PRIMARY CARE DATA FROM THE HEALTH IMPROVEMENT NETWORK (THIN)
Background
Since EPIC began in 1995, it has achieved unparalleled success as a facilitator for medical research on retrospective primary care databases. EPIC provides access to and advice on patient level data from the General Practice Research Database and the Health Improvement Network Data. Through its extensive experience, EPIC strives to provide an outstanding, flexible service to the worlds leading pharmaceutical companies and medical research organisations.
Founded by its Chairman Dr Alan Dean, a pioneer in computerised General Practice record keeping, EPIC has worked for many years analysing primary care data. Some of EPIC’s highly qualified team were part of the original VAMP group who started such research in the late 1980’s.
EPIC has contributed significantly to raising the profile of the GPRD as a valuable tool for healthcare providers, the pharmaceutical industry, and academia. The GPRD is now a recognised global medical research resource utilised extensively in the areas of drug safety, epidemiology and outcomes research.
EPIC is now able to offer access to The Health Improvement Network (THIN) database of anonymised primary care patient records. THIN data are an alternative resource of rich patient level information for medical research.
EPIC facilitates access to the data in a number of ways:-
· Providing on site datasets for organisations to conduct their own research
· Carrying out tailor made research on behalf of clients
· Supporting external researchers within EPIC’s own offices
· Helping obtain primary, secondary and lifestyle data directly via the GP and patients
THIN
The Health Improvement Network (THIN) was set up in 2002, and is a collaboration between two companies with an established name in primary care computing: EPIC, a non-profit making organisation that facilitates access to electronic research data, and Cegedim, a European healthcare software and research company and the parent company of INPS, who develop and supply the Vision general practice computer system to over 1800 GP practices in the UK.
INPS have written unobtrusive automatic data collection software for Vision practices that have joined THIN. The data are collected weekly by THIN and sent monthly to EPIC who, supply the THIN data to approved researchers for studies. Such research is approved by the nationally accredited ethics committee, South East Multicentre Research Ethics Committee (SE-MREC).
Data collection
THIN collects anonymous patient data, including past history and prescriptions, from practices using the Vision practice management system where it has already been recorded in the normal routine of the practice. Data collected from the practice system are anonymised at the collection stage; therefore, identifying information is never available to THIN. These data can be processed to provide researchers with access to demographic, medical, and prescription information at an individual patient level. In addition, there is information on referral to specialists, diagnostics, and laboratory results, some lifestyle characteristics, and other measurements taken in the GP practice. The data are organised in files by individual practice and provide a longitudinal record for each patient.
THIN data collection commenced in November 2002. More than 330 practices using Vision software have joined the scheme. Currently, the THIN dataset contains data from 333 practices with a total of over 5.7 million patients of which approximately 2.4 million are actively registered with the practices and can be prospectively followed. The remaining patients have historical data but have either left the practice or died. Most of these contributing practices have recorded several years of data on their system and where practices previously used VAMP systems, data entries extend as far back as 1985.
Quality of THIN Data
As well as establishing a new primary care research data resource, THIN is improving the data recording of contributing practices. In addition to providing free training to the practices, THIN data are subjected to computerised validation to quantify the completeness and accuracy of recording. These results are fed back to the practices with information on how to improve quality and correct omissions they are also supported by free training seminars to address any issues and encourage more efficient use of their computer systems. The comprehensive information on data quality of individual practices is shared with researchers so that, if appropriate, they can select subsets of practices with strengths in specific areas of research.
Advantages of THIN Data
One of the main advantages of such a non-interventional, retrospective database is that it allows maximum scope for researchers in terms of study design therefore case control; nested case control cohort and cross over designs are all supported. Another advantage is the size; as it includes data on over five million individuals. It is population-based, permitting calculation of incidence rates and, it is continually updated, permitting investigation of the effects of newly-marketed drugs. The THIN Data includes for some patients longitudinal data from 1985, thus making possible longitudinal studies and long-term follow-up of patient histories. Also access to general practitioners records and to hospital discharge summaries are available. The THIN data are not subject to recall bias or interviewer bias, as there is no reliance on patient recall or interviewers to collect the data. Finally, the THIN population is representative of the majority of the population within the UK, as 99% of adults are registered with a GP under the NHS.
Standards and scope of recording
A complete computerised record of the patient’s healthcare is built up whilst the patient is registered with their GP. Significant historical events are entered from a summary of the transferred paper notes. However, the THIN dataset is constructed from data collected from the GP’s medical record system and will reflect only those events that are deemed to be relevant to the patient’s care.
Diagnoses
All medical conditions and symptoms reported to GP are recorded on the computer during a consultation. These consultations build up long computerised medical histories for patients. Medical conditions are recorded using the Read Clinical Classification. This is a hierarchal system, and the codes may be cross-referenced to the International Classification of Diseases (contact the NHS Information Authority for further details). Information is routinely recorded on referrals to secondary care, including the type of specialty. Secondary information received by the GP is transcribed and entered retrospectively; this includes details on hospital admissions, discharge medication and diagnosis, outpatient consultation diagnosis, investigation and treatment outcomes.
Prescribing
Prescribing is particularly well recorded in terms of GP prescriptions since the computerised entry made by the doctor is also used as the prescription form with a copy being printed for the patient to present at the pharmacy. The computerised prescribing creates therapy records which are recorded using the Multilex coding system developed by First Databank. Prescriptions not issued from the computer, such as controlled drugs, immunisations, private prescriptions, and drugs prescribed during home visits should all be entered, but there is a possibility of under-recording with such items.
In most cases a medicine prescribed for the first time can be temporally linked with a medical event record (symptom or diagnosis) although there is not a permanent link between a therapy record entry and an entry in the medical records. Drugs prescribed by hospital doctors or other specialists will not appear in THIN data unless the treatment is to be continued. However, due to the constraints of specialist/hospital prescribing budgets many prescriptions issued outside of the GP practice will usually only be enough to cover the first 7 days. After this time the patient will be required to visit their GP for a further prescription, and again for any subsequent ones. These therefore will be entered into THIN data. Details of some prescriptions from ongoing outpatient specialist care (clinics etc.) may have been summarised by the GP, the degree of information will rely on its direct relevance to the GP management of the patient.
Additional Health
As well as GP consultations and details from other health care interventions, THIN data contains some information on lifestyle, including variables with a significant bearing on health such as smoking and alcohol intake, tests and lab results. Lab tests initiated by the GP are likely to be in the clinical data record. Many Vision practices are now electronically linked to the pathology laboratory. This makes it easier for the practice to store results in the patient’s electronic record. Those practices using a paper system of receiving results will manually code and enter any results that are considered to be important for patient care.
Postcode Variable Indicators (PVI)
EPIC can now provide anonymous postcode linked area based socioeconomic, ethnicity and environmental indices for studies using THIN Data. This information will provide important information about potential study bias and/or confounding which will enhance the usefulness of THIN Data for epidemiological, public health and health services research.
Uses
Since the database holds comprehensive prescribing and disease information it allows healthcare researchers to study the natural history of disease, monitor drug safety and carry out risk management studies. Prescribing analysis can examine therapy/treatment indication and use. In addition, these data can be used to perform outcomes research and pharmacoeconomic studies, including populating models.
2