The Impact of the Choice of Data Source in Record Linkage Studies Estimating Mortality in Venous Thromboembolism

Long-term healthcare use after postoperative complications: an analysis of linked primary and secondary care routine data

Background

Postoperative complications are associated with reduced long-term survival. We characterise healthcare use changes after sentinel postoperative complications.

Methods

We linked primary and secondary care records of patients undergoing elective surgery at four East London hospitals (2012-7) with at least 90 days follow-up. Complication codes (wound infection, urinary tract infection, pneumonia, new stroke, and new myocardial infarction) recorded within 90 days of surgery were identified from primary or secondary care. Outcomes were change in healthcare contact days in the 2 yr before and after surgery, and 2 yr mortality. We report rate ratios (RaR) with 95% confidence intervals and adjusted for baseline healthcare use and confounders using negative binomial regression.

Results

We included 49 913 patients (median age 49 yr [inter-quartile range {IQR}: 34-64]), 27 958 (56.0%) were female. Amongst 3883 (7.8%) patients with complications (median age 58 [IQR: 43-72]), there were 18.4 days per year in contact with healthcare before surgery and 25.3 days after surgery (RaR: 1.38 [1.37-1.39]). Patients without complications (median age 48 [IQR: 33-63]) had 12.3 days per year in contact with healthcare before surgery and 14.0 days after surgery (RaR: 1.14 [1.14-1.15]). The adjusted incidence rate ratio of days in contact with healthcare associated with complications was 1.67 (1.49-1.87). More patients (391; 10.1%) with complications died within 2 yr than those without (1428; 3.1%).

Conclusions

Patients with postoperative complications are older with greater healthcare use before surgery. However, their absolute and relative increases in healthcare use after surgery are greater than patients without complications.

Real-world effects of medications for stroke prevention in atrial fibrillation: protocol for a UK population-based non-interventional cohort study with validation against randomised trial results

INTRODUCTION

Patients with atrial fibrillation experience an irregular heart rate and have an increased risk of stroke; prophylactic treatment with anticoagulation medication reduces this risk. Direct-acting oral anticoagulants (DOACs) have been approved providing an alternative to vitamin K antagonists such as warfarin. There is interest from regulatory bodies on the effectiveness of medications in routine clinical practice; however, uncertainty remains regarding the suitability of non-interventional data for answering questions on drug effectiveness and on the most suitable methods to be used. In this study, we will use data from Apixaban for Reduction in Stroke and Other Thromboembolic Events in Atrial Fibrillation (ARISTOTLE)-the pivotal trial for the DOAC apixaban-to validate non-interventional methods for assessing treatment effectiveness of anticoagulants. These methods could then be applied to analyse treatment effectiveness in people excluded from or under-represented in ARISTOTLE.

METHODS AND ANALYSIS

Patient characteristics from ARISTOTLE will be used to select a cohort of patients with similar baseline characteristics from two UK electronic health record (EHR) databases, Clinical Practice Research Datalink Gold and Aurum (between 1 January 2013 and 31 July 2019). Methods such as propensity score matching and coarsened exact matching will be explored in matching between EHR treatment groups to determine the optimal method of obtaining a balanced cohort.Absolute and relative risk of outcomes in the EHR trial-analogous cohort will be calculated and compared with the ARISTOTLE results; if results are deemed compatible the methods used for matching EHR treatment groups can then be used to examine drug effectiveness over a longer duration of exposure and in special patient groups of interest not studied in the trial.

ETHICS AND DISSEMINATION

The study has been approved by the Independent Scientific Advisory Committee of the UK Medicines and Healthcare Products Regulatory Agency. Results will be disseminated in scientific publications and at relevant conferences.

Concordance of hospitalizations between Clinical Practice Research Datalink and linked Hospital Episode Statistics among patients treated with oral antidiabetic therapies

PURPOSE

The ability of the Clinical Practice Research Datalink (CPRD) to ascertain all-cause hospitalizations remains unknown. We determined the proportion of hospitalizations in CPRD that were also recorded in Hospital Episode Statistics (HES), and vice versa, among patients initiating oral antidiabetic (OAD) therapy.

