Methods

Aim of the guideline

The aim of this guideline is to provide recommendations to aid primary care health professionals in their management of patients with chronic stable angina. Where possible recommendations are based on and explicitly linked to the evidence that supports them. It does not deal with unstable angina or myocardial infarction. The development group assumes that health care professionals will use general medical knowledge and clinical judgement in applying the general principles and specific recommendations of this document to the management of individual patients. Recommendations may not be appropriate for use in all circumstances. Decisions to adopt any particular recommendation must be made by the practitioner in the light of available resources.

Responsibility and support for guideline development

Agency responsible for guideline development

The guideline was developed under the direction of Dr Martin Eccles, Senior Lecturer, Director of Primary Health Care Research, The Centre for Health Services Research, University of Newcastle upon Tyne.

Source of funding

The development of the guideline is principally funded by the Division of Research and Development of the former Northern Regional Health Authority, now the Northern and Yorkshire Health Authority. Contributions to the funding have been made by the Medical Audit Advisory Groups of Durham, Northumberland and South Tyneside Family Health ServiceAuthorities.

Steering group

The development of the guideline is supported by a technical advisory panel whose members are: Dr Jeremy Grimshaw, Programme Director, Health Services Research Unit, University of Aberdeen; Dr Ian Purves, Director of the Sowerby Unit for Primary Care Informatics, University of Newcastle upon Tyne; and Ian Russell, Professor of Health Care Sciences, University of York.

Guideline development group

It was important that the group contained individuals from appropriate sectors (Russell et al., 1993); the sectors approached were general practitioners, practice nurses, patient groups, secondary care physicians and public health physicians. These group members were there to ensure adequate relevant discussion of the evidence and the recommendations in the guideline.

The specialist resource was a consultant cardiologist. He worked closely with the research team, discussing the literature search strategy, reviewing and summarising the literature and feeding this information back to the group. The task of literature review was shared with the second consultant cardiologist (12 lead ECG, exercise testing), a number of the general practitioners (smoking, exercise, diet, pattern of care) and the practice nurse (cholesterol). The patient in the group reviewed the available patient information leaflets. The small group leader had the role of ensuring that the group worked effectively. The research team were responsible for the pursuit of the group task, undertaken by the principal investigator, and the resourcing of the group.

The guideline group membership consisted of:

Dr Phil Adams, Consultant Cardiologist, Newcastle upon Tyne (Specialist resource)

Dr Malcolm Aylett, General Practitioner, Wooler

Dr Lawrence Bryson, Consultant Cardiologist, South Tyneside

Dr Martin Eccles, Health Services Researcher, Centre for Health Services Research (Guideline development leader)

Mr Phil Flinders, Patient, Morpeth

Dr Hugh Geoghegan, General Practitioner, Cleveland

Dr Ed Holmes, General Practitioner, Durham (Small group leader)

Ms Carol Johnson, Practice Nurse, Wylam

Mr Paul McNamee, Health Economist, Morpeth

Dr John Myers, General Practitioner, Morpeth

Dr Gordon Pledger, Public Health Physician, Newcastle upon Tyne

Dr Eddie Selby, General Pactitioner, Middlesbrough.

Meetings took place during a weekday afternoon. This arrangement ensured that meetings were long enough (3.5 hours) to ensure sufficient time for the task but also that the meetings would be held at a time of day when group members had not already completed a full day's work. All group members were offered reimbursement of their travelling expenses, and general practitioners and practice nurses were offered reimbursement for any locum expenses incurred. As the skills provided by the small group leader and the specialist resource were central to the functioning of the group they were paid a fee in recognition of their work. A light sandwich lunch was available before the meeting for those who wanted it.

Evidence identification, review and synthesis

The search strategy

The aim of the evidence review was to identify and synthesise relevant evidence within the published literature to allow recommendations to be based on evidence wherever possible (West, 1992). The refinement of the final search strategy was influenced by a number of pragmatic considerations.

The search strategy was carried out using the electronic database MEDLINE and covered the period January 1985 to August 1994 (Appendix 2). This ten year period was a pragmatic decision influenced by the volume of papers and the time and resources available.

