López-Jiménez, Francisco

Abstract Aims Spectrum bias can arise when a diagnostic test is derived from study populations with different disease spectra than the target population, resulting in poor generalizability. We used a real-world artificial intelligence (AI)-derived algorithm to detect severe aortic stenosis (AS) to experimentally assess the effect of spectrum bias on test performance. Methods and results All adult patients at the Mayo Clinic between 1 January 1989 and 30 September 2019 with transthoracic echocardiograms within 180 days after electrocardiogram (ECG) were identified. Two models were developed from two distinct patient cohorts: a whole-spectrum cohort comparing severe AS to any non-severe AS and an extreme-spectrum cohort comparing severe AS to no AS at all. Model performance was assessed. Overall, 258 607 patients had valid ECG and echocardiograms pairs. The area under the receiver operator curve was 0.87 and 0.91 for the whole-spectrum and extreme-spectrum models, respectively. Sensitivity and specificity for the whole-spectrum model was 80% and 81%, respectively, while for the extreme-spectrum model it was 84% and 84%, respectively. When applying the AI-ECG derived from the extreme-spectrum cohort to patients in the whole-spectrum cohort, the sensitivity, specificity, and area under the curve dropped to 83%, 73%, and 0.86, respectively. Conclusion While the algorithm performed robustly in identifying severe AS, this study shows that limiting datasets to clearly positive or negative labels leads to overestimation of test performance when testing an AI algorithm in the setting of classifying severe AS using ECG data. While the effect of the bias may be modest in this example, clinicians should be aware of the existence of such a bias in AI-derived algorithms.

Introduction: Obesity is linked with heightened vulnerability to cardiovascular diseases including hypertension. Nonetheless, mechanistic studies addressing the effects of experimental weight gain on blood pressure are lacking. We sought to examine whether experimental weight gain raises ambulatory blood pressure in healthy individuals and identify any relationship between changes in blood pressure and changes in regional fat distribution. Methods: Twenty-six normal weight subjects were randomized to a 8-week period of weight gain through overfeeding (N=16; age 30.4±6.6 years, BMI 23.5±3.5 kg/m 2 ) or weight maintenance (N=10; age 27.1±7.7 years, BMI 23.6±2.7 kg/m 2 ). Measurements of body composition by dual-energy X-ray absorptiometry and abdominal computed tomographic scans and 24-h ambulatory blood pressure monitoring were obtained at baseline and at follow-up. Results: Overfeeding resulted in an increase in body weight of 3.7±1.5 kg ( p <0.001) in weight gainers, with increments seen in total (21994±8247.4 to 25180.7±8563.1 gr, p <0.001), visceral (61.6±32.7 to 75.5±30.9 cm 2 , p =0.002) and subcutaneous fat (135.5±77.4 to 167.9±82.9 cm 2 , p <0.001). No changes occurred in the maintenance group. Weight gainers exhibited an increase in 24-h systolic blood pressure at follow-up (113.7±8 to 117.7±7.9 mmHg, p =0.009) and mean blood pressure (MAP) (85.1±4.9 to 86.8±5.1 mmHg, p =0.02), while blood pressure was unchanged in controls. Changes in MAP were positively correlated only with changes in visceral fat (rho=0.452, p =0.02) but not with changes in weight or any other body composition measure. Conclusion: Modest weight gain leads to marked elevation in 24-h blood pressure in lean healthy subjects. The association between increased MAP and abdominal visceral fat accumulation suggests that visceral deposition of adipose tissue may contribute specifically and mechanistically to the enhanced risk of hypertension associated with weight gain.

