Authors: Richard A. Shellenberger, Stephanie Crass, Justin Jevicks, Ankita Badhwar, Jeremy Albright, Agara Kumar
Categories: Systematic Review/Meta-analysis
Source: CJC Open
Patients with obstruction to the left ventricular outflow tract from degenerative aortic stenosis (AS) usually do not become symptomatic until their disease becomes graded as severe. We sought to assess the accuracy of the physical examination for the diagnosis of AS of at least moderate severity.
A systematic review and meta-analysis of case series and cohorts of patients who received a cardiovascular physical examination prior to receiving a left heart catheterization or an echocardiogram. PubMed, Ovid MEDLINE, the Cochrane Library, ClinicalTrials.gov, and Embase were searched with a date range from inception through December 10, 2021, without language restrictions.
Our systematic review yielded 7 observational studies with adequate data to perform a meta-analysis on 3 physical examination assessments. Auscultating a diminished second heart sound (likelihood ratio [LR] = 10.87, 95% confidence interval [CI], 3.94-30.12, P < 0.05) and palpating a delayed carotid upstroke (LR = 9.04, 95% CI, 3.12-25.44, P < 0.05) are useful for detecting AS of at least moderate severity. The absence of a systolic murmur radiating to the neck (LR = 0.11, 95% CI, 0.06-0.23, P < 0.05) rules against AS of at least moderate severity.
Low-quality evidence from observational studies supports a diminished second heart sound and a delayed carotid upstroke as having moderate accuracy in diagnosing the presence of AS of at least moderate severity, whereas the absence of a murmur radiating to the neck is equally accurate in excluding this diagnosis.
Degenerative valvular heart diseases, such as aortic stenosis (AS) and mitral regurgitation, are now much more common than rheumatic heart disease in many countries worldwide.^1^ As valvular degeneration of the aortic valve leads to stenosis, the progressive increase in the obstruction to the left ventricular outflow tract can result in a decline in cardiac output, impaired exercise tolerance, and congestive heart failure.^2^ AS is graded from mild to moderate, and finally to severe, by this progression of the degree of valvular obstruction.^3^ The importance of clinical symptoms is paramount, as they are the strongest marker of mortality in patients with AS, which has a 50% two-year mortality without valve replacement with the development of syncope, chest pain, or congestive heart failure.^4^^,^^5^ Asymptomatic patients with severe AS also have been shown to have an increased risk of sudden death.6, 7, 8, 9 Unfortunately, the AS in many of these asymptomatic patients remains undetected, with most patients with AS remaining undiagnosed until they present with symptoms.^10^
The prevalence of AS increases steeply with advancing years of life, with a prevalence of only 0.2% in those aged 50-59 years, in contrast with the 9.8% prevalence seen in those aged over 80 years.^2^ In the US, approximately 1.5 million people have been diagnosed with AS, of which 500,000 have severe AS.^11^ Unfortunately, only one-half of these severe AS cases are noted to be symptomatic.^11^ The natural history of AS is one in which patients are generally unaware of their condition until they develop symptoms, and this development portends a 4-fold increase in mortality.^12^ Mounting evidence shows the clinical benefit of early intervention in asymptomatic severe AS, which emphasizes the importance of early and accurate detection.13, 14, 15 Physical examination findings have a long history in the diagnosis of AS, with the characteristic murmur first being described by James Hope in 1832.^16^ Evaluation of the precordial impulse, cardiac auscultation, and evaluation of the characteristics of the carotid impulse are longstanding approaches for diagnosing AS using physical exam. The rapid technological advances in the treatment of AS, combined with the curative nature of an aortic valve replacement, demonstrate the importance of early and accurate detection of AS. Given the lack of evidence to support any benefit from medical treatment to alter the progression of AS, an approach to identify patients in need of a valve replacement could improve outcomes as well as reduce costs. We performed our systematic review and meta-analysis to identify the most accurate physical examination findings for detection of AS of at least moderate severity.
