New York Heart Association Class and Kansas City Cardiomyopathy Questionnaire in Acute Heart Failure

This cohort study evaluates concordance between New York Heart Association class and Kansas City Cardiomyopathy Questionnaire overall summary score among patients with acute heart failure in China.


Introduction
The New York Heart Association (NYHA) classification has been widely used for heart failure (HF) symptom severity assessment since the 1920s. 1 It is ingrained in guidelines and clinical practice, and NYHA class has served as the benchmark for risk stratification, trial enrollment, and treatment candidacy determination. 2 However, there has been recognition of its potential limitations (eg, marked interobserver variability and limited reproducibility), 3,4 and its prognostic relevance has been questioned increasingly. 5,6Since the inception of NYHA class, patient-reported outcome (PRO)   measures have evolved substantially.9][10][11] However, comparisons between the KCCQ and NYHA classification among patients with acute HF in clinical practice remain unknown.Few studies have reported the prognostic value of PROs compared with NYHA and their conflicts, and the results have been affected by some limitations, such as inclusion of patients exclusively with chronic HF with reduced ejection fraction, 6 short-term outcomes, and use of a general health survey rather than an HF-specific instrument (eg, KCCQ). 12In 2015, a shorter version of KCCQ, KCCQ-12, reduced the original 23-item questionnaire to 12 items, taking half the time to complete and increasing feasibility, particularly in acute care settings.Given the rapid progress of symptom severity in HF, it would be informative to compare the concordance in NYHA class and KCCQ-12 and the long-term outcome values associated with their early changes.
Accordingly, using data from a nationwide cohort of patients with acute HF, we aimed to (1)   characterize the level of concordance between admission NYHA class and KCCQ and assess correlations between them, (2) describe patterns of changes in NYHA class and KCCQ over time, and (3) compare the association of changes in NYHA class and KCCQ for long-term outcomes.

Study Design and Population
We included patients who were enrolled in the China Patient-Centered Evaluative Assessment of Cardiac Events prospective HF study and had NYHA class and KCCQ data at admission and 1-month follow-up.The study enrolled patients hospitalized for HF between August 2016 and May 2018 from 52 diverse hospitals located in 20 provinces. 13Patients hospitalized for HF in these hospitals were consecutively registered if they were aged 18 years or older.Those who signed the informed consent were enrolled and followed up at 1, 6, and 12 months after discharge and annually thereafter until March 2022.The ethics committee of Fuwai Hospital and the local ethics committees at sites approved the study.The study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guidelines. 14

Data Collection
Patients' demographics, socioeconomic status, depressive symptoms, and cognitive function were collected by standardized questionnaire through interviews in person during hospitalization.Clinical characteristics (eg, systolic blood pressure [SBP] and heart rate), comorbidities, and treatments were obtained from the medical records of the index hospitalization.Left ventricular ejection fraction (LVEF) was uniformly measured during hospitalization, and patients were categorized into HF with reduced ejection fraction (LVEF Յ40%), HF with mildly reduced ejection fraction (LVEF 41%-49%), and HF with preserved ejection fraction (LVEF Ն50%).Laboratory tests at admission, including N-terminal pro-B type natriuretic peptide (NT-proBNP), creatinine, sodium, and potassium were analyzed at the central laboratory.The estimated glomerular filtration rate (eGFR) was calculated according to the adaptation of the Modification of Diet in Renal Disease. 15Depression was evaluated by the Patient Health Questionnaire-2, 16 and cognitive function was assessed by the Mini-Cog test. 17

NYHA Class and HF-Specific KCCQ Assessment
NYHA class was a prespecified 4-tier schema of class I to IV and assessed by clinicians at admission and 1 month.NYHA class IV indicates the worst functional status in HF and vice versa. 7The HF-specific PRO was measured within 48 hours of index admission and 1 month by KCCQ-12, which quantifies 4 health status domains: symptom frequency, physical limitations, social limitations, and quality of life. 11The KCCQ overall summary score (KCCQ-OS) was generated from the previously listed domains, ranging from 0 (worst) to 100 (best).

