Temporal Changes in Appropriateness of Cardiac Imaging
Temporal Changes in Appropriateness of Cardiac Imaging
This comprehensive evaluation of published data evaluated AUC for different cardiac imaging techniques from 103,567 tests grouped in 10 different cardiac imaging outcomes. Meta-regression was used to evaluate the temporal trend of appropriateness, based on the year of publication. In contrast to reports comparing the behavior of specific groups of physicians over time, the results are an indication of "real-world" practice at sites publishing their appropriate use data. There are 5 important observations. First, the improvement of appropriate use from the original to the revised versions may merely reflect easier classification of patients and a change in attribution of proportion of appropriate testing rather than a change in practice. Second, we found a temporal improvement in percent appropriateness for TTE, TEE, and CTA, but no evidence of a change in the number and proportion of appropriate testing for other modalities. Overall rates of appropriate use for CTA and stress echocardiography remain low, and those for SPECT are only modest. This implies a disconnection between clinical practice and AUC that warrants better understanding. Third, this limited change has not matched the reduction in imaging tests over the last 5 years, suggesting that physician use of AUC in the ordering process may not have played a major role in this reduction. However, an indirect role (through AUC influence on the decisions of radiology benefit managers) cannot be excluded. Fourth, the proportion of appropriate use presented here may well be shown to its advantage in these retrospective and largely unblinded evaluations, which for the most part were performed by physicians able to identify appropriate indications, even if this was not the primary reason for the test. Moreover, the evaluations were mainly performed at the point of service of academic medical institutions. There was substantial variation between observers, especially between physicians. Finally, although there was some evidence of publication bias for stress echocardiography 2011 and CTA 2010, bias was not identified for most scenarios.
The observed heterogeneity of the proportion of appropriate testing among reports might be expected on the basis of a wide variety of participant characteristics, study designs, and types of hospital and regions in the published data. However, the unique aspect of this work is its examination of the temporal variation in proportion of appropriate testing.
The drivers of test ordering are complex, and the persistent rate of w80% for TTE (and less for other modalities) perhaps testifies to a variety of influences that are not disease specific. These include factors individual to the patient, such as factors related to other comorbidities and situational considerations. These features may drive the request for testing in a situation when the test is considered "rarely appropriate" and explain the stability of this attribution. In addition, clinical practice guidelines may be discordant with AUC. Finally, the adjudication of appropriate testing is often inconsistent because appropriate and "rarely appropriate" reasons for testing may coexist in the same patient. The implication is that the ordering physician may choose an existing appropriate indication rather than the real clinical issue. This may be likely when the proportion of inappropriate (or "rarely appropriate") tests is reviewed as part of the accreditation process.
In this respect, the increment of "rarely appropriate" tests from the first to the second versions of stress echocardiography, SPECT, and CTA was a surprising finding of this meta-analysis. Interestingly, none of these tests showed a gradation of "rarely appropriate" use within the time frame of each edition; therefore, this finding likely reflects the change in criteria rather than a change in practice.
The use of AUC as a process to reduce costs neglects the fact that testing labeled as "sometimes" and "rarely appropriate" is very reasonable in some situations. Indeed, this is a shortcoming of the widespread use of radiology benefit managers as a tool to control the use of cardiac imaging: they are inflexible to situational demands. The application of AUC at the point of care (e.g., using electronic tools that help physicians to choose "appropriately") has produced similar results to the radiology benefit managers but has the same limitation.
Although the nuances of specific clinical scenarios make the AUC problematic for controlling testing, they are potentially valuable as a yardstick for education. The evidence available regarding the value of educational campaigns is contradictory. In the interpretation of responses to AUC campaigns, it should be noted that knowledge of AUC is only 1 component of test selection. Test selection is also influenced by the characteristics of health professionals, features of practice settings, incentives, linkage of AUC performance with accreditation or licensing bodies, patient factors, compatibility with existing practice and beliefs, and perceived quality of the guidelines. Moreover, how much repetition is required for an educational campaign to have a sustained effect is unclear.
In jurisdictions in which the laboratory is responsible for appropriate use, a strategy of laboratorybased audit is needed for the thousands of cardiac imaging requests that are submitted to the laboratory every year. The use of AUC to facilitate auditing is more likely to be effective than its application to individual test requests. Tests that are most likely to be of "maybe" or "rare" appropriateness include those requested in younger patients, those with previous tests, those who are outpatients, or tests that are reevaluations in patients who are asymptomatic or have no changes in clinical status. These situations might be used as markers of potential inappropriate use in individual patients.
Most cardiovascular imaging is performed in the community practice environment, whereas most AUC reports have been reported in academic medical centers. Nonetheless, the wide discussion of AUC over the last decade might be expected to influence all environments, and a practicespecific variation of some tests and not others seems unlikely. It is unclear whether changes in percent appropriateness reflect better test selection rather than observer-expectancy effect or coding of indications to satisfy AUC.
