General practitioners’ use of computers for prescribing and electronic health records: results from a national survey

Reports of problems with quality of care and patient safety usually describe inpatient settings,1 but as many as one in 10 general practice patients experienced an adverse drug event in the past 6 months.2 Health information technology (HIT) has been proposed as an important strategy to combat medical errors and quality-of-care deficits.3

The Australian Government has made HIT a cornerstone of efforts to improve health-care quality, safety, and efficiency, especially in general practice. In the late 1990s, incentives were provided to general practices to install computers and clinical software packages to be used for prescribing and transmission of clinical data.4 That initiative was at least partially responsible for the rise in the use of computers by general practitioners, from 15% in 1997 to 70% in 2000.5 Other government projects are addressing other aspects of HIT, such as standards and interconnectivity.6

Since a 2001 survey of general practices,7 there has been little description of how Australian GPs use computers in clinical care. We provide an updated view of GP computer use for a wide variety of clinical functions, especially prescribing and electronic health record functions. We studied the use of clinical computer software packages (commonly called “clinical packages”) by GPs, who, following the Medicare definition of a GP,8 derived at least 50% of their Medicare income from unreferred patient visits. We sought to answer two questions: Are GPs using the full range of functions available in their clinical packages?; and if not, What are possible explanations for the infrequent use of some functions, especially ones that may contribute to improved patient safety and quality of care?

The Information Branch of the Australian Government Department of Health and Ageing provided a stratified random sample of GPs from the Medicare GP database on 5 October 2005. To be eligible for selection, GPs must have submitted at least 375 Medicare claims in the quarter ending 30 June 2005.9 From an eligible population of 18 172 GPs, a stratified sample of 3000 was selected, consisting of 70% urban/regional GPs and 30% rural/remote GPs (we oversampled rural/remote areas to allow statistical comparisons between rural/remote and urban/regional GPs, but did not make such comparisons in this study). For this study, urban/regional was defined as Rural, Remote and Metropolitan Area categories 1–3, and rural/remote as categories 4–7.10

Of the 3000 GPs in the sample, 1186 returned completed questionnaires, giving a response rate of 39.5%. Our sample had a similar sex distribution to that in the Medicare database of 18 172 GPs in active practice, but there were modest differences in age, location of medical training, and state in which practices were located (Box 1). The significant difference between our respondents and the Medicare GP database for practice location was the result of our stratification, which ensured that 30% of the sample would be from rural and remote regions compared with 21% of GPs in the Medicare database.

Most practices had the computer software and hardware to perform administrative and clinical functions, and most (78.3%) had a high-speed internet connection (Box 2). Over half the practices (55.6%) had received a Practice Incentives Program (PIP) payment for information technology, and nearly a third (31.5%) had received the Broadband for Health incentive payment. Nearly 90% of GPs reported they use the computerised clinical package.

Virtually all GPs who used a clinical package reported regularly using electronic prescribing (98%), and high percentages also updated medication lists, checked for drug–drug interactions, and checked for drug allergies (Box 3A). Smaller percentages checked drug–disease interactions and recorded their reason for prescribing.

The most commonly used electronic health record functions were ordering laboratory tests, updating patient allergy information, and generating patient health summaries, all used by 83% or more of the GPs who used a clinical package (Box 3B). Less commonly used functions were creating and updating disease management plans, recording progress notes, accessing educational material for patients, and conducting clinical audits.

Patient lists, or registries, which are especially important for managing chronic conditions,18 were used relatively less often (58% or fewer of GPs who used clinical packages; Box 3C). Also, while GPs frequently used the automatic alerts for medication safety (eg, drug–drug interactions), they did not regularly use optional electronic decision-support functions during the consultation, such as review of prescribing information or assessment of risk factors (Box 3D). Among GPs using a clinical package, Medical Director was used by 73.1% (Box 3E).

Women used prescribing functions more often than men, though the difference was statistically significant only for drug–drug interaction checking (92.3% v 85.1%; P = 0.001). Similarly, women tended to use electronic health record functions, such as ordering laboratory tests (88.9% v 82.6%; P = 0.007) and running recall systems (81.6% v 75.4%; P = 0.02), more often than men. In contrast, men were the more frequent users of computers for generating lists of patients for all three items: patients with a specific condition (61.7% v 52.4%; P = 0.004), taking the same medication (42.7% v 32.7%; P = 0.002), and needing vaccination (45.0% v 38.6%; P = 0.05). Men were also somewhat more likely to access computerised information during the consultation, though the difference was statistically significant only for reviewing prescribing information (79.0% v 73.5%; P = 0.04).

Computerisation in Australian general practice has grown, with 90% of GPs now using a computerised clinical package. While prescribing is the most commonly used electronic function (98% of those who use a clinical package), medication safety functions, such as checking drug–drug interactions, are also frequently used. GPs also reported widespread use of electronic health record functions, including ordering laboratory tests (85%), updating allergy information (84%), and generating health summaries (84%). These results reflect significant progress since 2001, when, among GPs in computerised practices, 71% used electronic prescribing and 42% generated health summaries electronically.7 Comparable data on the percentage of GPs using a clinical package was not available in the 2001 study, although 86% of general practices had a computer at that time.

There are some notable gaps in the use of certain electronic clinical functions. Among doctors using clinical packages, only 65% regularly record electronically the reason for prescribing a medication. Similar percentages electronically record progress notes (64%) or access patient educational material (63%). Of GPs who use a clinical package, 43% use a computerised list of patients needing vaccines, and under 40% use a computerised list of patients taking the same medication. Yet, these less-used functions have been shown to improve the management of chronic illness, as well as more acute conditions.19 Finally, having up-to-date, research-based information to help with decision making has the potential to eliminate errors and improve the quality of care.20,21

The variation in use of electronic clinical functions may be related to the functions’ attributes, as suggested by diffusion theory22 and the technology acceptance model (TAM).16 Diffusion theory’s “relative advantage” (ie, improvement over existing methods) may explain the near-universal use of electronic prescribing, which eliminates legibility problems and attracts PIP payments. In contrast, the lower frequency of accessing electronic information during consultations may reflect low “usefulness” (TAM term for productivity),16 so that GPs may perceive that taking even 1 or 2 minutes to seek information on the computer will reduce their productivity. The lower adoption of electronic progress notes, on the other hand, may reflect the lack of perceived “ease of use”, another TAM construct. Unlike prescribing, which can be done with a few mouse clicks, most electronic progress notes require typing relatively large amounts of text, a skill that some GPs may lack.

Our study has several limitations. The 39.5% response rate may mean that the respondents were not representative of the population of Australian GPs. However, our analyses suggest that our respondents did not differ substantially on demographic and practice characteristics from the overall population of Australian GPs. Another limitation is that survey responses were self-reported, and we do not have independent verification of the computer uses reported by GPs.

Further work is necessary to examine whether GP reports of information technology use match their actual use. Also, the differences we found in use between men and women could be explored through in-person interviews and focus groups.

In conclusion, Australian general practice has achieved near-universal computerisation in less than 10 years. Electronic prescribing alone is likely to have improved efficiency and quality of care, and reduced medication errors. Improving adoption of other electronic functions is likely to lead to additional health gains, especially in managing chronic conditions.