Identifying typologies of General Practitioners through analysis of prescribing & diagnostic test ordering decisions

There is a longstanding interest in variation in the prescription of drugs between general practitioners (GPs) and between general practices. In some cases (such as antibiotics, benzodiazepines or opioid analgesics) there is potential overuse. In other cases (antihypertensives, statins) there is potential underuse of drugs. There is also great variation between GPs and between general practices in their use of diagnostic tests. The characteristics of the patients seen by these GPs or general practices explain only some of the variation. Less attention has been paid to the idea that individual prescribers (GPs) and general practices are important determinants of variation. That GPs have a propensity to prescribe and general practices have a high or low prescribing culture.

It has been demonstrated that used Facebook users’ “likes” can be used to reliably identify their personality characteristics and preferences.[1] This project proposes to characterise GPs’ using their prescribing and diagnostic test ordering decisions. It is intended that this should provide insights into variation in clinical behaviour between clinicians.

Two independent analyses of large datasets of prescribing data have observed prescribing rates of quite different types of drugs are strongly correlated. In the US benzodiazepine prescribing at the level of individual prescriber was highly correlated with antibiotic prescribing and prescribing of high-risk medications for the elderly.[2] As these are entirely different types of medications, this suggests a single underlying construct determines high prescribing for a range of medications. The authors suggested this might be due to an underlying personal characteristic, hypothesising that the physicians’ consultation style, use of education and communication in consultations might be a factor. They noted that female physicians were lower prescribers.

A similar UK study analysed two databases, one of primary care records, the other of prescription reimbursement data. It also found strong correlations at the level of general practice between antibiotic prescribing rates and prescribing rates of all other drugs.[3] These correlations were only modestly attenuated after adjustment for patient characteristics. In the prescription reimbursement database the adjusted regression model of prescription of all other medicines (i.e. everything except antibiotics) explained 56% of the variation in antibiotic prescription and in the primary care database 62.9% of the variation in antibiotic prescription. Rates of antibiotic prescription correlated strongly with prescription rates of specific drug classes unrelated to antibiotics, including: proton pump inhibitors, renin-angiotensin system drugs, antiplatelet drugs, drugs used in nausea and vertigo, and selective serotonin reuptake inhibitors. The authors concluded that propensity to prescribe was the single strongest determinant of antibiotic prescribing.

Previous research has characterised general practitioners (GPs) by their prescribing and diagnostic styles and used this to predict health outcomes in patients.[4] GPs were characterised as having an integrated style, with low diagnostic test, prescription and referral rates; an interventionist style with by high diagnostic test, prescription and referral rates; or a minimal diagnostic style characterised by low diagnostic test rates but high prescription and referral rates. Patients of general practitioners with an integrated style consulted less frequently and self-reported better health.

More widely, it has been observed that investigation of variation in health care use has overlooked the role of style of physician practice.[5]
There is evidence that GP prescribing is influenced by an underlying propensity to prescribe. This may also be true of diagnostic tests. These characteristics may be important determinants of health care use and of patient outcomes. The proposed doctoral thesis aims to will investigate whether clinical prescribing and diagnostic test behaviour can be used to characterise GPs and to predict health care utilisation by patients.

Objectives:
1. Identify a range of index prescribing decisions attributable to individual GPs and develop appropriate measures of prescribing rates for these index prescribing decisions.
2. Identify a range of index diagnostic test ordering decisions attributable to individual GPs and develop appropriate measures of diagnostic test rates for these index diagnostic test decisions.
3. Undertake statistical analysis to determine the number of factors underlying GP propensity to prescribe and to order diagnostic tests
4. Describe the variation in the propensity to prescribe and propensity to order diagnostic tests over time within the same GP and variation by general practice.
5. Investigate the relationship between propensity to prescribe and propensity to order diagnostic tests and health care use (visit frequency) in primary care

Data source:
Large dataset of anonymised electronic primary care records