|Home | About | Journals | Submit | Contact Us | Français|
Syncopal episodes are common among older adults; etiologies range from benign to life-threatening. We determined the frequency, yield, and costs of tests obtained to evaluate older persons with syncope. We also calculated the cost per test yield and determined whether the San Francisco Syncope Rule (SFSR) improved test yield.
Review of 2,106 consecutive patients 65 years and older admitted following a syncopal episode.
Electrocardiograms (99%), telemetry (95%), cardiac enzymes (95%), and head computed tomography (CT) (63%) were the most frequently obtained tests. Cardiac enzymes, CTs, echocardiograms, carotid ultrasounds, and electroencephalography all affected diagnosis or management in <5% of cases and helped determine etiology of syncope < 2% of the time. Postural blood pressure, performed in only 38% of episodes, had the highest yield with respect to affecting diagnosis (18-26%) or management (25-30%) and determining etiology of the syncopal episode (15-21%). The cost per test affecting diagnosis or management was highest for electroencephalography ($32,973), CT ($24,881), and cardiac enzymes ($22,397) and lowest for postural blood pressure ($17-$20). The yields and costs for cardiac tests were better among patients meeting, than not meeting, SFSR. For example, the cost per cardiac enzymes affecting diagnosis or management was $10,331 in those meeting, versus $111,518 in those not meeting, the SFSR.
Many unnecessary tests are obtained to evaluate syncope. Selecting tests based on history and examination and prioritizing less expensive and higher yield tests would ensure a more informed and cost-effective approach to evaluating older patients with syncope.
Syncope, defined as the sudden, transient loss of consciousness with spontaneous recovery, accounts for 1-3% of emergency department visits and up to 6% of hospital admissions.1,2 Older patients present more often with syncope, have higher rates of hospitalization, and greater morbidity, than younger patients.3 Evaluation of older patients following a syncopal episode is challenging because there is a wide spectrum of possible etiologies, ranging from benign to life-threatening conditions.4 Because of the varied causes of syncope, clinicians may pursue a range of diagnostic investigations. Despite thorough evaluations, however, the etiology of syncope frequently remains undetermined.5,6
Several authors have suggested schemes for evaluating syncope, based primarily on expert consensus rather than empirical evidence.7,8 Studies have attempted to reduce unnecessary testing by the use of algorithms to improve syncope evaluation methods. 9-13 The algorithms improved the percentage of patients in whom an etiology was identified but did not decrease the use of low yield testing or reduce the cost associated with diagnostic testing. The few studies examining the utility of individual tests found that cardiac enzymes,14,15 electroencephalography (EEG),16 head computed tomography (head CT)17 and carotid ultrasonography (carotid US)18 seldom identified the etiology. Neurological testing was less useful diagnostically than cardiac testing in one study.19 The contribution of magnetic resonance imaging (MRI), echocardiogram, telemetry, and other tests remains unknown as does cost of tests relative to their effect on diagnosis or management. Also unclear is whether the yield and cost-effectiveness of evaluation can be improved by identifying older adults presenting with syncope in whom test results are likely to affect diagnosis or management. Patient characteristics such as those used in The San Francisco Syncope Rule (SFSR) may serve this latter purpose.20 The SFSR was developed to improve prediction of the likelihood of serious outcomes in patients presenting with syncope and has been prospectively validated for this purpose.21
We determined how often diagnostic tests were obtained to evaluate older persons presenting with syncope and whether or not these tests helped establish the etiology of the syncopal episode or affected diagnosis or management. We also calculated the cost per test that affected diagnosis or management. Finally, we examined whether the San Francisco Syncope Rule (SFSR) was associated with the likelihood of test results affecting diagnosis or management.