Definition of Concordance and Discordance
We defined concordance and discordance through an established framework using cut points to categorize NYHA class and KCCQ-OS.Each scale was categorized into 4 levels from worst to best, and cut points for NYHA class (class IV, III, II, I) and KCCQ-OS (<25, 25-49, 50-74, 75-100) were chosen to be consistent with previous literature. 6Concordance was defined as having NYHA class and KCCQ-OS at the same level among these categories.Mild and moderate to severe discordance were defined by the level of NYHA class and KCCQ-OS differing by 1 level or 2 or more levels, respectively.

Outcomes
The main outcome was 4-year all-cause mortality.Other outcomes included 1-year all-cause mortality and the composite of cardiovascular (CV) death or first rehospitalization for HF.CV death was defined as sudden cardiac death, or death due to HF, cerebrovascular events, coronary heart disease, or other CV causes.Death events and underlying reasons were collected from death certificates, patient's relative interviews, or the national database of death causes, then centrally adjudicated by trained clinicians.

Statistical Analysis
Patients' characteristics stratified by level of discordance between admission NYHA class and KCCQ-OS (concordance, mild discordance, and moderate to severe discordance) were compared using the Kruskal-Wallis tests for continuous variables and χ 2 tests for categorical variables.The difference between groups was calculated by standardized mean difference (SMD), and absolute values less than 0.1 were considered small differences.
Kernel density estimations were used to describe the distribution of KCCQ-OS by NYHA class.
Silverman rule-of-thumb was applied for bandwidth selection. 18The overlap between classes was defined as the area of the intersection divided by the total area under both curves to quantify the similarity between distributions. 19We used Spearman correlation to compute the correlation coefficients between NYHA class and KCCQ with its domains at admission and 1 month.
Changes in NYHA class and KCCQ-OS between admission and 1 month were summarized in categorical and continuous analyses.Considering that a change of 5 or more points in KCCQ-OS is clinically significant, changes in KCCQ-OS were categorized into 10 or more points decline, 5 to 9 points decline, no significant change (<5-point change), 5 to 9 points improvement, and 10 or more points improvement.A logistic regression model was used to determine patient factors associated with specific discordance directionality between NYHA class and KCCQ-OS at admission and 1 month.
Model selection was based on backward elimination, and variables with a P value greater than .05were removed.
Cox proportional hazard models were used to separately evaluate the associations of improvements in NYHA class and improvements of 5 or more points in KCCQ-OS with outcomes.We adjusted the following variables in the models: age, sex, educational attainment, employment, smoking, depression status, cognitive function, SBP, NT-proBNP, eGFR, serum sodium, potassium, LVEF subtypes, atrial fibrillation, diabetes, chronic obstructive pulmonary disease (COPD), myocardial infarction, stroke, anemia, prior HF, cardiac resynchronization therapy (CRT), implantable cardioverter-defibrillator (ICD), and postdischarge medications.A sensitivity analysis was performed to evaluate whether improvements of 10 and 20 points in KCCQ-OS, which were considered  20 were associated with outcomes.We conducted subgroup analyses to examine the consistency of the associations.The subgroup parameters included age, gender, HF subtype (ie, new-onset HF or acute decompensated chronic HF), LVEF, NT-proBNP, eGFR, atrial fibrillation, diabetes, and COPD.

JAMA Network Open | Cardiology
In this study, missing variables ranged from 0% to 4.73% (ie, eGFR, NT-proBNP, LVEF, sodium, and potassium) as presented in eTable 1 in Supplement 1 and were imputed using multiple imputations by the Markov chain Monte Carlo method.All comparisons were 2-sided, and statistical significance was defined as P < .05.We performed statistical analyses using SAS version 9.4 (SAS Institute).Analysis was conducted from January to March 2023.