METHODS

We conducted a retrospective cohort study from October 2009 to September 2012 among OAD-treated patients registered with general practitioners who contribute to CPRD and consent to HES linkage. In CPRD, we identified initial hospitalizations for each calendar year by an Inpatient Referral, Consultation Type code, or Read code indicating an inpatient episode and determined if an admission date was recorded in HES within ±30 days. We then identified initial HES admission dates and determined if a hospitalization was documented in CPRD within ±30 days. Sensitivity analyses were conducted utilizing HES discharge, rather than admission, dates.

RESULTS

Among 8574 OAD-treated HES-linked patients in CPRD, 6574 initial hospitalizations across the study period were identified in CPRD, and 5188 (78.9% [95% CI, 77.9%-79.9%]) were confirmed by a HES admission date within ±30 days (median difference, ±3 days [IQR, 1-7 days]). Among 8609 initial hospital admissions in HES, 4803 (55.7% [95% CI, 54.7%-56.8%]) hospitalizations were recorded in CPRD within ±30 days (median difference, ±4 days [IQR, 1-9 days]). Similar results were observed using HES discharge dates.

CONCLUSION

A substantial minority of patient-level hospitalization data are nonconcordant between HES and CPRD. Pharmacoepidemiologic studies within CPRD that seek to identify hospitalizations should consider linkage with HES to ensure adequate ascertainment of inpatient events.

The accuracy of date of death recording in the Clinical Practice Research Datalink GOLD database in England compared with the Office for National Statistics death registrations

Purpose

It is not clear whether all deaths are recorded in the Clinical Practice Research Datalink (CPRD) or how accurate a recorded date of death is. Individual‐level linkage with national data from the Office for National Statistics (ONS) and Hospital Episode Statistics (HES) in England offers the opportunity to compare death information across different data sources.

Methods

Age‐standardised mortality rates (ASMRs) standardised to the European Standard Population (ESP) 2013 for CPRD were compared with figures published by the ONS, and crude mortality rates were calculated for a sample population with individual linkage between CPRD, ONS, and HES data. Agreement on the fact of death between CPRD and ONS was assessed and presented over time from 1998 to 2013.

Results

There were 33 997 patients with a record of death in the ONS data; 33 389 (98.2%) of these were also identified in CPRD. Exact agreement on the death date between CPRD and the ONS was 69.7% across the whole study period, increasing from 53.4% in 1998 to 78.0% in 2013. By 2013, 98.8% of deaths were in agreement within ±30 days.

Conclusions

For censoring follow‐up and calculating mortality rates, CPRD data are likely to be sufficient, as a delay in death recording of up to 1 month is unlikely to impact results significantly. Where the exact date of death or the cause is important, it may be advisable to include the individually linked death registration data from the ONS.

Long-term effectiveness of bisoprolol in patients with angina: A real-world evidence study

A cohort analysis using UK Clinical Practice Research Datalink (CPRD) was performed to compare the effects of bisoprolol, other β-blockers, and drugs other than β-blockers on the long-term risk of mortality and cardiovascular events in patients with angina. Adult patients first diagnosed with angina from 2000 to 2014, with ≥365 days of registration to first angina diagnosis and initiating monotherapies of bisoprolol, other β-blockers, or drugs other than β-blockers within 6 months of angina diagnosis were included. Incidence rates for each treatment cohort were compared using adjusted hazard ratio (HR) and 95% confidence intervals (CI) obtained from Cox regression analyses. Overall, 987 patients were treated with bisoprolol, 1348 with other β-blockers and 5272 with drugs other than β-blockers. Over the total follow-up (≤14 years), the HR of bisoprolol versus other β-blockers and drugs other than β-blockers for mortality was 0.45 (95% CI: 0.34-0.61) and 0.50 (95% CI: 0.38-0.66), respectively. The HR of bisoprolol versus other β-blockers for angina was 0.58 (95% CI: 0.50-0.68) and versus drugs other than β-blockers was 0.77 (95% CI: 0.68-0.88), respectively. For myocardial infarction, the HR of bisoprolol versus drugs other than β-blockers up to 14 years was 0.34 (95% CI: 0.23-0.52) and versus other β-blockers up to 5 years was 0.45 (95% CI: 0.27-0.75). At 5 years, the HR of bisoprolol versus other β-blockers, and drugs other than β-blockers, for arrhythmia was 0.60 (95% CI: 0.35-1.0) and 0.61 (95% CI: 0.40-0.93), respectively. In conclusion, long-term significant reduction in the risk of mortality and various cardiovascular events with bisoprolol versus other β-blockers, and drugs other than β-blockers, confirm treatment guidelines recommendation that bisoprolol is particularly well suited as the first-line treatment of angina in primary care.

