Family Practice Advance Access originally published online on July 27, 2006
Family Practice 2006 23(5):520-528; doi:10.1093/fampra/cml038
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Systematic review: prognosis of angina in primary care
a Royal Free University College Medical School, Department Of Primary Care and Population Sciences London, UK
b Royal Free and University College Medical School, Medical Library London, UK
c Barts and the London, Queen Mary, University of London Centre for Health Sciences, London, UK
Correspondence to Melvyn Jones, Royal Free and University College Medical School, Department of Primary Care and Population Sciences, Rowland Hill Street, London NW3 2PF, UK; Email: m.jones{at}pcps.ucl.ac.uk
Received 14 June 2005; Accepted 23 May 2006.
 |
Abstract |
|---|
 Top Abstract BackgroundOutcomes of interestResultsDiscussionREFERENCES |
|---|
Background. Angina is a common chronic condition, largely managed in primary care in the UK. Mortality data are predominately from population or hospital studies with little known about the prognosis of angina in general practice settings.
Objective. To describe the prognosis of angina in patients identified in primary care.
Methods
Design. Systematic review of cohort studies of angina in primary care.
Data sources. Medline, PsycINFO, EMBASE, CINAHL, HMIC, WOS, IBSS, UK National Research Register, notification via JISC, CHAIN.
Review methods
Selection criteria. Cohort studies of patients with angina, with >12 months of follow-up, recruited within primary care.
Validity assessment. Database searches and abstracts were reviewed independently by two authors. Papers were assessed according to criteria derived from the cohort methodological literature.
Data abstraction. Data were abstracted by two reviewers.
Data synthesis. Narrative summary. A quantitative synthesis was planned.
Main outcome measures. Total and cardiovascular death; non-fatal myocardial infarction (MI).
Results. Six studies fulfilled our selection criteria. The annual total mortality rate is 2.8–6.6%, an annual cardiovascular death rate of 1.4–6.5% and an annual non-fatal MI rate of 0.3–5.5%. A quantitative synthesis was not possible, because the studies were clinically heterogeneous.
Conclusions. The primary studies have value in determining the prognosis of patients with angina recruited in general practice; however, the studies are old, have small numbers of events and are clinically heterogeneous. The contemporary prognosis of angina in primary care remains a key question, and further research is, therefore, required to estimate the prognosis of angina in this setting and its determinants.
Keywords. Angina pectoris, primary health care, prognosis, systematic review.
|
Background |
|---|
|
TopAbstractBackgroundOutcomes of interestResultsDiscussionREFERENCES |
|---|
Angina pectoris is an important common condition with appreciable morbidity and mortality. Estimates of angina prevalence in UK are 4.8 and 3.4% for men and women, respectively, of all ages.1 The US prevalence is reported at 3.8%.2 The Quality and Outcome framework (QOF) monitoring system for the new UK GP contract recorded 174 000 incident cases of angina in England over 2 years, which gives an incidence rate of angina of 0.33% (0.17% pa).3
Community-based studies suggest that people with diagnosed angina have a better 5 year mortality than survivors of a myocardial infarction (MI; hazard ratios 3.5 and 6.8, respectively), compared with people without manifest ischaemic heart disease.4 The disease severity in terms of risk of mortality falls with years since diagnosis, with a hazard ratio of 3.5 in the first 5 years and 1.2 between 10 and 15 years.4 This finding is supported by the community-based Framingham study5 and from one primary care based study.6 Secondary prevention and revascularisation has improved prognosis and quality of life of people with angina.7,8 However, we know that patients with angina are infrequently referred from primary care to specialist services,9 although this is likely to change with new policy promoting referral of people with new onset angina to secondary care in the UK.10,11
There is a need for a systematic review of angina prognosis in primary care. Estimates of angina prognosis from community-based and hospital-based studies need to be complemented by studies based in primary care, where most patients labelled and diagnosed with angina are managed. A consequence of selective referral from primary to specialist care is that studies measuring prognosis of angina patients in hospital settings are unrepresentative of angina patients in the community. Patients included in this type of study will probably have more severe disease and a worse prognosis. They may also be referred in an early, acute stage, again when their prognosis may be worse, leading to a referral bias. At a population level, among those with angina symptoms, there is evidence that receiving a diagnosis of angina improves prognosis,12 so community-based studies may again appear to indicate a worse prognosis in comparison with a primary care cohort.