The study included all patients 65 years and older admitted to an acute care hospital, after presentation to the emergency department, between July 1, 2002 and December 31, 2006 with an admission or discharge diagnosis of syncope. Patients were identified based on the presence of an International Classification of Diseases–Clinical Modification (ICD 9-CM) code of 780.2 as a primary or non-primary diagnosis in the hospital billing records. Up to 10 diagnoses are listed, enhancing the likelihood that patients with syncope were identified. Based on review of the medical records, all patients with presumed loss of consciousness (LOC) were included. Patients in whom absence of loss of consciousness (e.g. near syncope) was documented were excluded. The study was approved by the School of Medicine’s Institutional Review Board. Consent was not obtained from participants because we reviewed existing data; in accordance with federal guidelines no subject identifiers were included in the data collected.
Records from 2209 admissions in 2009 patients were reviewed. We excluded 103 admissions because of the complete absence of laboratory data, imaging, electronic medical record, or paper chart. Admissions were included if partial data were available; there were 2106 admissions included for 1920 patients.
For each admission, emergency department, inpatient admission, and progress notes; discharge summaries; and laboratory and imaging data were abstracted. We employed methods recommended to ensure the validity and reliability of data collected, including a standardized abstraction form, precisely defined variables and criteria, and a pilot study of 60 charts to refine criteria.22
Data collected included patient age and gender; dates of admission and discharge; whether presumed loss of consciousness was documented in the record and whether the episode was witnessed; symptoms and activity at the time of episode; health conditions; cardiac and neurological examination findings; postural blood pressure recordings; and cardiac enzymes. Reported etiology of the syncopal episodes was ascertained from the discharge summary. If no etiology was reported in the discharge summary, then progress note documentation was used. Results of electrocardiogram, echocardiogram, head computed tomography, carotid ultrasound, stress testing, head magnetic resonance imaging, and electroencephalography were abstracted from the test reports and progress notes.
A second reviewer blindly abstracted a random sample of 40 admissions. To measure interrater agreement, we used the prevalence-adjusted, bias-adjusted kappa statistic.23,24 The mean PABAK statistic was 87% (±20%) for the diagnostic test variables.
An abnormal finding for imaging was defined as any abnormality, no matter how minor, not seen on prior testing as written in the test reports (for example, mild mitral regurgitation on echocardiogram and mild slowing on electroencephalography). If no mention was made of prior testing, the result was assumed to be new. Abnormal cardiac enzyme results were defined as any troponin-I level > 0.05 (the hospital’s reference value). For postural blood pressure, recordings were documented based on position (supine to sitting, supine to standing, or supine to sitting to standing). Postural blood pressure was defined using two sets of criteria, “strict” and “loose”. The strict criteria for postural hypotension was a drop in systolic blood pressure ≥ 20 mmHg, or a drop in diastolic blood pressure ≥10mmHg from lying to standing positions.25 The loose criteria for postural hypotension was ≥10 mmHg drop in systolic or diastolic pressure or a systolic pressure drop to ≤ 90mm Hg from lying to sitting or standing positions. This definition incorporated the variability in methods used to assess blood pressure changes and the wide range of definitions used in the literature, particularly for older patients.26,27,28
A test result was considered to have affected diagnosis or management if it was noted in any of test reports, progress notes, or discharge summaries that the test contributed to, confirmed or established any diagnosis or management decisions. This definition included documentation of negative and positive test results and all diagnoses, including those not related to syncope. We also recorded whether it was documented any where in the medical record that a test result helped determine the etiology of the syncopal episode. Examples of a test affecting diagnosis included an electrocardiogram identifying atrial fibrillation and postural blood pressure recordings identifying postural hypotension. Examples of a test affecting management included an electrocardiogram resulting in the management of atrial fibrillation with anti-coagulation and beta-blockers and postural blood pressure recordings resulting in the management of postural hypotension with hydration. The criteria detailed above for test results were defined independently such that a test result could be abnormal but not considered to have affected diagnosis or management or vice versa.