Discussion
This comprehensive evaluation of published data evaluated AUC for different cardiac imaging techniques from 103,567 tests grouped in 10 different cardiac imaging outcomes. Meta-regression was used to evaluate the temporal trend of appropriateness, based on the year of publication. In contrast to reports comparing the behavior of specific groups of physicians over time, the results are an indication of "real-world" practice at sites publishing their appropriate use data. There are 5 important observations. First, the improvement of appropriate use from the original to the revised versions may merely reflect easier classification of patients and a change in attribution of proportion of appropriate testing rather than a change in practice. Second, we found a temporal improvement in percent appropriateness for TTE, TEE, and CTA, but no evidence of a change in the number and proportion of appropriate testing for other modalities. Overall rates of appropriate use for CTA and stress echocardiography remain low, and those for SPECT are only modest. This implies a disconnection between clinical practice and AUC that warrants better understanding. Third, this limited change has not matched the reduction in imaging tests over the last 5 years, suggesting that physician use of AUC in the ordering process may not have played a major role in this reduction. However, an indirect role (through AUC influence on the decisions of radiology benefit managers) cannot be excluded. Fourth, the proportion of appropriate use presented here may well be shown to its advantage in these retrospective and largely unblinded evaluations, which for the most part were performed by physicians able to identify appropriate indications, even if this was not the primary reason for the test. Moreover, the evaluations were mainly performed at the point of service of academic medical institutions. There was substantial variation between observers, especially between physicians. Finally, although there was some evidence of publication bias for stress echocardiography 2011 and CTA 2010, bias was not identified for most scenarios.
Understanding Temporal Variations in Appropriateness
The observed heterogeneity of the proportion of appropriate testing among reports might be expected on the basis of a wide variety of participant characteristics, study designs, and types of hospital and regions in the published data. However, the unique aspect of this work is its examination of the temporal variation in proportion of appropriate testing.
The drivers of test ordering are complex, and the persistent rate of w80% for TTE (and less for other modalities) perhaps testifies to a variety of influences that are not disease specific. These include factors individual to the patient, such as factors related to other comorbidities and situational considerations. These features may drive the request for testing in a situation when the test is considered "rarely appropriate" and explain the stability of this attribution. In addition, clinical practice guidelines may be discordant with AUC. Finally, the adjudication of appropriate testing is often inconsistent because appropriate and "rarely appropriate" reasons for testing may coexist in the same patient. The implication is that the ordering physician may choose an existing appropriate indication rather than the real clinical issue. This may be likely when the proportion of inappropriate (or "rarely appropriate") tests is reviewed as part of the accreditation process.
In this respect, the increment of "rarely appropriate" tests from the first to the second versions of stress echocardiography, SPECT, and CTA was a surprising finding of this meta-analysis. Interestingly, none of these tests showed a gradation of "rarely appropriate" use within the time frame of each edition; therefore, this finding likely reflects the change in criteria rather than a change in practice.
Alternative Approaches to Reducing Cost
The use of AUC as a process to reduce costs neglects the fact that testing labeled as "sometimes" and "rarely appropriate" is very reasonable in some situations. Indeed, this is a shortcoming of the widespread use of radiology benefit managers as a tool to control the use of cardiac imaging: they are inflexible to situational demands. The application of AUC at the point of care (e.g., using electronic tools that help physicians to choose "appropriately") has produced similar results to the radiology benefit managers but has the same limitation.
Although the nuances of specific clinical scenarios make the AUC problematic for controlling testing, they are potentially valuable as a yardstick for education. The evidence available regarding the value of educational campaigns is contradictory. In the interpretation of responses to AUC campaigns, it should be noted that knowledge of AUC is only 1 component of test selection. Test selection is also influenced by the characteristics of health professionals, features of practice settings, incentives, linkage of AUC performance with accreditation or licensing bodies, patient factors, compatibility with existing practice and beliefs, and perceived quality of the guidelines. Moreover, how much repetition is required for an educational campaign to have a sustained effect is unclear.
In jurisdictions in which the laboratory is responsible for appropriate use, a strategy of laboratorybased audit is needed for the thousands of cardiac imaging requests that are submitted to the laboratory every year. The use of AUC to facilitate auditing is more likely to be effective than its application to individual test requests. Tests that are most likely to be of "maybe" or "rare" appropriateness include those requested in younger patients, those with previous tests, those who are outpatients, or tests that are reevaluations in patients who are asymptomatic or have no changes in clinical status. These situations might be used as markers of potential inappropriate use in individual patients.
Study Limitations
Most cardiovascular imaging is performed in the community practice environment, whereas most AUC reports have been reported in academic medical centers. Nonetheless, the wide discussion of AUC over the last decade might be expected to influence all environments, and a practicespecific variation of some tests and not others seems unlikely. It is unclear whether changes in percent appropriateness reflect better test selection rather than observer-expectancy effect or coding of indications to satisfy AUC.
Source...