The aim of this review was to estimate the prognosis of people with a diagnosis of angina in primary care.
The selection criteria for studies were as follows:
Inclusion criteria: Cohort studies of patients with angina with follow-up of at least 12 months, based in primary care.
Exclusion criteria: Non-cohort study designs, studies recruiting patients with unstable angina and cohort studies recruited outside primary care, e.g. within the general population or within hospitals.
Language: There were no language restrictions.
Time limits: Studies were included if they were published between January 1968 and December 2004. Earlier descriptive or methodological studies that relate to the primary studies are included only where they add useful data.
|
Outcomes of interest |
|---|
|
TopAbstractBackgroundOutcomes of interestResultsDiscussionREFERENCES |
|---|
Non-fatal MI, fatal MI, cardiovascular deaths and total mortality.
Definition of primary care
Accessible, often first contact, health care, usually provided within the community, which is either comprehensive, coordinated care involving sustained relationship with patients, or undifferentiated by age, gender, disease or organ. This includes comprehensive, co-ordinated care to particular subsets of the population sometimes for a fixed period or care that focuses on sustaining health rather than treating illness.13
|
Search strategy
We used standardized techniques14 to identify studies of prognosis15 and primary care.16 The strategy was developed in Medline, using both MeSH (thesaurus) terms and text words, (Fig. 1) and translated for other databases (available from authors). We searched on the following bibliographic databases:
|
National Library of Medicine (NLM) using Medline.
PsycINFO (The American Psychology Association's database).
EMBASE (A bibliographic database produced by Elsevier Science B.V. accessing international literature on pharmacology and biomedicine).
CINAHL (Cumulative Index to Nursing & Allied Health Literature).
HMIC (Health Management Information Consortium).
ISI WOK (Web of Knowledge) conference proceedings.
IBSS (International Bibliography of the Social Sciences).
The UK National Research Register was also searched, using the MeSH term ‘angina pectoris’ (accessed 20 February 2004). ‘Grey’ literature was obtained by searching for theses and conference abstracts from HMIC and WOK. Additionally we notified the Joint Information Systems Committee (JISC; www.jisc.ac.uk) and CHAIN (Contact, Help, Advice, and Information Network for effective health care). JISC and CHAIN are informal email networks for evidence-based care (Fig. 1).17
Citation tracking
Relevant articles were forward tracked through the Web of Knowledge and any articles identified by this process were reviewed. Most studies have reported earlier cross-sectional studies. Data from all papers reporting one of the included studies were used. Author names from identified studies were used as search terms to identify other studies using Medline. We wrote to first or corresponding authors and sought unpublished data (Fig. 2).
|
All primary studies that were subsequently included were initially found in the search of the NLM database.
Validity assessment
Database searches and abstracts were reviewed independently by two authors (MJ and GR) for papers meeting the inclusion criteria. Full text articles were obtained for possible and probable papers. Where there was disagreement, papers were reviewed and agreement was reached. A third reviewer (GF) was available for adjudication.
Papers were assessed by two reviewers according to consensus quality criteria.14,18–20 Criteria suggested by Lupacis were modified with the MOOSE consensus statement.19 This statement has a wider scope and specifies search strategies and methods of capturing the grey literature. We used these criteria first as inclusion/exclusion criteria and second, if feasible, for sensitivity analyses. In this review all studies of angina in primary care used a clinical diagnosis of angina, which would not fit the criterion of ‘explicit reproducible diagnosis’, so this filter was discarded.
Data from included studies were then abstracted independently by two reviewers (MJ and SC).
