The current population of Tayside is approximately 400 000 people and is comparatively small, even for the study of commonly prescribed drugs. However, drug exposure data in Tayside are only available from 1989 and cover only a limited set of drugs until January 1993 from when all dispensed prescriptions have been collected. Scottish doctors are conservative prescribers of new drugs, so new agents tend to penetrate the market a few years after their launch. This limits the ability to study new chemical entities, arguably the most important and interesting drug group to study. Offsetting these disadvantages, the profile of certain diseases, for example cardiovascular disease, is higher in Scotland than in other populations and consequently the prescribing of drugs used in the prevention and treatment of these diseases is proportionately higher.
Another weakness, but one that is common to many drug safety databases, is the inability to capture directly exposure to over-the-counter drugs or drugs prescribed in hospital. Perhaps more importantly, the diagnostic indication for prescribing is not available to the researcher. In some cases, the indication for drug use may be clear; for example, glyceryl trinitrate is used primarily for angina. However, difficulties arise when a drug has more than one indication for use, leading to misclassification of exposure or outcome. For example, beta andreoceptor blocking drugs can be given for indications varying from anxiety to hypertophic cardiomyopathy. This may be a potential source of error called confounding-by-indication that is difficult to adjust for in pharmacoepidemiologic research if the information is not available.
MEMO cannot contact patients directly to elicit information on possible confounding factors. However, with Ethical Committee approval GPs can do this in a collaborative manner. Primary care and hospital records can also be checked and some data on smoking and alcohol can be retrieved from them, although the quality does vary (Evans et al., 1998). This is also a method to identify outpatient diagnoses, which are not available electronically for record-linkage in MEMO. MEMO is therefore currently best suited for the study of serious drug toxicity that requires hospital admission.
One of the criticisms levelled at record-linkage studies is the inaccuracy of computerised medical diagnoses. The discharge diagnoses for SMR1 are abstracted from the clinical discharge summaries by specially trained coding clerks. These clerks on occasions have to interpret the "soft diagnoses'', such as symptoms, for which no cause can be found. In addition, non-standard terminology may be employed to describe an illness, for example eponymous terms, and so the coding of diagnoses may be imprecise. Computerised algorithms exist to detect and reject the most glaring errors, but errors of interpretation persist within any database. Several validation studies of the accuracy of hospital discharge data in Scotland have been performed comparing the coded diagnoses with diagnoses inferred by one or more senior doctors who have reviewed the original case records (Kohli and KnillJones, 1992; Park et al, 1992; Pears et al, 1992). The most pertinent of those studies carried out on Tayside data found 18% of internal medicine diagnoses to be clinically unacceptable (Pears et al., 1992). Since the publication of this study, steps have been taken, mainly for resource management reasons, to improve the diagnostic accuracy of computerised data by involving clinicians in quality control. This initiative has substantially improved the diagnostic accuracy of records.
Was this article helpful?
Suffering from Anxiety or Panic Attacks? Discover The Secrets to Stop Attacks in Their Tracks! Your heart is racing so fast and you don’t know why, at least not at first. Then your chest tightens and you feel like you are having a heart attack. All of a sudden, you start sweating and getting jittery.