTA) coronary calcium scoring system. METHODS: Calcium was quantified on CCTA images using a new patient-specific attenuation threshold: mean + 2SD of intra-coronary contrast density (HU). Using 335 patient data sets a conversion factor (CF) for predicting CACS from CCTA scores (CCTAS) was derived and validated in a separate cohort (n = 168). Bland-Altman analysis and weighted kappa for MESA centiles and Agatston risk groupings were calculated. RESULTS: Multivariable linear regression yielded a CF: CACS = (1.185 × CCTAS) + (0.002 × CCTAS × attenuation threshold). When applied to CCTA data sets there was excellent correlation (r = 0.95; p < 0.0001) and agreement (mean difference -10.4 [95% limits of agreement -258.9 to 238.1]) with traditional calcium scores. Agreement was better for calcium scores below 500; however, MESA percentile agreement was better for high risk patients. Risk stratification was excellent (Agatston groups k = 0.88 and MESA centiles k = 0.91). Eliminating the dedicated CACS scan decreased patient radiation exposure by approximately one-third. CONCLUSION: CCTA calcium scores can accurately predict CACS using a simple, individualized, semiautomated approach reducing acquisition time and radiation exposure when evaluating patients for CAD. This method is not affected by the ROI location, imaging protocol, or tube voltage strengthening its clinical applicability. KEY POINTS: • Coronary calcium scores can be reliably determined on contrast-enhanced cardiac CT • This score can accurately risk stratify patients • Elimination of a dedicated calcium scan reduces patient radiation by a third.

The REporting of studies Conducted using Observational Routinely-collected health Data (RECORD) statement.

Routinely collected health data, obtained for administrative and clinical purposes without specific a priori research goals, are increasingly used for research. The rapid evolution and availability of these data have revealed issues not addressed by existing reporting guidelines, such as Strengthening the Reporting of Observational Studies in Epidemiology (STROBE). The REporting of studies Conducted using Observational Routinely collected health Data (RECORD) statement was created to fill these gaps. RECORD was created as an extension to the STROBE statement to address reporting items specific to observational studies using routinely collected health data. RECORD consists of a checklist of 13 items related to the title, abstract, introduction, methods, results, and discussion section of articles, and other information required for inclusion in such research reports. This document contains the checklist and explanatory and elaboration information to enhance the use of the checklist. Examples of good reporting for each RECORD checklist item are also included herein. This document, as well as the accompanying website and message board (http://www.record-statement.org), will enhance the implementation and understanding of RECORD. Through implementation of RECORD, authors, journals editors, and peer reviewers can encourage transparency of research reporting.