Background: Regular exercise is considered a key element to maintain weight loss, but its role to induce weight loss has been controversial. Methods: In this retrospective study of 7237 individuals attending a worksite wellness center from September 2008 to October 2011 to assess predictors of weight loss at 12 months after joining. Upon joining the wellness center, participants completed a survey that assessed for the presence of cardiovascular risk factors, height, weight, self-perceived health behaviors and psychosocial factors using validated 11-point scale items. We limited the sample to 1927 overweight (25 < BMI ≤ 30) and obese (BMI > 30) participants who completed the survey in full at baseline and at 12 months (+/- 90 days). The outcome variable was weight change at 12 months after enrollment. Weight loss was defined as a loss of > 1kg. Results: Of 1927 participants, 998 (52%) were either overweight (n=573; 57%) or obese (n=425; 43%), and 60% were women. Mean (SD) age was 41 (11.5) yr, BMI 30.7 (5.5) kg/m 2 and they visited the wellness center 49 (90) times/year. At 12 months, 416 (42%) of the participants had lost >1 kg and 19% had lost > 5kg. The predictors of weight loss were: baseline BMI, [Odds Ratio (OR)= 1.56 (1.15-2.13); p < .0044] and [OR= 1.75 (1.26-2.44); p < .0010] for BMI between 30-35 and BMI ≥35, vs BMI between 25-29, respectively. In addition, visits/year [OR= 1.60 (1.19-2.16); p < .0010] and [OR= 1.85 (1.37-2.61); p < .0001] for those attending 13-52 sessions and > 52 visits/year, vs ≤ 12 visits/year, respectively. The association persisted after controlling for all covariates (ie, cardiovascular risk factors, self-perceived health behavior and psychosocial factors). The Table shows the results of the multivariate analysis assessing predictors for >1kg weight loss. Neither age, gender, perceived stress, self-perception of being overweight, following a healthy diet, nor history of hypertension, high blood cholesterol or diabetes predicted weight loss. Conclusions: A significant percentage of overweight or obese individuals lose weight after enrolling in a worksite wellness center. Frequency of attendance, baseline BMI, and support for maintaining a healthy living are associated with weight loss. These results suggest interventions to promote usage of wellness centers maybe be beneficial for the obese adult.

Introduction: Cardiac rehabilitation (CR) is an effective but highly underutilized therapy for heart disease. Efforts are underway to increase CR referral and enrollment, but little is known about the potential capacity for growth in CR utilization in the United States. To address this concern, we estimated the current national capacity of CR programs across the United States and assessed obstacles to potential growth. Methods: We surveyed all CR Program Directors listed in the American Association of Cardiovascular and Pulmonary Rehabilitation (AACVPR) database in November of 2012. Respondents reported current enrollment levels, current program capacity, program capacity given reasonable expansion, and obstacles to growth. Results: Of 812 Program Directors in the AACVPR database, 290 (36%) completed the full survey. Respondents represented somewhat larger programs compared to non-respondents (4.0 vs. 4.6 full time employees, p = 0.01), but were otherwise similar. Current enrollment levels demonstrated a median [IQ range] of 140 [75,232] patients per year and a current estimated capacity of 192 [100,300] patient per year. Respondents estimated that programs could expand services by 25% to 240 [141,380] patients annually, assuming a significant increase in patient demand and a feasible increase in resources. We estimate that if programs filled to their current capacity, national CR utilization rates would increase from 34.7% to 46.4% (95% CI, 43.9 to 48.9) of eligible US patients each year. Given feasible program expansion, utilization could increase to a maximum of 58.4% (95% CI, 55.2 to 61.6) of qualifying patients. Capacity projections based upon absolute patient numbers and national statistics revealed concordant results. The most commonly cited obstacles to increasing patient participation are shown in the figure. As seen, the majority (88%) are controllable system-related factors unrelated to patient behavior. Conclusions: Even with substantial expansion of all existing CR programs, there is currently insufficient capacity in current CR programs to meet national service needs. Solutions to this problem will likely include the creation of new CR programs, and new policies that improve reimbursement for CR and also embrace new models of CR delivery.