Patients were considered to have met the SFSR criteria if they had history of congestive heart failure, hematocrit <30%, abnormal electrocardiogram, shortness of breath, or systolic blood pressure < 90mmHg at presentation.20
Standard billing charges for this hospital were used to calculate the charge per test. For imaging and electrocardiography, this included professional fees associated with interpretation. Similar to other studies, we converted billing charges to costs by multiplying charges by the hospital’s cost to charge ratio,29,30 because the hospital calculates costs based on a patient’s admission rather than on individual tests or services provided. A cost to charge ratio for a given hospital is determined by dividing the cost incurred to the hospital for an admission divided by the amount charged to an admitted patient. A cost to charge ratio of 0.34 was used based on this hospital’s cost to charge ratio from the State of Connecticut’s Annual Report on the Financial Status of Connecticut’s Acute Care Hospitals for Fiscal Year 2007. This ratio was similar to the cost to charge ratio for syncope patients admitted during the study period of 0.35. The cost for telemetry was estimated as the difference in cost between a monitored and unmonitored bed. For postural blood pressure, we estimated $5 per test, assuming that it required five minutes at a nurse’s wage of $60 per hour, the highest nursing salary on a medicine service. Nurses most often obtain postural blood pressure recordings at this hospital. We defined the cost per test affecting diagnosis or management as the cost per test multiplied by the number of tests obtained divided by the number of test results that affected diagnosis or management.
Statistical analysis was performed using SAS (version 9.1). Yields were reported as percentages. Denominators were the number of tests obtained and the numerators were the number of tests in which findings were abnormal, affected diagnosis, affected management, or helped determine the etiology of the syncopal episode. We stratified patients into those meeting or not meeting SFSR criteria to compare testing results and cost per test affecting diagnosis or management. The Fisher exact test was used to compare test results in patients meeting and patients not meeting the SFSR. A two-sided P value < 0.05 was used to indicate statistical significance.
Characteristics of the 1920 patients are presented in Table 1. The mean age was 79 (±7.9) years; 53% were female. One hundred and sixty-three patients (8.5%) had two or more admissions for syncope during the four and a half year period, resulting in 2106 admissions. The most common preexisting health conditions included hypertension (66%), hyperlipidemia (32%) and coronary artery disease (32%). The most commonly reported etiologies were vasovagal and orthostatic hypotension. For 47% of episodes, the etiology was reported as unknown or not reported in the records. Ninety percent of admissions were to a general internal medicine service, 6% to a cardiology service, and 3% to neurology. We did not find any significant differences in testing between patients admitted to the different services.
The frequencies of tests obtained, abnormal findings, and yields are shown in Table 2. The most frequently obtained tests were electrocardiogram (99% of admissions), telemetry (95%), and cardiac enzymes (95%). Only 5% of admissions had abnormal enzymes defined as any elevation in troponin-I. Echocardiogram (63%) had the highest frequency of abnormal findings; most of these were minor structural changes such as mild mitral regurgitation. Only 2% of echocardiograms revealed findings, most often aortic stenosis, reported to have contributed to the syncopal episode. Similarly, for electrocardiogram and telemetry most findings were minor such as premature ventricular contractions. Telemetry results helped determine the etiology, such as atrial fibrillation or bradycardia, for 5% of syncopal episodes.
Postural blood pressure was performed in 38% of patients; only 24% of patients had recordings obtained lying to standing. Postural blood pressure had the highest yield with respect to affecting diagnosis (18% using strict criteria) and management (25% using strict criteria), and was the test most frequently reported to have helped determine the etiology of the syncopal episode. The tests with the lowest likelihood of affecting diagnosis or management or determining the etiology of the syncopal episode were head CT, carotid ultrasound, EEG, and cardiac enzymes. In the eight out of nine admissions in which cardiac enzymes helped determine the etiology of syncope, abnormal electrocardiogram changes were also noted.
Head CT affected diagnosis or management in only 28 of 1327 (2%) admissions; 25 of these involved clinically suspected neurologic disease such as brain metastases, new neurological symptoms, or recent head trauma. Similarly, 17 of the 20 cases in which MRI result affected diagnosis or management were suspected based on history or examination.