Patient Characteristics
In total, 2683 patients were included in this analysis (eFigure 1 and eTable 2 in Supplement 1

Correlations Between NYHA Class and KCCQ Measurements
The distributions of KCCQ-OS levels stratified by admission and 1-month NYHA class are presented in Figure 1.At admission, for KCCQ-OS, kernel density overlaps were 73.6% between NYHA II and III, 63.8% between NYHA II and IV, and 88.3% between NYHA III and IV.At 1 month, overlaps between NYHA class for KCCQ-OS varied from 20.9% (NYHA I vs IV) to 76.2% (NYHA I vs II).The correlations between NYHA class and KCCQ-OS with its domains were statistically significant at admission and 1 month.The magnitude of the correlation between NYHA class and KCCQ-OS was modest (r = 0.26) at admission and higher (r = 0.54) at 1 month (all P < .001)(eTable 7 in Supplement 1).

Change in NYHA Class and KCCQ From Admission to 1 Month
From admission to 1 month after discharge, 1924 patients (71.7%) experienced any improvement in NYHA class, 670 (25.0%) had no change, and 89 (3.3%) had any decline.For KCCQ-OS, the most common change was a 10-point or more improvement in 1699 patients (63.3%).A total of 204 patients (7.6%) had improvements of 5 to 9 points, 349 patients (13.0%) had no significant change, and 431 patients (16.1%) experienced a decline of 5 or more points (Figure 2).

Associations of Change in NYHA Class and KCCQ With Outcomes
There were 1057 (39.4%) deaths during the 4-year follow-up, of which 275 patients (10.2%) died within 1 year after discharge, and 958 (35.7%) had 1-year CV death or HF rehospitalization.In unadjusted analysis, any improvement in NYHA class and a 5-point or more improvement in KCCQ-OS were both associated with lower risks of all-cause mortality and the composite of CV death or HF rehospitalization.
After adjustment for covariates, there was no significant association between improvement in NYHA class and 4-year all-cause mortality (hazard ratio [HR], 0.91; 95% CI, 0.79-1.04),whereas an improvement of 5 or more points in KCCQ-OS was independently associated with a 16% lower risk of 4-year mortality (HR, 0.84; 95% CI, 0.74-0.96).The estimation of the composite outcome of CV death and HF hospitalization was greater for an improvement of 5 or more points in KCCQ-OS (HR, 0.64; 95% CI, 0.56-0.73)than for an improvement in NYHA class (HR, 0.81; 95% CI, 0.71-0.94).
The sensitivity analyses showed similar associations of improvement of 10 or more or 20 points in KCCQ-OS with outcomes (eFigure 2 in Supplement 1).No significant heterogeneity was observed in the associations of improvement in NYHA class and KCCQ-OS with 4-year mortality in most subgroups as shown in eFigure 3 in Supplement 1, including age, sex, HF subtype, LVEF group, eGFR, atrial fibrillation, diabetes, and COPD.To convert potassium to milliequivalents per liter, multiply by 1.0; to convert sodium to milliequivalents per liter, multiply by 1.0 a SMDs less than 0.10 are considered small.