Background: This is the first systematic review and meta-analysis to investigate the effect of resistance training (RT) on survival and other cardiovascular outcomes including ischemic heart disease events and stroke. Methods: An experienced librarian searched databases up to September 25 th , 2017, for randomized trials and cohort studies that evaluated the effect of RT on survival and cardiovascular events in the general population. The databases included Ovid MEDLINE, EMBASE, Cochrane Central Register of Controlled Trials, Cochrane Database of Systematic Reviews, and Scopus. Two investigators conducted the screening process independently and in duplicate. Cochrane tools were used to assess the risk of bias in clinical trials and observational studies. We calculated hazard ratios and 95% confidence intervals using RevMan and fixed and random effect models and had a subgroup analysis based on doses of RT and for the combination of RT and aerobic exercise (AE) vs no exercise. Results: The search identified 1429 studies from which 10 (one randomized trial) met the inclusion criteria, including 338,254 participants with a mean follow up of 8.14 years. The meta-analysis showed that RT, in comparison with no exercises, is associated with 24% lower all-cause mortality and 48% lower mortality when combined with AE. Based on subgroup analysis, performing 1-2 sessions of RT/week is associated with lower all-cause mortality by 28% (HR 0.72, 95% CI 0.66-0.78) whereas > 5 sessions of RT/week has no association with all-cause mortality (HR 0.99, 95% CI 0.76-1.31). Further, RT alone or combined with AE is associated with lower CV mortality compared to no exercise (Figures). Finally, RT alone also showed a borderline association with lower all-cancer mortality. Heterogeneity was present for several comparisons, and subsequent analysis will explore sources of this variability. Using study design-specific Cochrane risk of bias tools, no major sources of bias were identified in the included studies. One cohort study looked at the effect of RT on coronary heart disease events and found 23% risk reduction in men, while no study specifically assessed the effect of RT on cerebrovascular outcomes. Conclusion: RT is associated with lower all-cause, CV and all-cancer mortality. RT appears to have an additive effect when combined with AE.

Background: Cardiac Rehabilitation (CR) is an effective but highly underutilized treatment for heart disease. Prior studies have demonstrated that the use of strategies such as systematic referral (as part of an order set that does not depend on physician initiative or memory, usually with an opt-out option), liaison facilitated referral (a specific staff member who discusses the program, encourages attendance, and facilitates referral), an early appointment (< 2 weeks), and telephone appointment reminder calls can all improved enrollment rates. However, it is unclear how frequently these strategies are utilized across the United States, how often CR programs monitor their participation rates, and how often quality improvement (QI) projects designed to improve participation rates are performed. Methods: We surveyed all CR program directors in the American Association of Cardiovascular and Pulmonary Rehabilitation’s (AACVPR) database. We assessed basic program features, use of systematic referral, liaison facilitated referral, early appointments, and telephone reminder calls and QI projects. We also assessed self-reported referral, enrollment, and completion rates. Results: Response rate was 36% (290/812). Non-response bias analysis showed reasonable survey representativeness. Over the past 5 years, 49% of programs measured hospital referral, 21% office/clinic referral, 71% program enrollment, and 74% program completion rates. Self-reported rates (interquartile range) were 68 [32 to 90]% for hospital referral, 35 [15 to 60]% for office/clinic referral, 70[46 to 80]% for program enrollment, and 75[62 to 82]% for program completion. Programs reported utilizing a hospital-based systematic referral, liaison facilitated referral, or inpatient CR program 64%, 68%, and 60% of the time, respectively. Early appointments (< 2 weeks) were utilized by 35% and consistent phone call appointment reminders were utilized by 50%. Over the past 5 years, the percent or programs performing QI projects were 62% for hospital referral, 42% for clinic referral, 60% for program enrollment, and 57% for program completion. Measurement of participation rates was highly correlated with performing QI projects (p < 0.0001.) Conclusions: Although programs are aware of participation rate gaps, the monitoring of participation rates is suboptimal, quality improvement initiatives are infrequent, and proven strategies for increasing patient participation are inconsistently utilized. Awareness in outpatient clinical settings is particularly low. These issues likely contribute to the national CR participation gap and may prove to be useful targets for national QI initiatives.

Introduction: Clinical predictive value of transient prolonged ventricular conduction and repolarization during right ventricular (RV) pacing have not been studied. Hypothesis: Ventricular conduction and repolarization prolongation during RV pacing predict risk of arrhythmic and mortality events. Methods: Electrophysiology studies in 501 patients with cardiomyopathy (CM) were reviewed. Duration changes of QRS and QT interval during pacing from the RV apex were measured. Subjects were followed up for a mean of 3.2 years for arrhythmic events and total mortality. Results: Participants had a mean age 65 yrs, 80% male, 74% funcional class II or III, and a mean EF of 33.1%. A paced QRS duration > 190 ms was associated with more severe CM and in multi-variable analysis it was associated with a 3.5 times higher risk of ventricular arrhythmias or appropriate defibrillator therapy (HR 3.5, 95% CI 2.3-5.4, p<0.01) and 2.6 times higher risk of total mortality or arrhythmic events (HR 2.6, 95% CI 2.3-3.4, p<0.01) compared to a paced QRS duration <190 ms, independent of baseline EF or baseline QRS duration, and after adjustment for electrophysiology study results, baseline functional class, type of cardiomyopathy and other relevant clinical factors. Similarly a QTc interval during RV pacing > 620 ms was associated with a 2.7 times higher risk of arrhythmic events compared to a QTc interval during RV pacing < 620 ms. Conclusions: Prolongation of ventricular conduction (QRS > 190 ms) and repolarization (QTc > 620 ms) observed during RV pacing in subjects with CM are associated with higher risk of arrhythmic events, defibrillator therapies and all-cause mortality.