Analysis
Data from included studies were abstracted and means of annual total mortality, cardiovascular mortality and non-fatal MI rates, and their confidence intervals were calculated, if not reported. Forest plots were then constructed for the data. In addition to the narrative analysis, we planned to use a random effects model to produce a pooled estimate of mean event rates. We abstracted data and examined in detail, the Methods section of the included reports to look at study methods and the selection criteria for those recruited to the studies to ascertain whether there was clinical heterogeneity between studies.
|
Results |
|---|
|
TopAbstractBackgroundOutcomes of interestResultsDiscussionREFERENCES |
|---|
We reviewed 3411 abstracts and 136 full papers. Five abstracts in languages other than English potentially fulfilled our inclusion criteria and were translated into English. They did not fulfil our inclusion criteria and were excluded. Our correspondence with authors added no further data to the published reports.
Six papers fulfilled all our quality criteria (Table 2). The most common reasons for exclusion of full papers, identified as potential primary studies from the abstracts, were: studies examining a broader range of CHD diagnoses (e.g. post MI, undifferentiated chest pain), based in secondary care, cross-sectional or community surveys. A few studies were excluded as they had <12 months follow-up (Table 3).
|
|
Outcome 1: total mortality
All-cause mortality rate varied from 2.8 to 6.6% per annum. The studies were found to exhibit considerable clinical heterogeneity; therefore, a pooled estimate would not be statistically appropriate (Fig. 3).
|
Outcome 2: cardiovascular mortality
The cardiovascular death rate varied from 1.4 to 6.5% per annum (Fig. 4).
|
Outcome 3: non-fatal MI
The non-fatal MI rate varied from 0.3 to 5.5% per annum. We have only reported data from four studies as the non-fatal MI data were not reported in two of the studies (Fig. 5).
|
Clinical heterogeneity
We identified clinical heterogeneity in these studies. There is variation in cohort identification and recruitment in the primary studies. Lambert's study was a non-randomized comparison of beta blocker usage in angina (but reported total events).21–24 Hobkirk's study has an age range with a young upper limit (59 years).25,26 As most deaths and events would be expected to occur in the older age group, this may explain the reduced event rates in this study. The Mozaffarian study recruited patients with recall of a CHD diagnosis, (i.e. no independent confirmation of the diagnosis).27–29 However, there is evidence that there is good agreement between patient recall and clinical diagnoses.30 The Mozaffarian study also followed patients up on the basis of their Seattle Angina Questionnaire (SAQ) angina score and not specifically an angina diagnosis (Table 4). This study has the largest sample in this review and is the most robust methodologically, but the participants are different from the UK studies. All had served in the US armed forces, 98% were male and 15% were African Americans. This group will, therefore, have a different risk of cardiac events in comparison with other studies. Overall there is marked clinical heterogeneity between studies rendering a pooled estimate inappropriate.
|
|
Discussion |
|---|
|
TopAbstractBackgroundOutcomes of interestResultsDiscussionREFERENCES |
|---|
The main findings of this systematic review are the morbidity and mortality rates; the range of total mortality rate in angina patients identified in primary care is 2.8–6.6% per annum, cardiovascular death rates ranged from 1.4 to 6.5% per annum and the non-fatal MI rate ranged from 0.3 to 5.5% per annum.
There are some limitations in the primary studies, which may have an impact on the outcome rates we have reported. There is a wide range of percentage cardiovascular deaths (32–98%), which suggests there have been problems in recording this data. In Clarke's study,9 approximately one-third of patients had an unknown cause of death. It is likely that a high proportion of these deaths were cardiac in origin, and so this might explain the low percentage of cardiovascular deaths (32%) in this study compared with the other included studies.
The reports of percentage non-fatal MI rates show an almost 20-fold difference from the highest to lowest estimate. The lower estimate comes from Fry who collected data from patients starting in the 1950s.6 Fry's study may give a low estimate, due to under-recognition or under-recording of non-fatal MI as a cause. Historically, MI was largely a clinical diagnosis and patients were infrequently hospitalized with suspected MI. If we exclude Fry's non-fatal MI data the estimate range is a slightly narrower 1.8–5.5% pa.
