Improving CT-Derived Fractional Flow Reserve Analysis: A Quality Improvement Initiative

Objectives The aim of this study was to identify factors and quality improvement strategies to improve coronary computed tomography angiography (CCTA) studies referred for fractional flow reserve derived from CT angiography (FFRCT) analysis. Methods Thirty randomly selected CCTAs were analyzed for quality control. A uniform CCTA protocol was implemented by an in-house steering committee, emphasizing the importance of adequate heart rate control and nitroglycerine usage. Sixty additional randomly selected CCTAs were evaluated for quality at multiple time points during intervention, and FFRCT acceptance rate was analyzed at the conclusion. Results Prior to the implementation of this quality improvement program, our overall institution-specific percent acceptance rate was 76.1% for FFRCT compared to the national average of >95%. Post-intervention, this was improved to an average acceptance rate of 90% for FFRCT analysis. Conclusions Establishment and strict adherence to CCTA imaging protocols with appropriate training and adequate buy-in of CT technologists and nurses is a viable way of improving the quality of imaging and subsequent patient care.


Introduction
Current guidelines recommend the use of non-invasive anatomic imaging, including coronary computed tomography angiography (CCTA), for first-line testing in patients with suspected intermediate-risk stable coronary artery disease (CAD) [1,2]. CCTA and invasive coronary angiography have high accuracy for the detection of coronary artery stenoses based on anatomy evaluation, but they are less helpful for the identification of functionally significant, flow-limiting stenoses of approximately 50-90% [1,3]. Fractional flow reserve derived from CT angiography (FFRCT) has been shown to be an effective adjunct to CCTA in the diagnosis of flow-limiting stenosis and determining prognosis [4]. FFRCT utilization is now being recognized by large insurers who are covering the cost of this added service.
The purpose of our institution's CCTA-FFRCT program is to improve the accuracy of diagnosis and management of CAD, avoid additional unnecessary procedures, reduce cost, and improve patient and physician satisfaction. Although FFRCT has been shown to be a valuable tool in the evaluation of intermediate-range stenosis on CCTA, the analysis is highly sensitive to scanning protocol and artifacts [4]. It has been shown that image quality of CCTA is closely associated with the heart rate at the time of study acquisition [5]. The rejection rate of FFRCT in the literature ranges from 2% to 33% mainly due to differences in imaging acquisition, study incompletion such as missing best diastolic or systolic reconstructions for myocardial segmentation, patient-specific factors including body habitus and motion, and artifacts including calcium blooming, motion, and low contrast [4,6]. In contradistinction, the rejection rate at our facility of radiologist-referred FFRCT was discovered to be as high as 50%, limiting comprehensive evaluation of coronary artery stenoses and further patient management. produced as well as the overall acceptance of CCTA studies for FFRCT analysis. Our primary measure was the overall improvement of FFRCT analyses provided in order to maximize yield of the study and improve patient management.

Initial assessment
An initial review of a subset of 30 randomly chosen CCTA studies out of a total of 114 studies performed in July and August 2019 was completed by a sub-specialist cardiovascular radiologist to determine the overall quality of the studies in addition to factors limiting the utility of the studies. CCTA studies were assessed according to CAD-RADS criteria, a standardized method of assessing the highest-grade coronary artery lesion for adequacy of study: excellent (no artifacts), good (minor artifact but good diagnostic quality), acceptable (moderate artifacts), or poor/suboptimal (severe artifacts) [7]. The first three categories, excellent, good, and acceptable, were deemed as diagnostic and acceptable for FFRCT evaluation, whereas the last category, poor/suboptimal, was deemed non-diagnostic and therefore unsuitable for FFRCT evaluation referral. A non-diagnostic study was defined as poor quality in one or more coronary artery segments, which would further preclude FFRCT evaluation. In an attempt to optimize the quality of the studies, direct lines of communication were established with the CT technologists and nurses to reveal protocol and acquisition shortcomings. To evaluate potential issues, a question and answer session with a brief didactic lecture was given by an expert cardiovascular radiologist to the CT technologists and the nursing staff, including an overview of coronary artery imaging and the purpose of performing CCTA and FFRCT at our institution. From this discussion with the technologists and nurses, the lead cardiac imager noted several confounding factors described by the staff that were potentially leading to poor image quality. These included a lack of a standardized protocol regarding administration of nitroglycerine and beta-blockers and a lack of understanding of how to assess studies for quality control on the scanner.
The standardized Coronary Artery Disease -Reporting and Data System (CAD-RADS) method of assessment was used to evaluate the final images obtained for all patients. The scoring system ranges from CAD-RADS 0 (complete absence of stenosis) to CAD-RADS 5 (presence of at least one totally occluded coronary artery) and allows for specific recommendations to be included in the radiology impression for appropriate management. All CCTA examinations at our institution with a CAD-RADS of 3 (50-69% stenosis) were referred for FFRCT analysis unless otherwise indicated. Finally, a complete FFRCT analysis was performed on all studies meeting our hospital's inclusion criteria.

