General practice-based clinical trials

Each year, around 90% of the public see their general practitioner.1 General practice therefore should be an excellent environment to conduct community trials, because that is where the people are and, ideally, research should be conducted where it is to be applied.2 In this article, I discuss why general practice may be a difficult place to conduct trials for those from a non-primary care background and the means to overcome these difficulties. The article is based on a presentation at the MJA Clinical Trials Research Summit last year. The emphasis here is on conducting clinical trials in general practice, and not on research capacity building in general practice or on GPs’ selection of trials to participate in.

Clinical researchers from a non-primary care background most often want access to general practice for case finding. In a common scenario, a letter arrives in a GP’s in-tray from a reputable hospital or institute about an uncommon disease. The GP is not likely to have many patients listed with the disease, even if all of the patients’ diagnoses had been recorded in the GP’s clinical software. The request also has a long list of inclusion and exclusion criteria that make the whole process inefficient and time-consuming. Under these circumstances, general practice gets a reputation as a difficult place for clinical researchers to work in.

A more fruitful approach is to conduct research on a condition of interest to GPs and their patients. This means that GPs will see the research as clinically important and relative to their scope of practice. It is also important that the methods used recognise the complexity of general practice and are sufficiently flexible to adapt to different modes of practice (Box).

General practices are clinical services, mostly operating as small private businesses, and therefore are not set up as research centres. In recognition of this, workloads associated with clinical trials must not be onerous for GPs and practice staff, and reasonable practice payments should be made if there is use of practice facilities. Continuing professional education and development points should be offered for the Royal Australian College of General Practitioners (RACGP) and Australian College of Rural and Remote Medicine (ACRRM). It is best to incorporate these as much as practicable into the research endeavour, rather than as an add-on, to reduce GP and research staff workload. In the RACGP Quality Improvement and Continuing Professional Development Program, GPs now receive Category 1 points by participating in a trial, but must conduct extra individual work related to their own participating patients.

Engaging an experienced researcher in general practice as a chief investigator should increase GP involvement. Participating GPs should also be considered for co-investigator status, especially if they are involved with recruiting participants for the trial and the relationship is ongoing.

The study protocol should detail the general recruitment strategy for GPs and practices. Using an established general practice research network is ideal as it provides access to other GPs and practices with exposure to research. There are such networks scattered around the country, for example, the Victorian Primary Care Practice-Based Research Network, but unfortunately there are no national networks like those in Europe and North America.3 If a network does not exist, which is usually the case, a recruitment network needs to be established. Previously, Divisions of General Practice (DGP) membership databases were a good method of contacting GPs, and still are where they exist. Medicare Locals sites may provide the best means of contacting GPs in the future. As in any clinical trial, these methods need to be tested in pilot studies.

Potential trial participants can be identified by several methods. Investigators or their representatives can make direct contact with patients through posters in the waiting room, advertising in practice newsletters, or simply face-to-face in a waiting area or consulting room (after reception staff have informed patients of their presence). The direct contact approach is time-consuming, so it is usually reserved for studies where there is broad eligibility, such as a survey, rather than a clinical trial. Smoking cessation trials and interventional trials for common conditions may make use of this technique. Another approach is through GPs, clinic nurses or other staff. It is important that the staff involved are familiar with the trial protocol, and can identify appropriate individuals. Sampling bias is inevitable because individuals volunteer for the trial, but selection bias is reduced by recruiting participants from the community. Selection bias can be further reduced by keeping inclusion and exclusion criteria restricted to existing evidence-based indications and contraindications for an intervention.

Direct mailing or telephoning based on clinic databases is more efficient and will be discussed later. Indirect patient contact by appeals through media or advertisements to the general public may attract good numbers, but this approach is very likely to include large numbers of ineligible individuals. This approach is only suitable for trials where no direct GP contact is required, as failure to recruit GPs also renders their patients unable to participate.

Ethics committees may have concerns about the ongoing and comprehensive therapeutic relationship between GPs and their patients, but this really is no different to what happens in the secondary care sector. They also often raise privacy problems with accessing clinical lists for the purpose of contacting potential participants. This concern is shared by GPs and practice managers. It can be resolved by making the research assistant an agent of the practice, generating invitation letters on behalf of the GP rather than the study and not collecting patient information until volunteered by the patient at first study contact. Involving many GPs in a research endeavour for the first time also raises medicolegal queries from GPs about liability and practice accreditation. Trial insurance may be required. Researchers should make contact with medical defence organisations to make sure that the GPs’ regular practice indemnity will cover the trial and reassure GPs to this effect. Concerns about patient confidentiality and practice accreditation should be addressed in the same manner.

The second Australian National Blood Pressure study (ANBP2) involved 1594 general practices, 2681 GPs and 6083 participants.4 GP co-investigators were recruited through letters generated from DGP databases, supplemented by dinners, practice visits and cold-calling to non-responders. A manual search of health records was required to generate a list. GPs were asked to highlight their suitable patients. These patients were sent a letter of invitation from the GP and were asked to make an appointment with a research nurse at the practice. A screening visit was conducted by a study nurse at the practice before a face-to-face review by the GP co-investigator to assess the patient against the inclusion and exclusion criteria. Randomisation of patients was done by the research nurse. Using the same strategy, the Aspirin in Reducing Events in the Elderly study has recruited 867 practices, 1982 GPs and 9584 participants to December 2012.5 The screening-to-randomisation ratio in this trial is 77% (compared with 11% in the ANBP2) as a result of pre-screening by searching clinical practice databases and subsequent telephone calls by patients.

Providing researchers engage with GPs and practice staff, clinical trials can be conducted efficiently in the primary care setting. Results from clinical trials in the primary care setting are more generalisable to the community than those conducted in the tertiary care setting.