We have presented the data in an approximate chronological order, but this is not the strict publishing date order as the data collection covered large, and in some cases overlapping, periods of time. This is particularly the case for Fry's work, which recruited patients from the 1950s onwards, but was published in the mid 1970s.
MI before or after the diagnosis of angina, is highlighted by other investigators as an important prognostic factor.31 Only one study reported this association with a univariate odds ratio of 1 year morality of 1.4.29 Hobkirk's study reports data suggesting an odds ratio of subsequent cardiac events (cardiac death and non-fatal MI) of 1.2 in relation to previous MI.25 Diabetes is associated with an increased death rate in the Mozaffarian study (OR 1.52). Hobkirk indicates a 33% mortality over 4 years amongst the patients with diabetes suggesting a 2-fold increase over their total published mortality rate. Other known important prognostic factors such as heart failure, medication and smoking status are similarly poorly reported apart from the Mozaffarian study (so are not reported again in this review).27
Age, probably the most important predictor of outcome, is reported in three studies but using different analyses and differing age bands, making comparisons difficult (see Table 4). Overall these studies report an increasing death rate with age, but, interestingly, Fry reports a falling observed/expected death rate in older patients. This might suggest that the prognosis of angina for older patients is more to do with general age-related processes rather than their underlying coronary artery disease.
There are potential problems with systematic reviews of observational data such as confounding and selection bias, which can distort the findings. There are no validated quality criteria for such studies, but there are consensus statements, which we have used as study ‘quality filters’.14,18,19 The possibility of publication bias must also be considered.
For such an important condition with an enormous impact on individuals and health services, there are surprisingly few studies of prognosis from primary care populations. Those that do exist are of a lower quality and are relatively small. Most of the UK work is no longer contemporary and predates recent medical and surgical advances that have improved prognosis.
Relationship to other literature
The range in prognosis in our primary studies encompasses estimates from studies investigating angina prognosis in other settings or using other methodologies. For example, the Framingham study in the US, a population cohort, reported a 30% mortality rate over 8 years in those >55 with angina. This approximates to a death rate of 3.8% per annum. The study is now quite old and does report a higher initial death rate, falling with years from diagnosis, so such estimates of average death rates must be treated with caution.31
Tiernay's retrospective US cohort study patients with IHD, which was defined as MI, angina or coronary artery disease, but which was based in primary care found an 18% mortality over 5 years (which approximates to 3.6% pa).32 A UK study of fish oil exposure for male patients with angina identified by their GP and self-selected (and therefore excluded from this review) gave an approximate death rate of 3.7% pa.33
In the UK, the British Regional Heart Study group reported a total mortality for diagnosed angina of 3.0% per year.4 This was a population cohort of middle-aged men who were screened for cardiovascular symptoms and risk factors, including a doctor diagnosis of angina. A Swedish community study of angina reports 148 deaths over 16 years among 314 men with angina (but no pre-existing MI), which approximates to 3.0% deaths pa.34 A recent community study from Finland of patients with angina, defined as those on nitrates or who are ‘test positive’ (ECG or angiogram) reports data, which would suggest a lower death rate of those with angina of 1.6% pa.35
Not all other studies reported mortality in our range. For example, Gandhi's study of patients with angina recruited from primary care but required referral to a quasi-secondary care clinic, similar in nature to contemporary rapid-access chest-pain clinics (hence its non-inclusion in this review). They reported a 7% non-fatal MI rate per year and a 4% total death rate per year.36 These higher rates in comparison with our results would suggest a referral bias in favour of the sicker patient. However, two recent studies looking at angina prognosis in secondary care report quite low death rates of 1.1–1.67% pa (and 1.8% in angiographically confirmed angina).37,38
The implication for practice is that angina remains a disease with appreciable mortality and morbidity. The implication for research is that a contemporary cohort study of angina is required to examine the impact of changes in management (use of statins, anti-platelet agents and coronary artery procedures) away from tightly controlled trial environments. It is vital this occurs in a primary care setting, where most patients with angina receive their care.
