Analysis of Temporal Trends and Variation in the Use of Defibrillation Testing in Contemporary Practice

Key Points Question What are contemporary trends, institutional variation, and patient and hospital characteristics associated with defibrillation testing among patients undergoing first-time implantable cardioverter-defibrillator implantation? Findings In this cross-sectional study of 499 211 patients undergoing implantable cardioverter-defibrillator implantation, there was a significant decline in the use of defibrillation testing, yet there was a marked increase in institutional variation. Meaning There is variability in the deadoption of defibrillation testing which is not based on patient characteristics and could reflect differences in individual or institutional practices.


Introduction
Implantable cardioverter-defibrillators (ICDs) improve the survival of patients at high risk for sudden death. [1][2][3][4][5] Defibrillation testing (DFT) provides a means to evaluate the ability of a newly implanted device to detect and terminate ventricular fibrillation. Defibrillation testing was routinely used in the randomized clinical trials for which they are indicated, is included in the US Food and Drug Administration's labeling of ICDs, and the testing was adopted in routine clinical practice. 6,7 However, the benefits of DFT have not routinely been demonstrated. Several studies published between 2008 and 2012 found no association between routine testing and the efficacy of ICD shocks during follow-up or the risk of arrhythmic death. [8][9][10][11] Furthermore, contemporary ICDs rarely fail in their ability to detect and treat ventricular arrhythmias. Improvements in device technology provide defibrillation thresholds that are typically 10 to 20 J less than the maximum ICD output such that successful defibrillation occurs in more than 90% of clinical shocks. 6,[10][11][12][13][14][15][16] Complications of DFT, while relatively rare, can be serious. [11][12][13][14] Given the evolution of the evidence and guidelines, we sought to evaluate contemporary temporal trends and institutional variation in DFT at the time of initial ICD implantation and determine patient and hospital characteristics associated with DFT within the National Cardiovascular Data Registry (NCDR) ICD Registry between 2010 and 2015. The results could provide a perspective on the deadoption of a widely used clinical practice.

Data Source
Details of the NCDR ICD Registry have been reported previously. [17][18][19] The Centers for Medicare & Medicaid Services mandated that data on all Medicare primary prevention implants be entered into the NCDR ICD Registry until the data collection requirement ended on February 15, 2018. 17 The mandate was in effect during the study years for this analysis. The registry uses both a standardized data set and definitions, has requirements in place to ensure uniform data entry and transmission, and is subject to data quality checks. All data submissions are evaluated for errors and completeness. This information is summarized in an automated report that is sent to the participants after each data submission. The NCDR audit program, which includes hospital medical records reviews and blinded data abstractions, serves as an additional mechanism to assess the accuracy of the data and enables participants to identify areas for improved data entry. Use of the NCDR was approved by the ICD

Study Population
All patients undergoing initial implantation of a transvenous primary or secondary prevention ICD (single chamber, dual chamber, or cardiac resynchronization with defibrillator) between April 2010 and December 2015 were included. Patients were excluded if their ICD procedure was a replacement of a prior system, a generator replacement, or a lead revision. Patients younger than 18 years were included but represented only 0.35% of the cohort. Patients receiving a nontransvenous ICD system or a subcutaneous ICD, as well as those without documentation about whether DFT was performed, were also excluded.
infarction, cardiac tamponade, peripheral embolus, transient ischemia attack or cerebrovascular accident, urgent cardiac surgery) and in-hospital death were identified from the NCDR ICD registry.

