Design, setting and participants: Content analysis of advertisements displayed by Medical Director version 2.81 (Health Communication Network, Sydney, NSW) in early 2005; thematic analysis of a debate on this topic held on the General Practice Computer Group email forum (GPCG_talk) during December 2004.
Results: 24 clinical functions in Medical Director contained advertisements. These included 79 different advertisements for 41 prescription products marketed by 17 companies, including one generic manufacturer. 57 of 60 (95%) advertisements making a promotional claim appeared noncompliant with one or more requirements of the Code. 29 contributors, primarily GPs, posted 174 emails to GPCG_talk; there was little support for these advertisements, but some concern that the price of software would increase if they were removed.
In 1999, the Australian Government provided financial incentives for the uptake of computers by Australian general practitioners. This was followed by an increased proportion of GPs writing prescriptions with the aid of a computer, from around 50% in 1999 to more than 90% in 2004.1 One prescribing software package (Medical Director) achieved market dominance; it is the only Australian prescribing software containing pharmaceutical advertisements.
Medicines Australia’s Code of Conduct sets standards for the ethical marketing and promotion of prescription pharmaceutical products in Australia.2 Compliance with the Code is the responsibility of the pharmaceutical companies, not the software company. We aimed to assess the location, frequency and type of advertisements for prescription pharmaceuticals appearing in Medical Director, and their compliance with the relevant sections of the Medicines Australia Code of Conduct and associated guidelines.2,3 In addition, views on this topic posted to the General Practice Computer Group email discussion list (GPCG_talk)4 were thematically analysed. The GPCG is the peak body for general practice computing, established in 1997, funded by the Australian Government Department of Health and Ageing, auspiced by the Australian Medical Association until 2001, and then by the Royal Australian College of General Practitioners.
All clinical functions available in Medical Director version 2.81 (Health Communication Network, Sydney, NSW) were explored for the presence of advertising. We also used the advertising viewer program accompanying the software to sequentially access, count and study the advertisements in the program’s advertisement database. This database contains most, but not all, of the advertisements that appeared when using the software. For statistical analysis, we counted advertisements that differed in at least one characteristic: size of the image, animation or information.
A template for the objective content analysis of the advertisements was developed. It included an assessment of whether the following items required by the Code were present: product name (brand and generic), company name, promotional claims, a statement to review the product information, the Pharmaceutical Benefits Schedule (PBS) listing, referral to further information, identification and details of the references.
Certain clinical functions provided additional information around banner advertisements that was not present when the advertising viewer was used. For example, an advertisement might appear in one place with the adjacent statement “Before prescribing please refer to product information in this program by clicking on MIMS PI”, making the advertisement compliant with the Code. However, the same advertisement could appear in other places without such a statement (eg, in the patient educational leaflet browser, the MIMS pharmaceutical database or when viewing Consumer Medicine Information). Hence, we regarded information as present only if it was contained in the body of the advertisement.
The size of the advertisement, whether it was animated, and the legibility of the generic name were assessed. Our computer screens had a resolution of 1024 × 768 pixels. Three reviewers with extensive experience in critical appraisal of drug advertisements (K J H, A I V, E R) assessed the template criteria and compared each advertisement with relevant provisions of the Code independently. For objective measures, such as presence or absence of a statement, any difference between reviewers was resolved by re-examination of the advertisements and discussion. Legibility of the generic name was considered a more subjective assessment, and individual variation in this assessment was not modified.
All email postings to GPCG_talk with the subject “Ads in EHR software” were collected from the discussion sequence that occurred between 6 and 31 December 2004, and a thematic analysis performed manually. The emails were in response to an initial posting by one of the authors (K J H), suggesting that pharmaceutical advertisements in clinical software should be banned and replaced with unbiased information resources. Critical examination of the ensuing discussion was viewed as a means of exploring the views of a self-selecting group of physicians who were both articulate and confident in discussing this issue.
Advertisements were found in 24 clinical functions of Medical Director (Box 1). Some advertisements appeared randomly, while others were targeted toward the particular clinical function in use.