|
Acknowledgments |
|---|
ITA was partially funded by the British Heart Foundation Junior Research Fellowship (FS/03/011/15132). Sarah Cotter PhD who gave statistical advice for an earlier version of this paper died in 2005, we acknowledge her contribution and miss her.
|
Notes |
|---|
Jones M, Rait G, Falconer J and Feder G. Systematic review: prognosis of angina in primary care. Family Practice 2006; 23: 520–528
|
REFERENCES |
|---|
|
TopAbstractBackgroundOutcomes of interestResultsDiscussionREFERENCES |
|---|
1 Health survey for England 2003. Joint Health Surveys Unit, editor. (2004) (The Stationery Office, London).
2 Heart Disease and Stroke Statistics. (2006) American Heart Association American Stroke Association.
3 The Health and Social Care Information Centre (NHS). (2006) Quality and Outcomes Framework (QOF) for April 2004–March 2005, England Numbers on QOF disease registers, and unadjusted prevalence rates, by Strategic Health Authority with national summary; 2006. Available at: http://www.icservices.nhs.uk/qofdocuments/QOF0405_SHAs_Prevalence.xls.
4 Lampe FC, Whincup PH, Wanamethee SG, Shaper AG, Walker M, Ebrahim S. (2000) The natural history of prevalent ischaemic heart disease in middle aged men. Eur Heart J 21:1052–1062.
5 Kannel WB and Sorlie PD. (1978) Remission of clinical angina pectoris: the Framingham study. Am J Cardiol 42:119–123.[CrossRef][ISI][Medline]
6 Fry J. (1976) The natural history of angina in a general practice. J R Coll Gen Pract 26:643–646.[Medline]
7 Scandinavian Simvastatin Survival Study group (4S). (1994) Randomised trial of cholesterol lowering in 4444 patients with coronary heart disease. Lancet 344:1383–1389.[CrossRef][ISI][Medline]
8 Bucher HC, Hengstler P, Schindler C, Guyatt GH. (2000) Percutaneous transluminal coronary angioplasty versus medical treatment for non-acute coronary heart disease: meta-analysis of randomised controlled trials. BMJ 321:73–77.
9 Clarke KW, Gray D, Hampton JR. (1994) Implication of prescriptions for nitrates: 7 year follow up of patients treated for angina in general practice. Br Heart J 71:38–40.
10 New GMS contract. (2003) BMA/NHS Confederation.
11 National Service Framework for Coronary Heart Disease (Modern Standards & Service Models). (2000) London: Department of Health.
12 Hemingway H, Shipley M, Britton A, Page M, Macfarlane P, Marmot M. (2003) Prognosis of angina with and without a diagnosis: 11 year follow up in the Whitehall II prospective cohort study. BMJ 327:895.
13 Eldridge SM, Ashby D, Feder GS, Rudnicka AR, Ukoumunne OC. (2004) Lessons for cluster randomized trials in the 21st century: a systematic review of trials in primary care. Clin Trials 1:80–90.
14 Egger M, Davey Smith G, Altman DG. (2001) Systematic reviews in health care: meta-analysis in context. 2nd edn (BMJ, London).
15 McKibbon K, Eady A, Marks S. (1999) Evidence-Based Principles and Practice (Dekker, Hamilton, Canada).
16 Rosser WW, Starkey C, Shaughnessy R. (2000) The pond is wider than you think! Problems encountered when searching family practice literature. Can Fam Physician 46:103–108.
17 Russell J, Greenhalgh T, Boynton P, Rigby M. (2004) Soft networks for bridging the gap between research and practice: illuminative evaluation of CHAIN. BMJ 328:1174.
18 Laupacis A, Wells G, Richardson S, Tugwell P. (1994) Users' guides to the medical literature-V. How to use articles about prognosis. J Am Med Assoc 272:234–237.[CrossRef][ISI][Medline]
19 Stroup DF, Berlin JA, Morton SC, et al. (2000) Meta-analysis of Observational Studies in Epidemiology: A Proposal for Reporting. JAMA 283:2008–2012.