Approach to record linkage of primary care data from Clinical Practice Research Datalink to other health-related patient data: overview and implications

Record linkage is increasingly used to expand the information available for public health research. An understanding of record linkage methods and the relevant strengths and limitations is important for robust analysis and interpretation of linked data. Here, we describe the approach used by Clinical Practice Research Datalink (CPRD) to link primary care data to other patient level datasets, and the potential implications of this approach for CPRD data analysis. General practice electronic health record software providers separately submit de-identified data to CPRD and patient identifiers to NHS Digital, excluding patients who have opted-out from contributing data. Data custodians for external datasets also send patient identifiers to NHS Digital. NHS Digital uses identifiers to link the datasets using an 8-stage deterministic methodology. CPRD subsequently receives a de-identified linked cohort file and provides researchers with anonymised linked data and metadata detailing the linkage process. This methodology has been used to generate routine primary care linked datasets, including data from Hospital Episode Statistics, Office for National Statistics and National Cancer Registration and Analysis Service. 10.6 million (M) patients from 411 English general practices were included in record linkage in June 2018. 9.1M (86%) patients were of research quality, of which 8.0M (88%) had a valid NHS number and were eligible for linkage in the CPRD standard linked dataset release. Linking CPRD data to other sources improves the range and validity of research studies. This manuscript, together with metadata generated on match strength and linkage eligibility, can be used to inform study design and explore potential linkage-related selection and misclassification biases.

Approach to record linkage of primary care data from Clinical Practice Research Datalink to other health-related patient data: overview and implications

Record linkage is increasingly used to expand the information available for public health research. An understanding of record linkage methods and the relevant strengths and limitations is important for robust analysis and interpretation of linked data. Here, we describe the approach used by Clinical Practice Research Datalink (CPRD) to link primary care data to other patient level datasets, and the potential implications of this approach for CPRD data analysis. General practice electronic health record software providers separately submit de-identified data to CPRD and patient identifiers to NHS Digital, excluding patients who have opted-out from contributing data. Data custodians for external datasets also send patient identifiers to NHS Digital. NHS Digital uses identifiers to link the datasets using an 8-stage deterministic methodology. CPRD subsequently receives a de-identified linked cohort file and provides researchers with anonymised linked data and metadata detailing the linkage process. This methodology has been used to generate routine primary care linked datasets, including data from Hospital Episode Statistics, Office for National Statistics and National Cancer Registration and Analysis Service. 10.6 million (M) patients from 411 English general practices were included in record linkage in June 2018. 9.1M (86%) patients were of research quality, of which 8.0M (88%) had a valid NHS number and were eligible for linkage in the CPRD standard linked dataset release. Linking CPRD data to other sources improves the range and validity of research studies. This manuscript, together with metadata generated on match strength and linkage eligibility, can be used to inform study design and explore potential linkage-related selection and misclassification biases.

Burden of herpes zoster in 16 selected immunocompromised populations in England: a cohort study in the Clinical Practice Research Datalink 2000-2012

OBJECTIVES

Herpes zoster (HZ) is caused by reactivation of varicella-zoster virus which remains latent in individuals after a varicella infection. It is expected that HZ will be more frequent in immunocompromised (IC) individuals than in immunocompetent (IC-free). This study assessed the incidence rate (IR) of HZ in individuals with a wide set of IC conditions and in IC-free individuals.

