|
Systematic review and individual patient data meta-analysis of diagnosis of heart failure, with modelling of implications of different diagnostic strategies in primary care |
Mant J, Doust J, Roalfe A, Barton P, Cowie MR, Glasziou P, Mant D, McManus RJ, Holder R, Deeks J, Fletcher K, Qume M, Sohanpal S, Sanders S, Hobbs FD |
|
|
CRD summary The authors concluded that heart failure signs and symptoms were of limited use in a primary care setting. Plasma B-type natriuretic peptides, including N-terminal pro-B type natriuretic peptides, and electrocardiogram were useful for ruling out heart failure and chest X-rays could aid a positive diagnosis. The authors' conclusions should be interpreted with caution given the possibility of inappropriate statistical analyses. Authors' objectives To assess the accuracy of using clinical features and potential primary care investigations to diagnose heart failure. Searching MEDLINE and CINAHL were searched, without language restriction, for articles from inception to July 2006. Search terms were reported and reference lists of included studies and of review articles were searched. Grey literature databases were searched, as were the conference proceedings of the American College of Cardiology, American Heart Association, European Society of Cardiology (ESC), British Cardiac Society, Heart Failure Society of America, Royal College of Physicians, International Academy of Cardiology, International Heart Failure Society, and the Cardiac Society of Australia and New Zealand. Study selection Eligible for inclusion in the review were studies that compared clinical symptoms or signs, electrocardiogram (ECG), chest X-ray, or plasma or B-type natriuretic peptides (BNPs), which included N terminal pro BNP (NT-proBNP), with a reference standard. Reference standards were a clinical diagnosis, using for example ESC criteria, or echocardiographic diagnosis, using criteria for left ventricular systolic dysfunction or for heart failure with preserved systolic dysfunction. Studies were excluded if they included children; used an inappropriate index test (e.g. urinary natriuretic peptides) or reference standard (e.g. diastolic function alone); used a retrospective or case-control design; or did not provide results that could be extracted.
The clinical signs and symptoms included: a history of myocardial infarction, dyspnoea, orthopnoea, paroxysmal nocturnal dyspnoea, oedema, tachycardia, elevated jugular venous pressure, cardiomegaly, added heart sounds, lung crepitations, and hepatomegaly. Included studies were in primary care, emergency department, out-patient secondary care, and in-patient clinical settings or were screening studies.
Two reviewers selected studies for inclusion in the review and any disagreements were resolved by discussion with a third reviewer. Assessment of study quality Study quality was assessed using the Quality Assessment of Diagnostic Accuracy of Studies (QUADAS) criteria. The authors did not state how many reviewers performed this assessment. Data extraction Data were extracted to calculate the sensitivity and specificity, negative and positive predictive values, likelihood ratios, and their associated 95% confidence intervals. Authors of relevant studies were contacted for data that could not be extracted from the published papers.
Two reviewers extracted the data. Methods of synthesis Sensitivity and specificity were plotted on receiver operating characteristic (ROC) graphs. Sensitivity, specificity, diagnostic odds ratios, and positive and negative likelihood ratios were combined in a bivariate random-effects model. Tests for heterogeneity were not performed. A Youden index (percentage sensitivity plus percentage specificity minus 100), which is a measure of diagnostic accuracy, was calculated for each clinical symptom or sign and investigation. Results were grouped by index test (including type of assay) and reference standard, and by clinical setting.
The impact of clinical setting and that of the prevalence of heart failure in the study, on the results, were explored. Results of the review There were discrepancies in values between the abstract, the tables, and the main text of the review; the results in the main text of the review are reported here. Studies were of variable quality, but most of the criteria were met or were unclear.
Symptoms and Signs (15 studies): The Youden index ranged from 10 (added heart sounds) to 37 (dyspnoea). Dyspnoea (five studies) was the only symptom or sign with high sensitivity (83%, 95% CI 62 to 94), but it had poor specificity (54%, 95% CI 40 to 67). The remaining signs and symptoms (e.g. myocardial infarction) had relatively high specificity, but low sensitivity. There was considerable variation across studies, which was reported to be due to different definitions or ways of eliciting the symptoms or signs, or differences in the patient groups studied.
