Validity of cancer diagnosis in a primary care database compared with linked cancer registrations in England. Population-based cohort study

doi:10.1016/j.canep.2012.05.013

Cancer Epidemiology

Volume 36, Issue 5, October 2012, Pages 425-429

https://doi.org/10.1016/j.canep.2012.05.013 Get rights and content

Abstract

Aims

The present study aimed to evaluate the validity of cancer diagnoses and death recording in a primary care database compared with cancer registry (CR) data in England.

Methods

The eligible cohort comprised 42,556 participants, registered with English general practices in the General Practice Research Database (GPRD) that consented to CR linkage. CR and primary care records were compared for cancer diagnosis, date of cancer diagnosis and death. Read and ICD cancer code sets were reviewed and agreed by two authors.

Results

There were 5216 (91% of CR total) cancer events diagnosed in both sources. There were 494 (9%) diagnosed in CR only and 213 (4%) that were diagnosed in GPRD only. The predictive value of a GPRD cancer diagnosis was 96% for lung cancer, 92% for urinary tract cancer, 96% for gastro-oesophageal cancer and 98% for colorectal cancer. ‘False negative’ primary care records were sometimes accounted for by registration end dates being shortly before cancer diagnosis dates. The date of cancer diagnosis was median 11 (interquartile range −6 to 30) days later in GPRD compared with CR. Death records were consistent for the two sources for 3337/3397 (99%) of cases.

Conclusion

Recording of cancer diagnosis and mortality in primary care electronic records is generally consistent with CR in England. Linkage studies must pay careful attention to selection of codes to define eligibility and timing of diagnoses in relation to beginning and end of record.

Introduction

Electronic health records are an increasingly utilised resource for epidemiological research. In the UK, records from primary care databases have been used in studies of cancer diagnosis and prognosis [1]. Validation studies have confirmed the accuracy and completeness of UK electronic patient records with respect to several clinical conditions as well as pharmacological treatment, and death [2], [3], [4], [5], [6], [7], [8], [9]. In the UK, a national system for cancer registration aims to record all new cancer diagnoses. Cancer Registry (CR) data are considered to represent an accurate resource for studies of cancer incidence and prognosis [10]. The validity of cancer diagnoses in primary care electronic records in comparison with cancer registrations has not been well described.

The present study builds on an earlier analysis of data from the General Practice Research Database (GPRD) [1] that evaluated the incidence of cancer in patients presenting with four ‘alarm’ symptoms, haematuria, haemoptysis, dyphagia and rectal bleeding. At the time of the initial analysis linked cancer data were not available to ascertain the frequency and validity of cancer diagnoses recorded in the GPRD. In order to confirm our initial findings, we wanted to ascertain the validity of cancer diagnoses and dates of cancer diagnosis in GPRD. We have made use of the opportunity presented by a novel linkage between cancer registrations with primary care electronic records to compare data from the two sources. The present report therefore aims to evaluate the validity of cancer diagnoses in primary care electronic health records by comparing the occurrence, and timing of cancer diagnoses between GPRD with cancer registrations. We specifically evaluated diagnoses of lung cancer, colorectal cancer, cancer of the oesophagus or stomach and urinary tract cancers.

Section snippets

Cancer registry

Cancer registries in England represent the only available source of reliable population-based data on cancer incidence, prevalence and survival, excluding non-melanoma skin cancer which is not collected systematically. Information is collected on new diagnoses of cancer from hospitals, pathology laboratories, hospices, cancer treatment centres, cancer screening programmes, Hospital Episode Statistics (HES), cancer waiting times (CWT) and death certificates. Within hospitals, data are collected

Results

The initial GPRD sample comprised 83,841 participants. Fig. 1 provides a flowchart of the sample selection process. There were 173 (53%) out of the original 334 GPRD practices without any cancer occurrence recorded in the CR that were excluded from further analyses as not consenting to linkage. This left 158 (47%) eligible practices that participated in linkage between GPRD and CR. Consequently, 49% (N = 37,283) of participants with alarm symptoms (N = 76,143) and 46% (N = 5254) participants with

Discussion

The present study aimed to estimate the validity of four cancer diagnoses in GPRD in comparison with CR data as a reference source of cancer diagnoses in England. Overall, the present findings endorse the validity of colorectal, gastro-oesophageal, respiratory and urinary tract cancer diagnosis in the GPRD, including the timing of these diagnoses and subsequent mortality.