SETTING

A retrospective cohort study was conducted in England using data (January 2000 to March 2012) from the Clinical Practice Research Datalink with linkage to the Hospital Episodes Statistics.

PARTICIPANTS

A cohort of 621 588 individuals with 16 selected IC conditions and a gender/age-matched cohort of IC-free individuals were identified. The IC conditions included haematopoietic stem cell transplant (HSCT), solid organ transplant, malignancies, autoimmune diseases and users of immunosuppressive medications.

OUTCOMES

IR of HZ per 1000 person-years (PY) was estimated. Proportions of postherpetic neuralgia (PHN) and other HZ complications within 90 days of HZ onset were also estimated among patients with HZ. Risk factors for PHN in IC individuals with HZ were assessed by a multivariate regression model.

RESULTS

The overall IR of HZ in the IC cohort was 7.8/1000 PY (95% CI 7.7 to 7.9), increasing with age from 3.5/1000 PY (3.4-3.7) in individuals aged 18-49 years to 12.6/1000 PY (12.2-13.0) in individuals aged ≥80 years. This IR in the IC-free cohort was 6.2/1000 PY (6.1-6.3). The overall IR of HZ varied across IC conditions, ranging from 5.3 (5.1-5.5) in psoriasis to 41.7/1000 PY (35.7-48.4) in HSCT. The proportions of PHN and other HZ complications were 10.7% (10.2-11.1) and 2.9% (2.7-3.2) in the IC cohort, but 9.1% (8.7-9.5) and 2.3% (2.1-2.6) in the IC-free cohort, respectively.

CONCLUSION

IC population contributes to the public health burden of HZ in England. Vaccination might be the most preferable HZ preventive measure for the IC population.

Cancer recording in patients with and without type 2 diabetes in the Clinical Practice Research Datalink primary care data and linked hospital admission data: a cohort study

OBJECTIVES AND SETTING

Conflicting results from studies using electronic health records to evaluate the associations between type 2 diabetes and cancer fuel concerns regarding potential biases. This study aimed to describe completeness of cancer recording in UK primary care data linked to hospital admissions records.

DESIGN

Patients aged 40+ years with insulin or oral antidiabetic prescriptions in Clinical Practice Research Datalink (CPRD) primary care without type 1 diabetes were matched by age, sex and general practitioner practice to non-diabetics. Those eligible for linkage to Hospital Episode Statistics Admitted Patient Care (HES APC), and with follow-up during April 1997-December 2006 were included.

PRIMARY AND SECONDARY OUTCOME MEASURES

Cancer recording and date of first record of cancer were compared. Characteristics of patients with cancer most likely to have the diagnosis recorded only in a single data source were assessed. Relative rates of cancer estimated from the two datasets were compared.

PARTICIPANTS

53 585 patients with type 2 diabetes matched to 47 435 patients without diabetes were included.

RESULTS

Of all cancers (excluding non-melanoma skin cancer) recorded in CPRD, 83% were recorded in HES APC. 94% of cases in HES APC were recorded in CPRD. Concordance was lower when restricted to same-site cancer records, and was negatively associated with increasing age. Relative rates for cancer were similar in both datasets.

CONCLUSIONS

Good concordance in cancer recording was found between CPRD and HES APC among type 2 diabetics and matched controls. Linked data may reduce misclassification and increase case ascertainment when analysis focuses on site-specific cancers.

Performing studies using the UK Clinical Practice Research Datalink: to link or not to link?

The Clinical Practice Research Datalink (CPRD) is a repository of electronic medical records collected during routine primary care clinical practice in the UK, and is one of the most widely used sources of real-world data for healthcare research. Although CPRD provides access to comprehensive longitudinal patient records, the data does not fully capture diagnoses or outcomes occurring in secondary care and/or mortality. We provide here an overview of CPRD and the potential bias when using unlinked data in certain situations. Linkage of CPRD to other datasets can help to overcome these limitations. We discuss when to consider linkage to secondary care, disease-specific data sources or the official mortality data when conducting research using CPRD data.