All the searches carried out were limited to "english language" and to the age bands of 18 years and above using the MEDLINE options. Certain types of study were of more relevance to our purpose than others and so we conducted MeSH heading and free text searches using the terms: meta analysis; randomised controlled trials; reviews; cohort studies or case control studies. This also linked in with our need to grade levels of evidence (Sackett, 1986). The method of restricting the literature by limiting the search to mainstream journals (Sackett et al., 1991) was considered, but was rejected for two reasons: firstly it did not reduce the number of irrelevant documents substantially and resulted in rejecting papers on areas such as the management of care, which were already limited in number; and secondly because of the breadth of the subject the number of mainstream journals was still large. Additional specific MEDLINE searches were carried out in the following areas: decision making; isosorbide; myocardial infarction + secondary prevention; buccal.

In addition to the main MEDLINE search, references were identified from three other sources. The BIDS on-line database was searched (search terms "angina + management"). If there was no recent evidence in a clinically important area the specialist resource was asked to identify, from his personal knowledge, key articles published before 1985, the lower time limit of our electronic search. Also, although non-systematic reviews were not used for our evidence review, their reference lists were checked to ensure that we had identified relevant articles that they had cited. Relevant articles published pre-1985 were also identified in this way. We did not attempt to access the grey literature (unpublished reports and abstracts), nor did we identify letters in response to original articles.

Sifting the literature

The set of references generated by the search were sifted for relevance to the clinical area of the guideline. The initial sifting was done by a clinically qualified health services researcher on the basis of the article titles. Articles concerned with non-relevant clinical areas were removed as were editorials and letters. If there was any doubt about an article's relevance it was retained for the next round of sifting. This was performed by the specialist resource (the cardiologist) and was done on the basis of title and abstract (where this was available). This identified references that needed to be collected for more detailed assessment. The number of references rejected at each stage of this sifting process are listed in Table 1.

Table 1: Number of MEDLINE references remaining at various stages of the sifting process
Stage in the sifting process	1985-1994
All refs	38675
Restricted to English only	27982
Restricted to human and adult only	24194
Restricted by publication type	9165
Sifted on title (HSR)	1810
Sifted on abstract (consultant)	699
Methodological sift - accepted	286
Methodological sift - rejected	260
Not used	153

The total number of references identified was 957: 701 from MEDLINE (this includes two outside 1985-94 limit), 14 from BIDS, 87 from the specialist resource, and 155 from the citations of non-systematic reviews.

Reviewing the literature

The identified papers were reviewed against explicit methodological criteria set out in Appendix 3. These were informed by a number of sources (ter Riet, 1990a, 1990b; Kleijnen et al., 1991a, 1991b; Koes et al., 1991; Guyatt et al., 1994; Jaeschke, 1994a, 1994b). Due to the volume of literature, the reviewing task was extended beyond the single specialist reviewer by taking discrete areas of the literature and inviting group members to review these on behalf of the group. Reviewers were provided with explicit written instructions to guide them to ensure consistency (Appendix 4). The number of papers rejected at this stage and the reason for rejection are shown in Table 2

Table 2: Methodological criteria for rejecting papers
Reason for rejecting article	Number rejected	% of total sifted (683* papers)
Less than 80% follow-up	27	4
Group sizes smaller than 20	136	20
No control group	22	3
No method section (reviews)	82	12
Other design flaws	30	4
Other e.g. length of follow-up, population poorly sampled, etc	5	1
Total rejected on methodology	302	44
* 683 papers were identified as relevant to the task - a further 274 had been collected, but were not used because the areas were covered by good meta-analyses or because, after group discussion, the areas were felt to be outside the scope of the guideline.

Of the 683 papers 286 were accepted after the methodological sift. Certain methodological criteria were not regarded as absolute methodological flaws but were of sufficient importance to merit mention in the text. The commonest of these was an experimental study at risk of a Type II error. This means that the study had a negative result (it concluded there was no difference between groups) but it did not provide any evidence that the study had sufficient power to detect a difference were one there. This is referred to in the guideline as either "at risk of a Type II error" or "a negative study without a power calculation."

Synthesising the literature

Once individual papers had been checked for methodology and clinical significance, the information had to be synthesised. Each paper was categorised according to its study design reflecting susceptibility to bias. If a question could be answered by level I evidence provided by a good systematic review or randomised controlled trial then studies of weaker design were ignored. The evidence categories were adapted from the Canadian Task Force Classification (1979). Three categories were used:

I: based on well-designed randomised controlled trials, meta-analyses, or systematic reviews

II: based on well-designed cohort or case control studies

III: based on uncontrolled studies or consensus.