High sensitivity is extremely important when the aim is to identify severe but treatable disease such as the four cancer

Conclusion

The present study documented that cancer diagnoses in the GPRD are valid and accurate more than 90% of the time and can be used with reasonable confidence for studies of clinical care, auditing, and prevention on the four cancer diagnoses included in the present study.

Conflict of interest statement

None to be declared.

Funding

None.

References (15)

R. Jones et al.
Alarm symptoms in early diagnosis of cancer in primary care: cohort study using General Practice Research Database
BMJ
(2007)
H. Jick et al.
Validation of information recorded on general practitioner based computerized data resource in the United Kingdom
BMJ
(1991)
H. Jick et al.
Further validation of information recorded on a general practitioner based computerized data resource in the United Kingdom
Pharmacoepidemiol Drug Saf
(1992)
S.S. Jick et al.
Validity of the general practice research database
Pharmacotherapy
(2003)
N.F. Khan et al.
Validity of diagnostic coding within the General Practice Research Database: a systematic review
Br J Gen Pract
(2010)
E. Herrett et al.
Validation and validity of diagnoses in the General Practice Research Database: a systematic review
Br J Clin Pharmacol
(2010)
M.D. Manser et al.
Colorectal cancer registration: the central importance of pathology
J Clin Pathol
(2000)

There are more references available in the full text version of this article.

Cited by (109)