Pharmacological intervention in hypertension using beta-blockers: Real-world evidence for long-term effectiveness

The study objective was to compare the long-term incidence and risk of mortality and cardiovascular outcomes in patients with hypertension initiating bisoprolol, other β-blockers or other antihypertensive therapies. Cohort analysis using UK Clinical Practice Research Datalink (CPRD). Adult patients with first diagnosis of hypertension recorded between 2000 and 2014, with ≥365 days of registration to first event and initiating monotherapies of bisoprolol, other β-blockers or drugs other than β-blockers within 6 months of diagnosis were included. Incidence rates (IR) for each treatment cohort were compared using adjusted hazard ratio (HR) and 95% confidence intervals (CI) obtained from Cox regression analyses. Of 100,066 patients included, 539 were prescribed bisoprolol, 3701 other β-blockers, and 95,826 drugs other than β-blockers. Patients receiving bisoprolol had significantly increased survival from 2 up to <15 years (HR for <15 years 0.34; 95% CI 0.18-0.67) versus other β-blockers, and from 5 to <15 years (HR for <15 years 0.52; 95% CI 0.27-1.00) versus drugs other than β-blockers. Over time, the risk of arrhythmia was higher in the bisoprolol cohort versus other β-blockers, and risks of arrhythmia and angina were higher versus drugs other than β-blockers. No differences in the risk of embolism, stroke, and myocardial infarction (MI) were found between cohorts. Over time, mortality and cardiovascular outcome IRs decreased in each cohort. In conclusion, bisoprolol showed sustained benefit on survival, evident from 2 years after treatment initiation versus other β-blockers, and from 5 years versus drugs other than β-blockers, providing long-term evidence supporting the use of bisoprolol in patients with hypertension in primary care.

Comparative safety of direct oral anticoagulants and warfarin in venous thromboembolism: multicentre, population based, observational study

Objective To determine the safety of direct oral anticoagulant (DOAC) use compared with warfarin use for the treatment of venous thromboembolism.Design Retrospective matched cohort study conducted between 1 January 2009 and 31 March 2016.Setting Community based, using healthcare data from six jurisdictions in Canada and the United States.Participants 59 525 adults (12 489 DOAC users; 47 036 warfarin users) with a new diagnosis of venous thromboembolism and a prescription for a DOAC or warfarin within 30 days of diagnosis.Main outcome measures Outcomes included hospital admission or emergency department visit for major bleeding and all cause mortality within 90 days after starting treatment. Propensity score matching and shared frailty models were used to estimate adjusted hazard ratios of the outcomes comparing DOACs with warfarin. Analyses were conducted independently at each site, with meta-analytical methods used to estimate pooled hazard ratios across sites.Results Of the 59 525 participants, 1967 (3.3%) had a major bleed and 1029 (1.7%) died over a mean follow-up of 85.2 days. The risk of major bleeding was similar for DOAC compared with warfarin use (pooled hazard ratio 0.92, 95% confidence interval 0.82 to 1.03), with the overall direction of the association favouring DOAC use. No difference was found in the risk of death (pooled hazard ratio 0.99, 0.84 to 1.16) for DOACs compared with warfarin use. There was no evidence of heterogeneity across centres, between patients with and without chronic kidney disease, across age groups, or between male and female patients.Conclusions In this analysis of adults with incident venous thromboembolism, treatment with DOACs, compared with warfarin, was not associated with an increased risk of major bleeding or all cause mortality in the first 90 days of treatment.Trial registration Clinical trials NCT02833987.

Factors influencing the development of primary care data collection projects from electronic health records: a systematic review of the literature

BACKGROUND

Primary care data gathered from Electronic Health Records are of the utmost interest considering the essential role of general practitioners (GPs) as coordinators of patient care. These data represent the synthesis of the patient history and also give a comprehensive picture of the population health status. Nevertheless, discrepancies between countries exist concerning routine data collection projects. Therefore, we wanted to identify elements that influence the development and durability of such projects.

METHODS

A systematic review was conducted using the PubMed database to identify worldwide current primary care data collection projects. The gray literature was also searched via official project websites and their contact person was emailed to obtain information on the project managers. Data were retrieved from the included studies using a standardized form, screening four aspects: projects features, technological infrastructure, GPs' roles, data collection network organization.