We have registered our systematic review and meta-analysis through the International Prospective Register of Systematic Reviews (PROSPERO; CRD42022312606). Our study is reported in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) guideline.^17^
Our search strategies were established by using medical subject heading (MESH) terms and text words, and search terms that were selected based on common indexing practices and tested repeatedly to capture all potentially relevant studies. PubMed, Ovid MEDLINE, the Cochrane Library, ClinicalTrials.gov, and Embase were searched by 2 authors (S.C. and R.S.) for a date range of inception through March 10, 2022, without language restrictions. Additional studies were found by author and reference tracking.
After removing duplicates, 2 investigators (S.C. and R.S.) screened titles and abstracts for eligibility criteria. Full-text reviews were performed on studies that contained data on patients having a cardiac physical examination prior to having an echocardiogram or a left heart catheterization as the reference standard. Consensus was achieved for all phases of study selection, and any disagreements were reconciled by discussion.
Inclusion criteria were prospective or retrospective observational studies in which patients received a cardiac physical examination prior to receiving diagnostic confirmation by using a reference standard imaging test. We accepted an echocardiogram, as well as a left heart catheterization. as a gold-standard diagnostic imaging test for AS. Exclusion criteria included the no record of physical examination findings that were blinded to the results of the reference standard confirmatory diagnostic test; no confirmation of the physical examination with a reference standard; and no report of the number of patients in the study group or report of only percentages or likelihood ratios (LRs).
Studies that met inclusion criteria were then selected for data extraction and statistical analysis. Study characteristics were identified and included the study description; patient demographics; percentage of patients with moderate and severe AS; physical examination elements studied; and details of the reference standard confirmatory diagnostic test (Table 1). Two investigators independently assessed the quality and risk of bias of all included studies using the Newcastle-Ottawa Scale (NOS) for assessing the quality of nonramdomised studies in meta-analyses available from the National Institutes of Health.^18^ The certainty of evidence was evaluated for each study by using the Grading of Recommendations, Assessment, Development, and Evaluation (GRADE) assessment.^19^
Meta-analyses were performed when sufficient studies were available to pool sensitivities, specificities, and diagnostic odds ratios for each of the different measures. Table 1 outlines the study characteristics of the included studies for the 4 physical examination findings for which we have collected data. Meta-analyses of studies reporting sensitivity and specificity of diagnostic modalities generally employ bivariate methods that model both simultaneously. We have included forest plots showing the values of these statistics for individual studies. The pooled results are illustrated using hierarchical summary receiver-operating characteristic curves (Supplemental Fig. S1). Diagnostic odds ratios (DORs) also are presented using traditional univariate meta-analysis methods. Either a random-effects or fixed-effects pooling method was employed, depending on the value of the I^2^ statistic. Outlier studies were deleted, and the analysis was revised when funnel plots suggested the potential for publication bias (Supplemental Fig. S2). Analysis was performed using the meta and mada packages in R, as well as the metandi module for Stata.20, 21, 22 The Institutional Review Board at Trinity Health Ann Arbor Hospital deemed this project to be exempt from ethical approval.
After screening 3822 titles and abstracts, 44 studies were selected for full text review (Fig. 1). Seven studies met the inclusion criteria and had data extracted for statistical analysis from a total of 868 patients (Table 1).23, 24, 25, 26, 27, 28, 29 We were able to collect data on 4 physical examination findings and we review each evaluation separately.
Figure 1 Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) flow diagram.