The costs per test affecting diagnosis or management are shown in Table 3. This cost was highest for EEG ($32,973), head CT ($24,881), and cardiac enzymes ($22,397) and lowest for postural blood pressure ($17-$20). Examples of the cost per test that helped determine the etiology of syncope include $99,525 for head CT, $77,144 for cardiac enzymes, $65,946 for EEG and ($23- $33) for postural blood pressure.
As shown in Table 4, with the exception of cardiac stress testing, cardiac test results were much more likely to have affected diagnosis or management or helped determine the etiology of the syncopal episode in patients meeting the SFSR than in patients not meeting criteria. The costs per cardiac test affecting diagnosis or management also were much higher among patients not meeting the SFSR than patients meeting the SFSR. For example for cardiac enzymes, the cost per test affecting diagnosis or management was $10,331 in those meeting, versus $111,518 in those not meeting, the SFSR.
In this study, we found that cardiac and neurologic tests were commonly obtained in the evaluation of syncope in older patients despite minimal effect on diagnosis or management. In the few cases in which neurological tests were helpful, neurological conditions were suspected based on history or examination. Cardiac testing also had low yields overall. Conversely, postural blood pressure recordings had the highest yield, but were performed in only about a third of admissions and frequently performed inadequately. Application of the San Francisco Syncope Rule markedly improved yields and lowered costs without compromising identification of persons with life threatening cardiac conditions.
As in previous studies, vasovagal episodes and orthostatic hypotension were the most frequently reported etiologies. The lack of an etiology in almost half of patients despite extensive testing was also similar to prior reports of older adults.1,4 Algorithm-based evaluations, which were not used in the study hospital, result by design in higher percentages of patients with a diagnosed etiology. 9-13
The lowest likelihood of useful test results and, therefore, the highest costs per yield were incurred by EEG, head CT, and cardiac enzymes. Although only troponin-I was used to define abnormal results,31,32 the total cost per set of cardiac enzymes included creatine kinase and creatine kinase MB. If troponin-I alone was obtained the cost per cardiac enzymes affecting diagnosis and management would decrease from $22,397 to $4813. Postural blood pressure measurements represented the lowest cost per test affecting diagnosis and management at $17. This figure maybe lower or higher in actual practice based on who is performing the blood pressure measurements but the magnitude is likely accurate for comparison to other testing costs.
Many of our findings are consistent with prior studies. Grossman et al. and Link et al. found that serial cardiac enzymes had little impact on diagnosis in syncope.14,15 Head CT, carotid US, and EEG are all known to rarely identify lesions contributing to syncope.16-18 Our findings confirm these earlier findings that neurological imaging is not warranted in the evaluation of syncope unless a neurological disease or event is suspected. As has previously been shown, our results suggest that the etiology of syncope can often be determined solely by history 33 The high yield of postural blood pressure recordings in our study supports guideline recommendations that the initial evaluation of syncope should entail history, physical examination, electrocardiogram and postural blood pressure measurements. 3, 7, 8,34
Strengths of our study include a large sample size, standardized medical abstraction, consistent definitions and criteria, and blinded re-abstraction to ensure reliability. The evaluation of syncope entails identifying the presence of underlying diseases in addition to determining the etiology of the syncopal episode. Therefore, we complemented existing research by determining the affect the tests had on establishing any diagnosis or affecting any management decision.
