Assessment of inpatient multimodal cardiac imaging appropriateness at large academic medical centers

Advances in cardiac imaging have led to vast improvements in our ability to diagnose heart disease. These advances have led to an increase in cardiac imaging utilization and healthcare spending in this area as a result [1, 2]. In response to concerns regarding increased utilization, the American College of Cardiology Foundation (ACCF), in conjunction with other specialty societies, published appropriate use criteria (AUC) for various imaging modalities, including transthoracic (TTE) and transesophageal echocardiography (TEE), radionuclide imaging and diagnostic cardiac catheterization [3‐5]. These initially classified studies as ‘appropriate’, ‘may be appropriate’ or ‘rarely appropriate’, but subsequently updated the terms to ‘appropriate’, ‘may be appropriate’ and ‘rarely appropriate’ [6]. Previous retrospective studies using the AUC to classify imaging have identified rates of up to 22 % for TTE in the category of rarely appropriate, from a variety of clinical settings [7‐10]. Appropriateness studies of single-photon emission tomography myocardial perfusion imaging (SPECT), and cardiac catheterization have shown similar rates of rarely appropriate use [11‐13].

While most studies have focused on assessing appropriate use rates for individual cardiac imaging modalities, no prior studies have examined the rates of appropriateness of multiple cardiac imaging modalities simultaneously. To better understand appropriateness of multi-modal cardiac imaging in the acute care setting, we performed a retrospective review of in-patients undergoing a range of cardiac investigations at multiple large academic medical centers in Toronto, Canada, using the AUC for the most common cardiac imaging modalities. We hypothesized that the combined rate of ‘rarely appropriate’ use for several combined imaging modalities would be comparable to prior studies of single imaging modalities in other countries.

In this study we have presented data showing the application of AUC across a variety of imaging modalities in the in-patient population from three distinct academic teaching hospitals in a single payer healthcare system. To our knowledge, this is the first study to look specifically at the ordering practices of physicians in the acute care setting with respect to the range of cardiac imaging at their disposal. We have demonstrated that the recent versions of the AUC are capable of classifying the vast majority of clinical indications for cardiac imaging in the modern acute care setting. We also found that the combined rate of rarely appropriate ordering for cardiac investigations was low in all three study sites and between specialties, despite varying use of each imaging modality (Fig. 1).

A recognised limitation of many previous studies using AUC is that it can be difficult to determine from the written record the exact clinical indication for a particular test. In a large, single centre investigation into the appropriate use of transthoracic echocardiography, Bailey et al. [14], were able to achieve high levels of clinical reasoning determination (92 % of the cases) by undertaking a comprehensive review of admission history and physical examination, specialist consults, laboratory results and discharge summaries. In our work, we believe that by reviewing the daily chart entries in addition to the admission consult and electronic data, we were better able to appreciate the medical team’s reasoning and thus were confident in our ability to determine the clinical indication for testing in all the cases we reviewed. We were able to find a suitable clinical category for all but one the cases reviewed in the most up to date AUC for each imaging modality. This is similar to many other studies in the US and Europe which have found that the most recent AUC for echocardiography, SPECT and diagnostic catheterization are able to capture the vast majority of clinical situations and generally have high rates of classification [7‐9, 12, 13, 15, 16].