Initial medication trial
To evaluate the effect of pre-scan medication use on overall CCTA image quality at our institution, the radiology nurses were instructed to give every patient nitroglycerine and a beta-blocker, if required and not contraindicated, to achieve a heart rate of <70 and preferably <60. All patients were to receive 0.8 mg of sublingual nitroglycerine and the additional premedication instructions for heart rate ( Table 1).

Heart rate (beats per minute)
Metoprolol dose and route of administration  *Oral medications were to be administered 3-12 hours prior to CCTA. **Metoprolol PRN was to be an IV push of 5 mg metoprolol every 5 minutes up to eight doses for a target heart rate of 60 beats per minute and held for a systolic blood pressure of less than 100 mg.

CCTA, coronary computed tomography angiography
The nurses were then instructed to contact the cardiovascular radiologist on call if there was a contraindication to heart rate control or nitroglycerine administration. A list of contraindications for both nitroglycerine and beta-blocker administration was provided ( Figure 1).

FIGURE 1:
Nursing checklist provided to the radiology nurses to be completed prior to nitroglycerine and metoprolol administration for the CCTA examination.

CCTA, coronary computed tomography angiography
A second random sample of 30 CCTA cases performed during the month of November 2019 was reviewed by the same expert cardiovascular radiologist to assess for early post-intervention changes following the initial medication changes to confirm that improvement could be seen in the visual quality score.

Standardized protocol creation and implementation
An in-house steering committee was created comprising radiology administration, cardiovascular radiologists, cardiologists, an emergency medicine physician, CT technologists, and radiology nursing staff. The purpose of the committee was to make the CCTA acquisition protocol uniform. A proposed imaging pathway was developed with consensus by the steering committee members, which emphasized the need for consistent medication administration prior to scanning. An electronic order set was created within our electronic medical record to aid in the consistency of this protocol.
The steering committee recommended that one-on-one training be provided to the technologists regarding appropriate scanning parameters, contrast dosages, and injection rate ( Table 2). Monthly meetings were conducted by the same cardiovascular radiologist with the head technologists to discuss problem areas and concerns.
Siemens Force Dual Source CT; image during inspiration -acquire imaging from the carina through apex of the heart; run a test bolus; most patients will receive retrospective gating for function and also get coronary artery calcium scan   Figure 2 is an overall timeline demonstrating our intervention strategy and assessments, and Figure 3 demonstrates the overall CCTA acquisition workflow created. Figure 4 illustrates the individual improvement steps that were performed in order to reach our goal of improved CCTA acquisition for successful FFRCT analysis.

Analysis after medication administration education
The second analysis was performed to evaluate the effect of the metoprolol and nitroglycerine trial. The quality assessment again consisted of 30 CCTAs, this time with 21 diagnostic studies and only 9 studies with one or more poor quality coronary artery segments. Of the nine non-diagnostic studies, three (33%) were due to cardiac motion, five (55%) were due to respiratory motion, and one (11%) was due to suboptimal contrast bolus timing.
Given the success of the initial pre-scan medication trial at improving imaging quality, a standardized protocol was created and implemented by the in-house steering committee. One-on-one training with the CT technologists, including training for improved breath holds, was implemented to further improve protocol compliance.
One-sample correlation power calculation for follow-up duration revealed a power of 0.633, and the power calculation for patient number (n = 63) revealed a power of 0.99 ( Figures 5A, 5B). A linear weighted correlation showed a moderately strong linear correlation between months after the initiative began and the proportion of accepted studies (R = 0.697; p < 0.001) ( Figure 6).  Further FFRCT analysis was requested from pre-intervention until two months following the intervention (Figure 7). Our FFRCT acceptance rate increased to an average of 90% over the last three months of data collection.