Statistical Analysis
Baseline characteristics were compared between patients who did and did not have DFT performed at the time of implantation using χ 2 tests for categorical variables and 2-tailed, unpaired t tests for continuous variables. Continuous variables are presented as means (SDs) and categorical variables as frequency and percentage. Patient, hospital, and clinician factors were compared using the χ 2 test for categorical variables and t test for continuous variables. Missing values were rare (Յ1%) for all of the factors under consideration except left-ventricular ejection fraction (1.5%) and QRS interval duration (milliseconds) (7.1%); dummy variables indicating missing values for these factors were created and included in the models; missing values were imputed as the most frequent value of the categorical variables and the median of the continuous variables in the models. A multivariable analysis was performed to identify the patient and hospital factors associated with the use of DFT. Variables in the model were selected based on clinical experience and prior literature and included patient demographics, clinical characteristics, physician training, and hospital characteristics.
Complications and death were adjusted using similar variables.
To assess temporal changes in DFT, we examined its use since 2010 by quarters and tested the trend using the Cochran-Armitage trend test. Subgroup analysis was then performed to assess trends in specific patient populations, including primary and secondary prevention indication, and patients with structural abnormalities, including hypertrophic cardiomyopathy and congenital heart disease.
We also assessed use in patients according to their estimated risk for an elevated defibrillation threshold based on a risk score previously derived and validated in the NCDR ICD Registry. 20 This score is calculated based on the following characteristics: age younger than 70 years ( Hospital variation in the use of DFT was assessed by stratifying hospitals into quartiles of all DFT across the study years. To assess the institutional level variation in use of DFT, we determined the distribution and interquartile ranges of DFT in the entire cohort and in each year. To quantify the extent to which variation was explained by hospital level effects, the hospital-specific median odds ratio (MOR) was calculated using a validated method for the entire cohort and for each year. 21 Hierarchical logistic regression models were used to determine the between-hospital variance of DFT use for clustering of patients within hospital performed and the MOR was calculated. The MOR represents the odds that a randomly selected patient receiving DFT at a hospital with high testing rates would be tested than if he or she had received care at a hospital with low DFT rates. Findings were considered significant at P < .05. Data analysis was performed from May 20, 2015, to August 15, 2019, using SAS, version 9.4 (SAS Institute Inc).

Temporal Trends and Variation in DFT
Temporal trends in the use of DFT are depicted in

Discussion
This study evaluating the use of DFT among 499 211 patients in the NCDR ICD Registry in 1794 US hospitals has several findings. First, the use of DFT at the time of ICD implantation appeared to decline significantly between 2010 and 2015. Second, there seemed to be variation in the use of DFT  The median (center line) and interquartile range (top and bottom borders of box) of defibrillation testing with median odds ratio for each study year. Whiskers bars indicate 95% CI; circles, outliers. based on patient, clinician, and hospital characteristics, and the omission of DFT was possibly associated with the severity of comorbidities. Third, the temporal decline in DFT appeared to be accompanied by an increase in hospital-level variation.

Trends in DFT
Historically, DFT was performed in all patients undergoing ICD implantation. However, with improvements in technology, current-generation ICDs are capable of high-energy, biphasic, and tunable defibrillation waveforms such that successful defibrillation occurs in approximately 90% of clinical shocks. 6,10 A post hoc analysis of the Sudden Cardiac Death in Heart Failure Trial published in 2008 demonstrated that baseline DFT did not indicate the possibility of long-term mortality or shock efficacy. 8 The Shockless Implant Evaluation trial assessed the effect of DFT in patients undergoing initial left-sided ICD implantation. 22 The no-testing arm was noninferior to the testing arm for the primary outcome of arrhythmic death or failed appropriate shock. A second randomized trial found that forgoing DFT was noninferior to routine DFT using a primary end point of average first-shock efficacy for ventricular tachycardia and ventricular fibrillation episodes. 23  Many questions remain unanswered regarding DFT owing to limited data to provide evidencebased recommendations. It is reasonable to withhold DFT among patients with severe heart failure, valvular disease, or atrial arrhythmias and subtherapeutic anticoagulation. Alternatively, it is reasonable to perform testing among patients with right-sided implants or those with advisory leads at the time of generator change, and it is recommended among patients undergoing subcutaneous ICD. 24,25 Further questions remain regarding how best to approach patients whose conditions were poorly represented in trials (hypertrophic cardiomyopathy or congenital heart disease), those who are at very high risk for recurrent arrhythmias (secondary prevention indication), or those at risk for an elevated defibrillation threshold (amiodarone use, right-sided generator implants).
We found that, while DFT has declined overall, testing in certain populations varied. For instance, patients with a secondary prevention indication and structural abnormalities were more likely to undergo testing compared with those with a primary indication and those without structural abnormalities. This difference likely reflects the fact that these patients are at higher risk of appropriate ICD therapies, in which case testing is believed to provide reassurance that the system will abort ventricular arrhythmias when needed. Alternatively, the number and severity of comorbidities appeared to be related to the omission of DFT, which reflects that patients who are sick and/or frail were preferentially chosen to not undergo DFT. Prior factors that appeared to be associated with the absence of DFT include advanced age, a history of heart failure, advanced NYHA class status, atrial arrhythmias, lower ejection fraction, a wider QRS interval, and a higher serum creatinine level. 26 However, paradoxically, these patients who are not undergoing DFT may be more likely to benefit. Further studies are needed to determine which patients gain the most benefit from DFT. practicing in regions where they trained, or a monetary incentive to performing testing as there is a separate Current Procedural Terminology code for DFT that can be charged. In addition, the differences in regional use of DFT may reflect changes in payment structures in which reimbursement is bundled and there might be less incentive to perform additional procedures.