Several identical advertisements had multiple entries in the database; others only appeared once in the database, but appeared preferentially in relevant clinical functions. The advertising viewer displayed 79 different advertisements for 41 prescription pharmaceutical products marketed by 17 companies, including one generic manufacturer. There were 20 large advertisements (750 × 500 pixels; all static images) and 59 banner advertisements (480 × 75 pixels; 36 static images; 23 animated, containing multiple images).
The generic name was judged illegible by one or more reviewers in 44 of 79 (56%) advertisements, representing a potential breach of Section 3.10.6 of the Code (Box 2). Some animated advertisements were also adjudicated to be in potential breach of Section 3.10.7 (Box 2). Sixty advertisements made a promotional claim, including 41 (69%) of the banner advertisements and 19 (95%) of the large advertisements. If a claim is made, the Code requires additional information to be present. Fifty-seven (95%) advertisements appeared noncompliant with one or more of these requirements (Box 3).
Most promotional claims included broad slogans such as “Your first choice CCB”, “Power Plus”, “Power you can trust”, “You can’t keep a good woman down”. We did not systematically assess the accuracy of all claims. However, certain claims (Box 4 and Box 5) appeared to be in breach of Section 1.3 (Box 2). In addition, advertisements positioned in clinical functions designed to be shared with patients appeared to be in potential breach of Section 3.10.10 (Box 2).
The topic of “Ads in EHR software” on GPCG_talk attracted 29 contributors, primarily GPs, who posted 174 emails on this topic over 4 weeks. Initial coding of the debate indicated that, with one exception, there was little support for advertisements in clinical software. However, some contributors were worried that legislative measures to remove pharmaceutical promotion from prescribing software would increase its price. Others noted that software with equivalent functionality is available at a similar cost but, in the absence of standards for electronic health records, it is not easy to change software. There was support for balancing the selective information provided by promotional material with more objective sources of therapeutic information such as Therapeutic guidelines12 and the Australian medicines handbook13 and the Australian adverse drug reactions bulletin.14
Subsequently, more detailed analysis revealed five major themes (Box 6). Views were expressed in support of both sides of each of these themes.
Our analysis of pharmaceutical advertisements in prescribing software raised a number of methodological issues. Assessment of legibility proved problematic, as reviewers of differing age and visual acuity varied in their assessment. Nevertheless, all agreed that up to half the advertisements had illegible generic names. The Code states that information in advertisements should be legible, but provides no legibility criteria for electronic media. Like others, we believe that the generic name (and other important information) should be reproduced at the same size, font, colour and background as the brand name in all advertisements, labelling, product and consumer information.15,16
Despite the difficulties inherent in counting the advertisements, our study showed that doctors who use Medical Director (and patients who observe the screen) are exposed to many pharmaceutical advertisements. Furthermore, many of the advertisements were judged noncompliant with one or more requirements of the Code. Common problems included missing information, illegible generic names, and claims that were unsubstantiated or appeared not to be in accord with the published literature.
Lack of appropriate PBS listing information was a particular concern, as Medicines Australia and its member companies agreed to implement an initiative in the 2002 federal budget that all pharmaceutical promotional items would include detailed PBS information.3
Although direct-to-consumer advertising of prescription pharmaceuticals to consumers is prohibited in the Therapeutic Goods Act 1989 (Cwlth), the Code appears to condone it in Section 3.10.10 (Box 2). Many clinical functions in Medical Director, if shared and discussed with patients, are likely to assist them. However, most of these functions contain advertisements for prescription pharmaceuticals.
Given the array of problems we found, it seems that many pharmaceutical companies may not be providing advertisements in compliance with the Code, despite the efforts of Medicines Australia. In addition, finalised Code complaints show that some companies have been associated with repeated code breaches over several years, despite the sanctions applied by Medicines Australia.17-19 This failure of the self-regulatory process has important public health implications. Pharmaceutical promotion has been shown to influence physicians’ prescribing20 and to result in PBS cost blowouts due to prescribing of more expensive drugs.21 Pharmaceutical promotion in prescribing software, occurring at the time of physician–patient decision-making, may be more powerful than promotion in medical journals, gimmicks and give-aways.