A total of 6 studies with 815 patients evaluated a delayed carotid upstroke to assist in the diagnosis of AS.23, 24, 25, 26, 27, 28 The summary of pooled meta-analysis data for the delayed carotid upstroke in regard to sensitivity, specificity, and LRs is reported in Table 2. Univariate analysis was conducted for DORs employing a random-effects pooling method, owing to the moderate heterogeneity of studies. The pooled odds ratio based on all 6 studies is large and 13.81 (95% confidence interval [CI], 6.85-27.83; Fig. 2). Two studies^25^^,^^27^ were identified as having particularly large odds ratios and may be introducing publication bias. The analysis was performed without the outliers, and the remaining 4 studies still produced a large and significant pooled DOR of 11.06 (95% CI, 5.62-21.75; Table 2). We identified only one study reporting interobserver reliability for this physical examination finding, which was a kappa score of 0.26.^24^
Figure 2 Diagnostic odds ratios (ORs) for physical examination findings in the diagnosis of moderate aortic stenosis meta-analysis. CI, confidence interval.
Six studies with 801 patients reported data on the diminished intensity of second heart sound (S2).23, 24, 25, 26, 27^,^^29^ Pooled sensitivities, specificities, and LRs are reported in Table 2. A moderate amount of heterogeneity among studies is also present; therefore, DORs were calculated using the random-effects pooling method. The pooled odds ratio is large and significant—16.39 (95% CI, 7.85- 34.20; Fig. 2). The funnel plot indicates the possibility of publication bias; therefore, the studies^25^^,^^27^ that generated the 2 largest odds ratios were excluded, and the pooled odds ratio for the remaining 3 studies was smaller but still significant (DOR = 11.86, 95% CI, 6.67-21.08). In searching for data on interobserver reliability for this physical sign (diminished S2), we found only one study that reported a kappa score of 0.54.^24^
We report data from 4 studies with 662 patients that evaluated patients for the presence of a systolic murmur radiating to the neck for the diagnosis of AS.23, 24, 25, 26 Once again, significant variability among studies exists. Unlike the prior 2 measures, whose specificities exceeded their sensitivities, murmur radiating to the neck is more sensitive and less specific (Table 2). The between-study heterogeneity was minimal, and therefore, the fixed-effects model was employed. The pooled odds ratio was large and significant (DOR = 23.03, 95% CI 12.08-43.92, P < 0.05; Fig. 2). The funnel plot did not reveal evidence of bias from large outliers; the study with low precision^25^ reported the smallest odds ratio. We identified only one study that reported interobserver reliability of this physical examination sign—a kappa score of 0.33.^24^
We found only 2 studies, with 216 patients, from which we extracted data for a late-peaking systolic ejection murmur in the diagnostic evaluation of AS.^26^^,^^27^ With only 2 studies available, no bivariate meta-analysis was performed.
The relevance of the bedside diagnosis of AS dates back to at least 1832.^16^ We chose to perform our systematic review and meta-analysis because of the importance of early detection of AS, in an attempt to improve the unfavourable natural history of this degenerative valvular heart disease. To our knowledge, we are the first to perform a systematic review with a meta-analysis on the accuracy of the physical examination for the diagnosis of AS. Our study has identified 3 physical examination signs that are particularly useful in providing accuracy in the clinical bedside examination for the detection of AS of moderate or greater severity. Two of the examination findings we studied have significantly positive LRs, and thus are accurate to help make a determination in the presence of AS. The diminished S2 was found to have an LR of 10.87 (95% CI, 3.94- 30.12, P < 0.05), and palpating a delayed carotid upstroke had an LR of 9.04 (95% CI, 3.123-25.435, P < 0.05; Table 2). One physical examination finding was very good at discriminating against the diagnosis of AS—a systolic murmur radiating to the neck, which had an LR of 0.11 (95% CI, 0.056-0.232, P < 0.05; Table 2). Thus, the lack of detection of a systolic murmur radiating to the neck is a very accurate aid in ruling out AS. These data provide moderate support for the accuracy of the physical examination to diagnose AS of at least moderate severity. These interpretations of the use of LRs are supported by McGee, who has explained that an LR of 10 increases by 45% the probability of a disease or condition being present when a test or sign is positive or present; thus, it has moderative positive predictive value.^30^^,^^31^ Conversely, an LR of 0.1 decreases by 45% the probability of a disease or condition being present in the absence of that test or sign, and thus, it has moderate negative predictive value.^30^^,^^31^