Validation and impact of algorithms for identifying variables in observational studies of routinely collected data
2024, Journal of Clinical Epidemiology
Among observational studies of routinely collected health data (RCD) for exploring treatment effects, algorithms are used to identify study variables. However, the extent to which algorithms are reliable and impact the credibility of effect estimates is far from clear. This study aimed to investigate the validation of algorithms for identifying study variables from RCD, and examine the impact of alternative algorithms on treatment effects.
We searched PubMed for observational studies published in 2018 that used RCD to explore drug treatment effects. Information regarding the reporting, validation, and interpretation of algorithms was extracted. We summarized the reporting and methodological characteristics of algorithms and validation. We also assessed the divergence in effect estimates given alternative algorithms by calculating the ratio of estimates of the primary vs. alternative analyses.
A total of 222 studies were included, of which 93 (41.9%) provided a complete list of algorithms for identifying participants, 36 (16.2%) for exposure, and 132 (59.5%) for outcomes, and 15 (6.8%) for all study variables including population, exposure, and outcomes. Fifty-nine (26.6%) studies stated that the algorithms were validated, and 54 (24.3%) studies reported methodological characteristics of 66 validations, among which 61 validations in 49 studies were from the cross-referenced validation studies. Of those 66 validations, 22 (33.3%) reported sensitivity and 16 (24.2%) reported specificity. A total of 63.6% of studies reporting sensitivity and 56.3% reporting specificity used test-result-based sampling, an approach that potentially biases effect estimates. Twenty-eight (12.6%) studies used alternative algorithms to identify study variables, and 24 reported the effects estimated by primary analyses and sensitivity analyses. Of these, 20% had differential effect estimates when using alternative algorithms for identifying population, 18.2% for identifying exposure, and 45.5% for classifying outcomes. Only 32 (14.4%) studies discussed how the algorithms may affect treatment estimates.
In observational studies of RCD, the algorithms for variable identification were not regularly validated, and–even if validated–the methodological approach and performance of the validation were often poor. More seriously, different algorithms may yield differential treatment effects, but their impact is often ignored by researchers. Strong efforts, including recommendations, are warranted to improve good practice.
Examining methodology to identify patterns of consulting in primary care for different groups of patients before a diagnosis of cancer: An exemplar applied to oesophagogastric cancer
2023, Cancer Epidemiology
Current methods for estimating the timeliness of cancer diagnosis are not robust because dates of key defining milestones, for example first presentation, are uncertain. This is exacerbated when patients have other conditions (multimorbidity), particularly those that share symptoms with cancer. Methods independent of this uncertainty are needed for accurate estimates of the timeliness of cancer diagnosis, and to understand how multimorbidity impacts the diagnostic process.
Participants were diagnosed with oesophagogastric cancer between 2010 and 2019. Controls were matched on year of birth, sex, general practice and multimorbidity burden calculated using the Cambridge Multimorbidity Score. Primary care data (Clinical Practice Research Datalink) was used to explore population-level consultation rates for up to two years before diagnosis across different multimorbidity burdens. Five approaches were compared on the timing of the consultation frequency increase, the inflection point for different multimorbidity burdens, different aggregated time-periods and sample sizes.
We included 15,410 participants, of which 13,328 (86.5 %) had a measurable multimorbidity burden. Our new maximum likelihood estimation method found evidence that the inflection point in consultation frequency varied with multimorbidity burden, from 154 days (95 %CI 131.8–176.2) before diagnosis for patients with no multimorbidity, to 126 days (108.5–143.5) for patients with the greatest multimorbidity burden. Inflection points identified using alternative methods were closer to diagnosis for up to three burden groups. Sample size reduction and changing the aggregation period resulted in inflection points closer to diagnosis, with the smallest change for the maximum likelihood method.
Existing methods to identify changes in consultation rates can introduce substantial bias which depends on sample size and aggregation period. The direct maximum likelihood method was less prone to this bias than other methods and offers a robust, population-level alternative for estimating the timeliness of cancer diagnosis.
Characteristics of patients with lung cancer in clinical practice and their potential eligibility for clinical trials evaluating tyrosine kinase inhibitors or immune checkpoint inhibitors
2022, Cancer Epidemiology
In- and exclusion criteria of randomized clinical trials (RCTs) aim to include a homogeneous study-population. This study compared characteristics of lung cancer patients from phase III RCTs evaluating tyrosine kinase inhibitors (TKIs) or immune checkpoint inhibitors (ICIs) with characteristics of lung cancer patients in a real world setting in the United Kingdom.