‘The low rates of rarely appropriate use we observed likely represent the fact that the acute change in clinical status that brings most patients to hospital often justifies in-hospital cardiac imaging. This is reflected in the acuity of the most common indications for testing in each modality, which broadly included a new diagnosis of heart failure (TTE AUC #70), suspicion for infective endocarditis (TTE AUC #52 and TEE AUC #108), confirmed or suspected ACS with adjunctive work up (diagnostic catheterization AUC #3, TTE AUC #24 and SPECT AUC #6), new arrhythmia (TEE AUC #112) or worsening symptoms in patients with known coronary disease (SPECT AUC #30). This has been noted in previous studies, such as that by Matulevicius et al. [10], who noted higher rates of appropriate trans-thoracic echo in in-patients compared to out-patients. Studies looking at perfusion stress imaging have generally not separated in-patient and out-patient groups however, the most common reasons for rarely appropriate studies in these investigations usually involve asymptomatic, low risk patients or those with known stable coronary disease [11, 13, 17]. As hospitalised patients tend to be symptomatic and have a greater number of risk factors than the out-patient population, it is unlikely that many in-patients will be in a low risk category. The rate of rarely appropriate diagnostic angiograms seen here was also lower here than that reported elsewhere. Studies by Hannan and Mohareb [12, 18] (who also performed their study in Ontario) both found higher rates of rarely appropriate use - 24.9 and 10.8 % respectively; however both of these studies excluded both investigation of acute coronary syndrome and valvular heart disease, which made of 72 % of our studies and are deemed appropriate by the AUC in all circumstances. A large study by Chan and colleagues assessed appropriateness of percutaneous coronary interventions and found an appropriateness rate of 98.6 % in patients having PCI for an acute indication, which is consistent with the low rarely appropriate rate in our study [19].

To our knowledge, this is the first study to investigate appropriateness of cardiac imaging across a range of modalities, and so provide a holistic picture of cardiac imaging appropriateness at individual centers. In our study, we found preferential usage of different imaging modalities between sites, with site 1 performing relatively few catheterizations compared to site 2, and site 3 performing more SPECT and less TEE compared to the other sites. The reasons for this could include site specific specialization, e.g. one site is the cardiac surgery center for all three sites and two of the three sites have specialized heart failure clinics, variable availability of cardiac imaging investigations, staff preference, or site specific factors which could not be accounted for in this study. With these variables in mind, studies that examine only one modality per site may miss important ordering patterns across other areas of cardiac imaging. For example, a study may note a high level of rarely appropriate use of stress echo, but miss the fact that, due to higher staff familiarity, they order many more SPECT with a high level of appropriateness. Recognising the range of imaging techniques in use, the ACCF has most recently published disease specific imaging AUC [20, 21] rather than modality specific guidelines to aid clinicians in choosing between the options available to them. This approach is useful for patients with chronic diseases, such as heart failure or chronic ischemic coronary disease, but would not capture patients having imaging without those diseases. With this in mind we believe that assessing and reporting on combined cardiac imaging appropriateness across modalities could serve as a useful tool for clinical departments to determine comprehensive imaging ordering patterns and appropriateness. Moreover, a combined appropriateness rate for all cardiac imaging modalities could become an important quality improvement indicator that takes into account site specific practicalities and preferences. Although low levels of rarely appropriate use were noted across the board in this study on acute care inpatients (6 and 3 % respectively for GIM and Cardiology), this may not be the case in other clinical environments, particularly in the ambulatory care environment. A quality improvement indicator could then be used to identify areas of improvement and to help evaluate interventions designed to improve appropriate cardiac imaging ordering.

Our study has important limitations that warrant mentioning. As our data was collected retrospectively from the written record, it is possible that the clinical indication for a test was not fully captured. However, we believe that we were able to draw on sufficient information from each case (admission note, consults, electronic requests and daily chart entries) to minimise this risk. Also, unlike non-invasive cardiac imaging, diagnostic catheterization requested by the GIM service would usually require an inpatient Cardiology consult before it could take place. Although this obviously affects the diagnostic catheterizations we have reported as being performed under the GIM team, we felt that this group of patients represented a different sample from those directly under the Cardiology service and that it was important to analyse them as such. In addition, as the GIM team would remain the most responsible physician and it is unusual for the Cardiology service to refuse an angiogram that has been specifically requested. Finally, this study was conducted at academic health sciences centers, whose results may differ from community practices.

In conclusion, the ACCF AUC are able to capture the vast majority of clinical scenarios for multiple cardiac imaging modalities in this multi-centered retrospective study on general cardiology and GIM inpatients. Overall, the appropriateness ordering rate is high across all imaging modalities. We recommend that further work aimed at improving appropriate utilization of cardiac imaging resources should focus on investigations performed in the out-patient setting.