Our study was designed to identify the most accurate physical examination elements for evaluating patients with a systolic heart murmur suspected of having AS. We fully recognize that several other physical signs also should be used when evaluating patients with systolic murmurs. Several physical signs that historically have been studied in the diagnosis of AS were identified in our systematic review; however, we did not have enough data for these to be included in our meta-analysis. A palpable delay between the apical and carotid impulse had excellent negative predictive value (LR of 0.05) in only one study.^32^ The brachioradial delay has similarly excellent negative predictive value (LR of 0.04), also in only one study.^33^ Finding a sustained apical impulse has modest positive predictive value (LR of 4.1).^28^ A late-peaking murmur provides the best diagnostic evidence of any of the murmur characteristics that have been studied. Data from a small sample from 4 studies revealed modest positive predictive value when this sign is present (LR of 3.7), as well as modest negative predictive value of its absence (LR = 0.2).^23^^,^^25^^,^^26^^,^^34^ The fourth heart sound has not been shown to have significant diagnostic value for evaluating suspected AS.^25^^,^^35^ We believe our study has identified the most accurate physical examination findings that are useful to include in the assessment of patients with systolic heart murmurs. Our systematic review and meta-analysis introduces data that, to our knowledge, have not been presented elsewhere.The clinical implications of an accurate physical examination being used to enhance the early diagnosis of AS include a more expedient implementation of aortic valve replacement, which may decrease both disease burden and associated costs, and most importantly, lead to lower morbidity and mortality.13, 14, 15^,^^36^
Our study has several limitations. We only identified 8 studies that met our eligibility criteria. In 5 of the studies, the sample size was less than 100, and in the other 3, the largest sample size was 231 (Table 1). Risk of bias was moderate to high in all the selected studies (Table 3). Selection bias was high, as none of the studies had a control group for comparison, and most of the case series or cohorts were comprised of patients with known AS. Independence between the findings of the physical examination and the results of the reference standard test occurred in 4 of the 8 studies, which indicates the possibility of bias in the reporting of outcomes.^23^^,^^24^^,^^26^^,^^27^ Only one study we identified reported on the interobserver reliability for the 3 physical examination findings in our meta-analysis. We have identified age as a possible confounder for the physical finding of a delayed carotid upstroke, as the transmission velocity of the carotid pulse has been found to be higher in older, compared with younger, study subjects.^37^ Given that the prevalence of AS increases with increasing age, studies with age-matched control patients without AS could shed more light on the effects of age on the carotid upstroke. Moderate heterogeneity was present in the meta-analyses of the delayed carotid upstroke and the diminished intensity of S2, which means the results of the studies for these findings had a moderate degree of variability. The meta-analysis for murmur radiating to the neck had low heterogeneity and was the best estimate of a true single effect size in our study. And finally, due to the fact that all our studies are observational in nature, the quality of evidence using the Grading of Recommendations, Assessment, Development, and Evaluation (GRADE) assessment is low.^19^
In conclusion, in our review of 8 studies, which included a total of 951 patients, we found low-quality evidence for both the positive and negative predictive value of physical examination findings for the presence and absence of AS of at least moderate severity. Presence of a diminished S2 and a delayed carotid upstroke, when systolic heart murmurs are evaluated, indicate a modest increase in the pretest probability that a patient has moderate or greater severity AS. The absence of a murmur radiating to the neck provides modest improvement in the posttest probability that a patient does not have at least moderate AS. Further high-quality studies with control patients in carefully matched cohorts are necessary to shed more light on the accuracy of the physical examination in enhancing the early detection of AS.
The Institutional Review Board at Trinity Health Ann Arbor Hospital deemed this project to be exempt from ethical approval.
The authors have no funding sources to declare.
The authors have no conflicts of interest to disclose.