RESULTS

The literature search allowed identifying 36 routine data collection networks, mostly in English-speaking countries: CPRD and THIN in the United Kingdom, the Veterans Health Administration project in the United States, EMRALD and CPCSSN in Canada. These projects had in common the use of technical facilities that range from extraction tools to comprehensive computing platforms. Moreover, GPs initiated the extraction process and benefited from incentives for their participation. Finally, analysis of the literature data highlighted that governmental services, academic institutions, including departments of general practice, and software companies, are pivotal for the promotion and durability of primary care data collection projects.

CONCLUSION

Solid technical facilities and strong academic and governmental support are required for promoting and supporting long-term and wide-range primary care data collection projects.

Risk of infections in patients with gout: a population-based cohort study

To investigate the risk of various types of infections (pneumonia and urinary tract infection (UTI)), and infection-related mortality in patients with gout compared with population-based controls. A retrospective cohort study was conducted using data from the UK Clinical Practice Research Datalink (CPRD). All patients with a first diagnosis of gout and aged >40 years between January 1987-July 2014, were included and matched with up to two controls. Time-varying Cox proportional hazards models were used to estimate the risk of infections and mortality. 131,565 patients and 252,763 controls (mean age: 64 years, 74% males, mean follow-up of 6.7 years) were included in the full cohort. After full statistical adjustment, the risk of pneumonia was increased (adj. HR 1.27, 95% CI 1.18 to 1.36), while the risk of UTI (adj. HR 0.99, 95% CI 0.97 to 1.01) was similar in patients compared to controls. No differences between patients and controls were observed for infection-related mortality due to pneumonia (adj. HR 1.03, 95% CI 0.93 to 1.14) or UTI (adj. HR 1.16, 95% CI 0.98 to 1.37). In conclusion, patients with gout did not have decreased risks of pneumonia, UTI or infection-related mortality compared to population-based controls.

Assessing recording delays in general practice records to inform near real-time vaccine safety surveillance using the Clinical Practice Research Datalink (CPRD)

Purpose

Near real‐time vaccine safety surveillance (NRTVSS) is an option for post‐licensure vaccine safety assessment. NRTVSS requires timely recording of outcomes in the database used. Our main objective was to examine recording delays in the Clinical Practice Research Datalink (CPRD) for outcomes of interest for vaccine safety to inform the feasibility of NRTVSS using these data. We also evaluated completeness of recording and further assessed reporting delays for hospitalized events in CPRD.

Methods

We selected Guillain–Barré syndrome (GBS), Bell's palsy (BP), optic neuritis (ON) and febrile seizures (FS), from January 2005 to June 2014. We assessed recording delays (e.g. due to feedback from specialist referral) in stand‐alone CPRD by comparing the event and system dates and excluding delays >1 year. We used linked CPRD‐hospitalization data to further evaluate delays and completeness of recording in CPRD.

Results

Among 51 220 patients for the stand‐alone CPRD analysis (GBS: n = 830; BP: n = 12 602; ON: n = 1720; and FS: n = 36 236), most had a record entered within 1 month of the event date (GBS: 73.6%; BP: 93.4%; ON: 76.2%; and FS: 85.6%). A total of 13 482 patients, with a first record in hospital, were included for the analysis of linked data (GBS: n = 678; BP: n = 4060; ON: n = 485; and FS: n = 8321). Of these, <50% had a record in CPRD after 1 year (GBS: 41.3%; BP: 22.1%; ON: 22.4%; and FS: 41.8%).

Conclusion

This work shows that most diagnoses in CPRD for the conditions examined were recorded with delays of ≤30 days, making NRTVSS possible. The pattern of delays was condition‐specific and could be used to adjust for delays in the NRTVSS analysis. Despite low sensitivity of recording, implementing NRTVSS in CPRD is worthwhile and could be carried out, at least on a trial basis, for events of interest. © 2017 The Authors. Pharmacoepidemiology & Drug Safety Published by John Wiley & Sons Ltd.