Hospital-Level Variation in Use of DFT
However, given that hospitals in the fourth quartile (most use of DFT) had only an 11% change in use of DFT, while those in the first quartile had a 30% change in use of DFT over the study years, it seems that clinicians who routinely perform DFT have not changed their practice while others have markedly reduced their use of DFT.
The reasons for describing this change is that a fundamental aspect to quality improvement is reducing unwarranted variation in care. Identifying hospital-level variation is a first step in understanding the extent to which patient factors influence variation in care compared with institutional factors, which can serve to be processes for interventions in attempts to improve the quality of care.

Outcomes After DFT
In secondary analysis, we did not find evidence that DFT was associated with harm; in fact, there was a higher risk of adverse events among patients who did not undergo DFT. However, this outcome should not be interpreted as indicating that DFT is beneficial and/or without risk, and this finding is in direct contrast to prior literature suggesting that, while rare, serious adverse events occur during DFT. 13 The results of our study may be explained by 2 reasons: the number and severity of measured comorbidities was associated with the lack of DFT, such that patients who are sick and/or frail were preferentially chosen to not undergo DFT, and an effect-cause relationship, in which DFT was not performed because of an intraprocedural complications.

Evolution of Practice Patterns
The evolution of medicine occurs through a process of adoption and/or deadoption of practices.
More often, there is a focus on adoption of newer interventions that outperform established practices. However, there must also be deadoption of established standards, not because a better intervention has been developed, but because what was once thought to be beneficial is not. A wellrecognized example of this was the previously held belief that suppression of ventricular ectopy following acute myocardial infarction, which was an independent predictor of mortality, would improve arrhythmic death. However, the use of class 1C drugs, which successfully reduced ventricular ectopy, also resulted in an excess of mortality; thus, there was a deadoption of the use of class 1C drugs among patients with structural heart disease. 27 There are numerous other examples of deadoption, which are best seen in the evolving changes in clinical guidelines. 28,29 The present study may provide a perspective on the pace of deadoption of a once-held belief that DFT needs to be performed during ICD implantation. Despite mounting evidence that omission of DFT is safe, there is still significant variation in its use, with some institutions continuing to apply DFT frequently, which is independent of patient characteristics. As with adoption, deadoption requires a focus on the cultural aspects of practice, which have evolved from not only evidence-based practices, but also personal experiences, conflicts, and biases.

Limitations
This study has limitations. Certain factors should be considered in the interpretation of the findings.
First, the reasons behind decisions to perform or not perform DFT are not gathered from the NCDR ICD Registry. Second, the decision on use of DFT in a small portion of patients was based on their upper limit of vulnerability, which does not necessarily have the same risks as the lowest energy required (ie, patients do not need to have induction of ventricular fibrillation during the upper limit of vulnerability). In addition, defibrillation threshold testing is likely a misnomer for many centers in which a single shock of 10 to 15 J or greater is an acceptable defibrillation safety margin. Third, high-risk features exist for an elevated defibrillation threshold, including right-sided generator