FIGURE 7: FFRCT analysis pre-intervention (November) through intervention phase and follow-up.
Number of cases is represented by the bars (right-sided graph legend), and the acceptance rate is represented by the line graph (left-sided graph legend). Bar graph generated using ggpubr: 'ggplot2' Based Publication Ready Plots, R package version 0.3.0 (https://CRAN.R-project.org/package=ggpubr). Initial rejection rates for FFRCT analysis reported throughout the literature range from 13% to 33%, with significant cardiac motion commonly cited as a reason for rejection [4,6]. Pontone et al. found that temporal resolution, section thickness, and heart rate are independent predictors of rejection for CCTA FFRCT analysis [5,9]. They postulated that their own study had a low rejection rate of 2.9% because of the use of dual-source technology and wide-coverage single-source scanners. While prior studies have shown that the use of nitrates and beta-blockers improves accuracy of the FFRCT analysis, overall analysis of success rates remain moving targets [12][13][14]. Vasodilation plays a role in improving overall analysis and visualization of the entire length of the coronary artery; however, this also must be balanced with decreased cardiac motion for quality assessment [14]. These studies emphasize that many factors must be optimized and controlled in order to produce a high-yield study that can be referred for FFRCT analysis, but specific protocols for imaging acquisition and specific factors influencing the analysis remain incompletely understood.

Limitations
We took a holistic approach to correcting the CCTA quality issues at our institution. Thus, we are limited in our evaluation regarding the extent to which each of the individual factors that we adjusted for affected the outcome measure of rejection/acceptance of FFRCT analysis.
A significant limitation present in this study is the sample size of each group. However, we felt that a random sample of 30 studies, representing approximately 30% of the CCTA studies/month, was an adequate representation. Continuing data acquisition of prospective patients sent for FFRCT analysis, as well as increasing the retrospective sample data, could increase the overall power of the study.
Finally, we do not want to underestimate the need for CT technologist and radiology nurse buy-in and retraining. When the issues were brought to their attention and they were subsequently given training, this likely controlled for many confounders that were problematic in FFRCT analysis acceptance rate and CCTA quality.

Conclusions
Prior to intervention, our institution-specific overall FFRCT analysis acceptance rate was lower than average. Initial review demonstrated that 16 CCTA studies were deemed poor quality, of which eight were secondary to motion. Given that motion was a key contributory factor in visual quality inspection in addition to the success of FFRCT analysis, other unidentified factors needed to be controlled for in a holistic and protocoldriven manner.
Additional measures that were also taken during the education process, which certainly contributed to improved outcomes, were not directly assessed, including education on appropriate use of vasodilators. This likely resulted in better visualization of coronary vasculature. Moreover, a checklist was provided to the nursing staff to be completed prior to CCTA, which encouraged documentation of baseline cardiovascular status as well as relevant medical questions, factors that may affect or preclude the CCTA. Encouraging the nurses and technicians to call for any questions also worked to allow more open communication with the cardiac imagers, especially with more complex cases.
Providing a standardized protocol for adequate heart rate control allowed us to decrease cardiac motion but also provided more strict acquisition parameters to control for other unknown variables, which proved to be an important aspect of improving the FFRCT analysis acceptance rate. In short, our holistic approach and newly issued standard protocol likely controlled for other factors, thus improving our FFRCT analysis acceptance rate.

Additional Information Disclosures
Human subjects: All authors have confirmed that this study did not involve human participants or tissue. Animal subjects: All authors have confirmed that this study did not involve animal subjects or tissue.

Conflicts of interest:
In compliance with the ICMJE uniform disclosure form, all authors declare the following: Payment/services info: All authors have declared that no financial support was received from any organization for the submitted work. Financial relationships: All authors have declared that they have no financial relationships at present or within the previous three years with any organizations that might have an interest in the submitted work. Other relationships: All authors have declared that there are no other relationships or activities that could appear to have influenced the submitted work.