20 Elphick HE, Tan A, Ashby D, Smyth D. (2002) Systematic reviews and lifelong diseases. BMJ 325:381–384.
21 Lambert DMD. (1976) Effect of propranolol on mortality in patients with angina. Postgraduate. Med J 52:(suppl 4): 57–60.
22 Lambert DMD. (1972) Beta blockers and life expectancy in ischaemic heart disease. Lancet 1:793–794.[ISI][Medline]
23 Lambert DMD. (1977) Long term effects of propranolol on morbidity and mortality in patients with angina pectoris. Cardiovasc Med pp. 1977; 253–260.
24 Lambert DMD. (1978) Protection from morbidity and mortality from first infarcts. Aust Fam Physician pp. 1978; Special Issue June: 5–10.
25 Hobkirk DW. (1985) Follow up study of angina in patients aged 30 to 59 in general practice. Br Med J (Clin Res Ed) 290:1477–1479.
26 Research Committee Northern Regional Faculty Royal College of General Practitioners. (1982) Study of angina in patients aged 30 to 59 in general practice. BMJ 285:1319–1321.[ISI][Medline]
27 Mozzaffarian D, Bryson CL, Spertus JA, McDonell MB, Fihn SD. (2003) Anginal symptoms consistently predict total mortality among out patients with coronary artery disease. Am Heart J 146:1015–1022.[CrossRef][ISI][Medline]
28 Fihn SD, Burman M, McDonell MB, Henikoff J. (2002) Sub optimal management of chronic stable angina in a primary care setting. J Gen Intern Med 17:(suppl. 1): 192.
29 Spertus JA, Jones P, McDonell M, Fan V, Fihn SD. (2002) Health status predicts long-term outcome in outpatients with coronary disease. Circulation 106:43–49.
30 Lampe FC, Walker M, Lennon LT, Whincup PH, Ebrahim S. (1999) Validity of a self reported history of Doctor diagnosed angina. J Clin Epidemiol 52:73–81.[CrossRef][ISI][Medline]
31 Kannel WB and Feinleib M. (1972) Natural history of angina pectoris in the Framingham Study: prognosis and survival. Am J Cardiol 29:154–161.[CrossRef][ISI][Medline]
32 Tierney WM, Takesue BY, Vargo DL, Zhou XH. (1996) Using electronic medical records to predict mortality in primary care patients with heart disease: prognostic power and pathophysiologic implications. J Gen Intern Med 11:83–91.[ISI][Medline]
33 Burr ML, Ashfield Watt PAL, Dunsatn FDJ, Fehilly AM, Breay P. (2003) Lack of benefit of dietary advice to men with angina: results of a controlled trial. Eur J Clin Nutr 57:193–200.[CrossRef][ISI][Medline]
34 Rosengren A, Wilhelmsen L, Hagman M, Wedel H. (1998) Natural history of mycardial infarction and angina pectoris in a general population sample of middle aged men: a 16 year follow up of the primary prevention study, Goteborg, Sweden. J Int Med 244:495–505.[CrossRef][ISI][Medline]
35 Hemingway H, McCallum A, Shipley M, Manderbacka K, Martikainen P, Keskimaki I. (2006) Incidence and prognostic implications of stable angina pectoris among women and men. JAMA 295:1404–1411.
36 Gandhi MM, Lampe FC, Wood DA. (1995) Incidence, clinical characteristics and short term prognosis of angina pectoris. Br Heart J 73:193–198.
37 Daly CA, De Stavola B, Sendon JLL, et al. (2006) Predicting prognosis in stable angina-results from the Euro heart survey of stable angina: prospective observational study. BMJ 332:262–267.
38 Hjemdahl P, Eriksson SV, Held C, Forslund L, Nasman P, Rehnqvist N. (2006) Favourable long term prognosis in stable angina pectoris: an extended follow up of the angina prognosis study in Stockholm (APSIS). Heart 92:177–182.
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||