There are limitations to our study. First, we report the retrospective experience of a single hospital – although comparison to previous studies suggests that this experience is representative of other hospitals. 3,14-19 Second, we only included patients who presented to the emergency department and were subsequently admitted, because these patients undergo more extensive testing compared to other settings. Third, because we used the ICD 9-CM code of 780.2 to identify syncope admissions it is possible that we may have missed some patients with syncope without this ICD 9-CM code. However, as we included patients with an admission or discharge, primary or non-primary diagnosis, the number of missed patients is likely small and should not affect our results. Fourth, all clinical decisions may not have been documented in the medical record. For example, we likely underestimated the contribution of negative results to diagnosis or management, as only 3% of test results that were reported to have affected diagnosis or management were negative results. Fifth, we did not evaluate tests performed after the hospitalization, such as loop recorders and tilt table testing, and commonly performed laboratory testing, such as hematocrit and glucose. Finally, our calculation of costs using a cost to charge ratio is an estimation based on charges- an approach to cost calculation used by other studies in the absence of hospital estimates of cost. The hospital’s cost to charge ratio utilized in the calculation was nearly identical to the cost to charge ratio of patients admitted with syncope, which supports the accuracy of our estimation. Our calculation of costs may have underestimated total costs because it does not include all tests and procedures performed nor the cost of hospitalizations, estimated to be between $7460 and $9950 per admission.
Perhaps the most concerning finding of this study is the extent to which unhelpful, and presumably unnecessary, testing in the evaluation of syncope continues despite the compelling evidence against the practice dating back 20 years.7-19 The current study complements earlier work by showing the high costs associated with this unnecessary testing. Extrapolating our results nationally, assuming approximately 460,000 hospitalizations per year for syncope,29 yearly costs associated with the most commonly obtained tests may be nearly 6 billion dollars. Investigators have shown that easy availability of low risk testing contributes to the overuse of resources.35 The frequency of syncope and wide availability of low risk testing makes its an important source of revenue for hospitals. Unnecessary testing is a significant contributor to rising healthcare costs and has been proposed as a target for cost savings.35,36
Our results suggest how clinicians might be more selective when obtaining tests to evaluate syncope. One goal of the evaluation of syncope is to detect conditions, particularly life-threatening ones such as arrhythmias, which may be present in patients with syncope. In this study we found the SFSR criteria to be helpful in identifying patients likely to benefit from cardiac testing. Validation studies of the SFSR have had conflicting results.21,37 Other criteria have been proposed to predict adverse outcomes38 and may be helpful as well in serving as predictors of diagnostic yield. Our results suggest that using patient characteristics such as the SFSR may help determine in whom certain tests, particularly cardiac enzymes, and perhaps telemetry, are indicated, resulting in a marked savings costs. Further research is needed to determine reliable and feasible criteria to screen patients presenting with syncope.
Because almost one quarter of older patients who experience a syncopal episode suffer serious injuries such as a hip fracture during the episode,39 another goal is to identify non-life threatening, but treatable, etiologies such as postural hypotension. Even if other etiologies are suspected, assessment and management of postural hypotension is warranted. Our study suggests that inexpensive postural blood pressure testing is greatly underutilized, resulting in many missed opportunities to institute effective treatment strategies such as medication reduction.
Instituting evidence-based diagnostic guidelines, as have been developed by the European Union, might lessen the extent of unnecessary testing.7 Basing subsequent testing on the results of the initial history and examination, and prioritizing higher yield tests would ensure a more informed and cost-effective approach to evaluating older patients with syncope.
We are indebted to the following individuals from the Yale University School of Medicine; Peter Charpentier, M.P.H. for assistance in developing the database for this project, to Denise Acampora, M.P.H. for her help with human investigation approval and access to medical records, Jane Mallory, B.S. for manuscript preparation, and Leo Cooney, M.D. for his guidance and review of this manuscript. They received no compensation for assistance given on this project.
Financial Disclosures: This work was supported by grants from Yale University, the American Foundation for Aging Research, and the Hartford Foundation and by the Claude D. Pepper Older Americans Independence Center at Yale School of Medicine (#P30AG21342) from the National Institute on Aging.
This project was supported in part by the Yale Pepper Center (P60AG10469) from the National Institute on Aging and the American Foundation for Aging Research.
Presented in abstract form at the Annual Meeting of the American Geriatrics Society, Washington DC, April 30, 2008; Yale Student Research Day May 6, 2008.