Similar concerns about pharmaceutical promotion in prescribing software were posted to the GPCG_talk discussion forum. Those who contribute opinions to the GPCG_talk forum are a self-selected group of computer enthusiasts who are not shy of engaging in an often robust exchange of views; although they are unlikely to be representative of all GPs, their opinions are of value. The Australian Consumers’ Association (ACA) has also conducted an online poll of consumers on this subject, and this revealed a high level of disquiet.22 The ACA poll can be criticised for asking leading questions and being subject to manipulation by people voting more than once. Regardless, both GPCG_talk and the ACA poll show that we are not the only ones perturbed by pharmaceutical promotion in prescribing software.
Our concern is not just the apparent violations and Code inconsistencies, although we hope that Medicines Australia will address these issues. We are also concerned that an industry that spends 2–3 times as much money on marketing as on research and development23 distorts the information flow to health professionals and consumers, creating unhealthy and expensive prescribing habits as well as consumer expectations of a “pill for every ill”.24 A Dutch study concluded that, on balance, pharmaceutical marketing is welfare-negative because it results in doctors prescribing more expensive drugs.25 Just as public health campaigns have progressively reduced promotional avenues for the tobacco and alcohol industries, we believe the same should happen with pharmaceutical promotion — drug advertisements in clinical software are a good place to start. Without such action, the current unsustainable growth of PBS costs is unlikely to moderate.
In conclusion, a substantial proportion of advertising in Medical Director appears to lack compliance with the Medicines Australia self-regulatory Code of Conduct. Compliance is the responsibility of the pharmaceutical companies, not that of the software companies. Some will argue that the solution to this problem is to tighten the Code with respect to prescribing software and encourage better compliance by pharmaceutical companies; another consideration might be for software companies to play a role in assessing compliance. However, given the potential public health consequences of promoting prescription pharmaceuticals through software shared with patients, we argue that pharmaceutical promotion should be eliminated from prescribing software. In addition, as contributors to GPCG_talk noted, there is an urgent need for the Australian Pharmaceutical Advisory Council and the Australia Health Information Council to formulate a health information policy that ensures that Australian best-practice information resources are incorporated into clinical computer systems in both hospital and general practice.
1 Clinical functions accompanied by advertisements
Prescribing, including select drug by class, dose calculator, print prescription
Physical activity prescription
Record blood pressure
Cardiovascular risk calculator
Record height/weight/waist circumference
Record blood glucose
Record INR (international normalised ratio)
Calculate respiratory function
Mental state examination
Renal function calculator
Medical imaging ordering
Depression recovery scale
Asthma action plan
Drug resource lookup; ie, MDRef (drug list), MIMS PI (drug information), MIMS CMI (consumer medicine information)
Patient education material
Medication/diagnosis suggests cardiac problems
2 Extracts from the Medicines Australia Code of Conduct2
All information, claims and graphical representations provided to health care professionals and members of the general public must be current, accurate, balanced and must not mislead either directly, by implication, or by omission. Claims must be referenced where there is a possibility that a reader may be misled if the source of the reference is not disclosed.
3.10.10 As prescribing software packages may include information which could be shared with patients by prescribers, the content of any advertisements, particularly graphics, must be such as to not cause any offence, alarm or concern or give rise to unrealistic expectations of benefit when viewed by members of the general public, including children.
3.10.11 A company shall not negotiate or accept any offer from a software manufacturer to achieve a trigger or mechanism that results in the preferential presentation of its own product or a less favourable presentation of a competitor’s product in a way that would directly influence a prescriber’s choice.
3 Compliance of advertisements*
4 Examples of potential breaches of Section 1.3 of the Medicines Australia Code of Conduct (false or misleading claims)
5 Example of a potentially misleading claim
The claim “a large body of clinical evidence showing no significant increase in cardiovascular risk” is arguably misleading in view of official warnings from the Therapeutic Goods Administration that celecoxib may increase the risk of cardiovascular events.7
6 Representative comments from the GPCG_talk electronic discussion forum, grouped into themes