A retrospective study was conducted using the Clinical Practice Research Datalink GOLD. Patients (N = 9239) with a first ever lung cancer registration between 2014 and 2018 were identified. Eligibility for inclusion was assessed for twelve RCTs (evaluating TKIs or ICIs). Reasons for potential exclusion and the number of unmet criteria were assessed for each RCT independently. OS was assessed using Kaplan-Meier and Cox proportional hazards analyses.
The proportion of potentially eligible patients was 74.3% and 51.9% for TKI and ICI RCTs, respectively. History of another malignancy, renal insufficiency or concomitant drug-use were main reasons for exclusion. OS was considerably longer for potentially eligible patients. Hazards ratios varied from 1.17 (95% confidence interval, 1.11–1.24) to 1.35 (1.20–1.42) across the RCTs.
This study showed that a considerable proportion of lung cancer patients in a real-world setting would have been ineligible for participation in phase III RCTs and that potentially ineligible patients experienced a shorter OS.
Long-Term patterns of cancer incidence among patients with and without type 2 diabetes in the United Kingdom
2022, Diabetes Research and Clinical Practice
Studies using contemporary cohorts are needed to assess the association between type 2 diabetes and cancer.
Using the United Kingdom Clinical Practice Research Datalink, we matched patients with type 2 diabetes between 1988 and 2019 to patients without type 2 diabetes. Poisson regression models were fit to estimate incidence rate ratios (IRRs) with 95% confidence intervals (CIs) for cancer. In secondary analyses, we determined whether the strength of the association varied with calendar time and whether patients with type 2 diabetes had a higher incidence of being diagnosed with multiple cancers during the follow-up period.
890,214 patients with type 2 diabetes were matched to an equal number of patients without type 2 diabetes. Patients with type 2 diabetes had a higher cancer incidence than patients without type 2 diabetes (IRR 1.19, 95% CI 1.18–1.21). The IRR was higher 2010 onwards (IRR: 1.25, 95% CI: 1.23–1.28) compared with the association in previous years. Overall, patients with type 2 diabetes had a 5% higher incidence of being diagnosed with multiple cancers (IRR: 1.05, 95% CI: 1.04–1.07).
The results of this large population-based study indicate that type 2 diabetes is associated with an increased risk of several cancers.
Trends in time to cancer diagnosis around the period of changing national guidance on referral of symptomatic patients: A serial cross-sectional study using UK electronic healthcare records from 2006–17
2020, Cancer Epidemiology
UK primary-care referral guidance describes the signs, symptoms, and test results (“features”) of undiagnosed cancer. Guidance revision in 2015 liberalised investigation by introducing more low-risk features. We studied adults with cancer whose features were in the 2005 guidance (“Old-NICE”) or were introduced in the revision (“New-NICE”). We compared time to diagnosis between the groups, and its trend over 2006—2017.
Clinical Practice Research Datalink records were analysed for adults with incident myeloma, breast, bladder, colorectal, lung, oesophageal, ovarian, pancreatic, prostate, stomach or uterine cancers in 1/1/2006–31/12/2017. Cancer-specific features in the year before diagnosis were used to create New-NICE and Old-NICE groups. Diagnostic interval was time between the index feature and diagnosis. Semiparametric varying-coefficient analyses compared diagnostic intervals between New-NICE and Old-NICE groups over 1/1/2006–31/12/2017.
Over all cancers (N = 83,935), median (interquartile range) Old-NICE diagnostic interval rose over 2006–2017, from 51 (20–132) to 64 (30–148) days, with increases in breast (15 vs 25 days), lung (103 vs 135 days), ovarian (65·5 vs 100 days), prostate (80 vs 93 days) and stomach (72·5 vs 102 days) cancers. Median New-NICE values were consistently longer (99, 40–212 in 2006 vs 103, 42–236 days in 2017) than Old-NICE values over all cancers. After guidance revision, New-NICE diagnostic intervals became shorter than Old-NICE values for colorectal cancer.
Despite improvements for colorectal cancer, scope remains to reduce diagnostic intervals for most cancers. Liberalised investigation requires protecting and enhancing cancer-diagnostic services to avoid their becoming a rate-limiting step in the diagnostic pathway.
Glucagon like peptide-1 receptor agonists and the risk of skin cancer among patients with type 2 diabetes: Population-based cohort study
2024, Diabetic Medicine

View all citing articles on Scopus

View full text

Validity of cancer diagnosis in a primary care database compared with linked cancer registrations in England. Population-based cohort study

Abstract

Aims

Methods

Results

Conclusion

Introduction

Section snippets

Cancer registry

Results

Discussion

Conclusion

Conflict of interest statement

Funding

Alarm symptoms in early diagnosis of cancer in primary care: cohort study using General Practice Research Database

BMJ

Validation of information recorded on general practitioner based computerized data resource in the United Kingdom

BMJ

Further validation of information recorded on a general practitioner based computerized data resource in the United Kingdom

Pharmacoepidemiol Drug Saf

Validity of the general practice research database

Pharmacotherapy

Validity of diagnostic coding within the General Practice Research Database: a systematic review

Br J Gen Pract

Validation and validity of diagnoses in the General Practice Research Database: a systematic review

Br J Clin Pharmacol

Colorectal cancer registration: the central importance of pathology

J Clin Pathol