In the early meetings tables summarising the evidence were circulated to the group members before meetings. Initially the evidence was discussed and synthesised into recommendations by the group within meetings. In later meetings this process was conducted by the reviewer and the research team outside of group meetings and provisional recommendations were made. These were circulated for comment as a postal exercise to ensure time for finalising recommendations within the meetings. The evidence was synthesised using qualitative methods. These involved summarising the content of identified papers into brief statements that the group felt accurately reflected relevant evidence. Qualitative (meta-analysis) techniques were not used as we were dealing with studies other than randomised controlled trilas. Recommendations were derived using informal consensus methods. While it is inevitable that the process of interpreting evidence involves value judgements, by making this process explicit we made the scientific basis of these judgements as clear as possible. The relationship between strength of recommendation and category of evidence is shown below.

A: directly based on category I evidence

B: directly based on category II evidence

or extrapolated recommendation from category I evidence

C: directly based on category III evidence

or extrapolated recommendation from category I or II evidence

The strength of recommendation was shown as A, B or C after each recommendation.

Areas without evidence

It was accepted that there would be areas without literature evidence where recommendations had to be made and that consensus would be required to deal with these areas.

External review of the guideline

Dr R Baker, Director, Eli Lilly National Clinical Audit Centre, University of Leicester

Dr TA Carney, General Practitioner, Hexham, Northumberland

Professor S Cobbe, Walton Professor of Medical Cardiology, University of Glasgow

Dr A Farmer, General Practitioner, Thame, Oxfordshire

Dr G Feder, Senior Lecturer, Academic Department of General Practice and Primary Care, Medical Colleges of St Bartholomew's and The London Hospitals

Professor KAA Fox, Duke of Edinburgh Professor of Cardiology, University of Edinburgh

Dr JK Inman, General Practitioner, Syston, Leicester

Professor DG Julian, Emeritus Professor of Cardiology, University of Newcastle upon Tyne

Dr D Reid, Consultant Cardiologist, Freeman Hospital, Newcastle upon Tyne.

Their comments influenced the style and content of the guideline; however, the guideline remains the responsibility of the development group.

Piloting/pre-testing of the guideline

Scheduled review of the guideline

References

Canadian Task Force (1979) Canadian Medical Association Journal 121:1193-1254.

Guyatt, G.H., Sackett, D.L. and Cook, D.J. (1994) Users' guide to the medical literature II How to use an article about therapy or prevention. B. What were the results and will they help me in caring for my patients? Journal of the American Medical Association 271:59-63.

Jaeschke, R., Guyatt, G. and Sackett, D.L. (1994a) Users' guide to the medical literature III: How to use and article about a diagnostic test. A. Are the results of the study valid? Journal of the American Medical Association 271:389-391.

Jaeschke, R., Guyatt, G.H. and Sackett, D.L. (1994b) Users' guide to the medical literature III: How to use an article about a diagnostic test. B What are the results and will they help me in caring for my patients? Journal of the American Medical Association 271:703-705.

Kleijnen, J., Knipschild, P. and ter Riet, G. (1991a) Clinical trials of homeopathy. British Medical Journal 302:316-323.

Kleijnen, J., ter Riet, G. and Knipschild, P. (1991b) Acupuncture and asthma: a review of controlled trials. Thorax 46:799-802.

Koes, B.W., Bouter, L.M., Beckerman, H., van der Heijden, G.J.M.G. and Knipschild, P.G. (1991) Physiotherapy exercises and back pain: a blinded review. British Medical Journal 302:1572-1576.

Russell, I., Grimshaw, J. and Wilson, B. (1993) Scientific and methodological issues in quality assurance. Proceedings of the Royal Society of Edinburgh 101B:77-103.

Sackett, D.L. (1986) Rules of evidence and clinical recommendations on the use of antithrombotic agents. Chest 89(2):2S-3S.

Sackett, D.L., Haynes, R.B., Guyatt, G.H. and Tugwell, P. (1991) Surveying the medical literature to keep up to date. In: Clinical epidemiology: a basic science for clinical medicine. London: Little, Brown & Company, p. 359-378.

ter Riet, G., Kleijnen, J. and Knipschild, P. (1990a) Acupuncture and chronic pain: a criteria based meta-analysis. Journal of Clinical Epidemiology 43:1191-1199.

ter Riet, G., Kleijnen, J. and Knipschild, P. (1990b) A meta-analysis of studies into the effect of acupuncture on addiction. British Journal of General Practice 40:379-382.

West, R. (1992) Assessment of evidence versus consensus or prejudice. Journal of Epidemiology and Community Health 46:321-322.