ECG (11 studies): Using broad criteria for abnormality, ECGs achieved a relatively high sensitivity (89%, 95% CI 77 to 95), but only moderate specificity (56%, 95% CI 46 to 66). They were either read by a cardiologist or automatically.
Chest X-ray (nine studies): Chest X-rays were insensitive for detecting any sign of heart failure (68%, 95% CI 40 to 88; five studies) and for detecting increased cardiothoracic ratio (67%, 95% CI 53 to 78; six studies), but they had moderate specificity (83%, 95% CI 66 to 93, for signs; 76%, 95% CI 65 to 84, for cardiothoracic ratio).
BNP (20 studies): Studies examining BNP showed consistently high sensitivity (93%, 95% CI 91 to 95), but varying specificity (74%; 95% CI 63 to 83). Four studies, in general practice, showed a slightly lower sensitivity, but similar specificity. Sixteen studies examining NT-proBNP showed high sensitivity (93%, 95% CI 88 to 96), but specificity varied and was lower than that of BNP (65%, 95% CI 56 to 74). Eight studies of NT-proBNP, in general practice, showed a slightly lower specificity. There was no statistically significant difference in the diagnostic accuracy of NT-proBNP compared with BNP (DOR 1.20, 95% CI 0.30 to 4.80; six studies).
BNP had greater diagnostic accuracy than ECG, with a relative diagnostic odds ratio for ECG to BNP of 0.32 (95% CI 0.12 to 0.87; four studies). There was no difference in diagnostic accuracy between NT-proBNP and ECG (seven studies).
Individual patient data analyses were undertaken, but the results are not reported as they did not assess the effectiveness of the diagnostic tools. Authors' conclusions Symptoms and signs of heart failure had varying specificity, but their poor sensitivity limited their use in ruling out disease in a general practice setting. BNP and ECG had relatively high sensitivity and were useful for ruling out heart failure. Chest X-ray had the highest specificity and could be used for a positive diagnosis of heart failure. CRD commentary The review addressed a clear research question and was supported by detailed inclusion criteria. The search strategy was appropriate, with no language restriction, and attempts were made to locate unpublished material, which reduced the risk of language and publication bias. The quality assessment tool was appropriate, but the authors did not state how many reviewers made this assessment, which means that reviewer error and bias cannot be ruled out. Adequate details of the primary studies were provided. There was considerable variation between studies in their patient characteristics, methods, and quality, which means that the pooling of their results might have been inappropriate. There were also discrepancies in the values between the abstract, the tables and the main text of the review, which makes it unclear which values were correct.
The authors' conclusions reflected the evidence presented, but should be interpreted with caution given the possibility of inappropriate statistical analyses. Implications of the review for practice and research The authors did not state any implications for practice and research, based only on the results of the systematic review. Funding NIHR Health Technology Assessment programme Bibliographic details Mant J, Doust J, Roalfe A, Barton P, Cowie MR, Glasziou P, Mant D, McManus RJ, Holder R, Deeks J, Fletcher K, Qume M, Sohanpal S, Sanders S, Hobbs FD. Systematic review and individual patient data meta-analysis of diagnosis of heart failure, with modelling of implications of different diagnostic strategies in primary care. Health Technology Assessment 2009; 13(32): 1-207 Indexing Status Subject indexing assigned by NLM MeSH Aged; Aged, 80 and over; Diagnosis, Differential; Female; Heart Failure /diagnosis /metabolism; Heart Function Tests /methods; Humans; Male; Middle Aged; Natriuretic Peptide, Brain /analysis; Practice Guidelines as Topic; Primary Health Care /methods; State Medicine AccessionNumber 12009107773 Date bibliographic record published 07/10/2009 Date abstract record published 21/07/2010 Record Status This is a critical abstract of a systematic review that meets the criteria for inclusion on DARE. Each critical abstract contains a brief summary of the review methods, results and conclusions followed by a detailed critical assessment on the reliability of the review and the conclusions drawn. |
|
|
|