Background : Studies have shown an association between obesity and total mortality in patients with and without coronary artery disease. Less is known regarding the impact of obesity after cardiac arrest. Obese survivors of ventricular fibrillation (VF) out-of-hospital cardiac arrest (OHCA) may have worse outcomes than nonobese patients due to resuscitation challenges, long-term risks of cardiovascular disease, and other comorbidities. Methods : All patients who presented with a VF OHCA in Rochester Minnesota from November 1990-September 2006 were included. Patients were classified by weight (BMI <25, 25–30, >30). ICD shocks for VF/VT were determined by review of subsequent device interrogations. Results : 226 patients (age 63.7±13.5) were treated for VF OHCA with an average call-to-shock time of 6.3±1.8 minutes. 99 patients (44%) survived to hospital discharge with neurologic recovery. Data to calculate BMI were available in 213(95%). There was no difference between the relative distribution of body weight between the hospital survivors [<25(32%), 25–30(37%), >30(31%)] and nonsurvivors [<25(32%), 25–30(44%), >30(28%)], p=0.711. There was no difference in ICD implantation rates between weight groups (p=0.251). The patients were followed for an average of 5.8±4.4 years. The 5-year survival was 80±5%. 5-year survival was lower in underweight/normal patients [<25, 71% (63–79) compared to heavier patients (25–30:88% (82–94), p=0.001]. The survival difference persisted after adjustment for age (p=0.02). Malignancy was the most common cause of death in the underweight group. The 5-year survival free of ICD shocks was 61±7% with no weight-based difference in shocks. Conclusion : There was no apparent weight-based influence on resuscitation survival after VF OHCA. Patients who are under to normal weight had a lower long-term survival and represent a high-risk population primarily due to noncardiac diseases.

Central obesity has been associated with adverse events in patients with cardiovascular disease (CVD). We undertook this study to evaluate predictors of overweight awareness, desire to lose weight and weight loss attempts in patients with CVD and central obesity. Association of physician-rendered diagnosis of overweight with weight loss attempts was also evaluated. Data from National Health and Nutrition Examination Survey (NHANES) 1999 –2004 were used. CVD was defined as self-referred history of coronary artery disease or stroke. Central obesity was determined based on a waist circumference > 102 cm in men and > 88 cm in women. Motivational factors considered as determinant for behavioral change that were evaluated included: participants’ awareness of their overweight status; desire to lose weight; any weight loss attempts in last year. We examined demographic, anthropometric and clinical determinants of each stage of change using multivariate models. Of the 1484 participants with CVD that were identified in NHANES, 824 had central obesity (56%). 76% of centrally obese participants were aware of their overweight status and 79% were desirous of losing weight. Despite this awareness and desire, only 41% of centrally obese had attempted weight loss in the last year. Only 61% reported that they had been informed that they were overweight by physicians. On multivariate analysis, after adjusting for sociodemographic factors and body mass index, physician rendered diagnosis of overweight was a significant predictor of weight loss attempts (OR=2.5, 95% CI 1.3– 4.9, p= 0.006). Awareness of one’s overweight status strongly predicted desire to lose weight in multivariate models (OR=107.6, 95 % CI 42.9 –270.0, p<.0001) and desire to lose weight was linked to recent attempts to lose weight (OR 10.4, 95 % CI 2.8– 39.2, p= 0.0005), confirming that the Transtheoretical Model of behavioral change applies to the study population. In a nationally representative sample of participants with CVD and central obesity, physician-rendered diagnosis of overweight emerged as an important predictor of weight loss attempts. Efforts to promote weight loss should focus on interventions based on Transtheoretical Model of behavioral change.