Discussion
In this contemporary nationwide cohort study of patients with acute HF, we found that about two-thirds of patients exhibited discordance between the NYHA class and KCCQ scale upon admission, with the vast majority of discordances related to a disproportionally better KCCQ-OS.
Most patients experienced improvements in NYHA class and KCCQ-OS at 1 month after discharge, with substantial variability of KCCQ across NYHA class.KCCQ-OS improvement was independently associated with lower risks of all-cause mortality and the composite CV death or HF rehospitalization, whereas the change in NYHA class was not associated with 4-year mortality risk.
To our knowledge, this is the first study to provide a comparison between NYHA class and KCCQ-12 in patients with acute HF.Moderate to severe discordance between the NYHA class and KCCQ-OS was common in one-fifth of patients, and KCCQ was more sensitive for detecting changes in health status and estimating outcomes.One prior study 6 reported that KCCQ's prognostic value conflicted with NYHA class, but the evidence of comparisons was generated from stable patients with chronic HF with reduced ejection fraction.Another study 12 explored the discordant relationship in the ASCEND-HF trial but used a general health survey (EQ-5D), and it was limited to short-term outcomes (ie, 30-day and 180-day mortality) and nonsimultaneous assessments (eg, NYHA was documented before decompensation whereas EQ-5D was reported after admission).Greater attention should be paid to the fact that symptoms reported by clinicians and patients often differ, and it may affect clinical decisions for optimal therapeutic options.
The present study was further strengthened by evaluating the changes in NYHA class and KCCQ status in long-term outcomes using data from a clinical practice-based acute HF population in China.
Health status measurements for prognostication purpose have been reported in US and European trials, 3,21,22 yet data on comparing NYHA and KCCQ changes among patients with HF in a clinical population setting are not well characterized, especially in developing countries where health status might vary substantially by socioeconomic status and social class. 23,24Additionally, considering symptoms change rapidly in acute HF, it would be of great practical importance to understand the  patterns of early change in functional class after events and the impact of health status fluctuation on subsequent risk of events.Our data showed that improvements in KCCQ-OS were independently associated with the sheer magnitude of the 47% decreased risk of 1-year mortality and 16% decreased risk of 4-year mortality, which was amplified by the lack of a significant association between NYHA change and 4-year survival.Yet, despite a decade of advocacy for these measurements, health systems have been hesitating to adopt KCCQ within HF standard care due to various barriers. 25,26A recent trial of outpatients with HF showed that sharing KCCQ results with physicians enhanced the accuracy of health status assessment and boosted patients' perception that clinicians understood their symptoms better. 27Integrating the clinician-assigned NYHA class and patient-reported KCCQ in routine care can lay the foundation for more efficient patient-centered care and improve health care quality.
The mechanisms underlying the disassociations between NYHA class and KCCQ scale remain unclear.Despite significant correlations, meaningful changes in KCCQ-12 were not reliably reflected in NYHA class.Possible explanations include interphysician variability in assessing NYHA class and different emphasis in various measurements.Similar conflicts between clinician-reported outcomes and PROs have been reported in other disease settings. 6,28,29For instance, Arnold and colleagues 28 found 42% of patients with coronary disease had more frequent angina than those recognized by physicians, and a marked variation of 0%-86% in underrecognition across physicians was observed.
Moreover, patients may be more willing to provide details on their symptoms when using selfreported questionnaires to ensure adequate attention is given.In many cases, clinicians tend to consider that patients are in less pain than they reported after taking patients' medical history. 30,31ditionally, our study identified several patient factors, including sex, employment, marriage, psychological factors, and comorbidities such as COPD that were associated with disagreement between clinician-reported and patient-reported outcomes.Prior studies demonstrated that women reported poorer health status than men, and it is plausible that the gender gap may be attributed to factors such as biological, sociobehavioral, and psychological inequalities, and women may have more chronic conditions that are nonfatal but are debilitating to be linked with self-reported health problems. 32,33Likewise, current findings of employment and marriage status associated with discordance may suggest societal factors contributing to the clinician vs patient interpretation of

Figure 1 .
Figure 1.Distribution of KCCQ-OS by NYHA Class at Admission and 1 Month

Figure 2 .
Figure 2. Change From Admission to 1 Month in NYHA Class and KCCQ-OS

Table 1 .
Admission Characteristics by the Degree of Admission Discordance Between NYHA and KCCQ-OS (continued) Unadjusted and adjusted hazard ratios indicate the risk of clinical outcomes associated with improvement in NYHA class and KCCQ-OS score.Models were adjusted for age, sex, educational attainment, employment, smoking, depression status, cognitive function, systolic blood pressure, N-terminal pro-B type natriuretic peptide, estimated glomerular filtration rate, serum sodium, serum potassium, left ventricular ejection fraction subtypes, atrial fibrillation, diabetes, chronic obstructive pulmonary disease, myocardial infarction, stroke, anemia, prior heart failure, cardiac resynchronization therapy, implantable cardioverterdefibrillator, and postdischarge medications.CV indicates cardiovascular; HF, heart failure; KCCQ-OS, Kansas City Cardiomyopathy Questionnaire overall summary; NYHA, New York Heart Association.34.Lin MH, Chen LJ, Huang ST, et al.Age and sex differences in associations between self-reported health, physical function, mental function and mortality.Arch Gerontol Geriatr.2022;98:104537.doi:10.1016/j.archger.2021.10453735.Joseph SM, Novak E, Arnold SV, et al.Comparable performance of the Kansas City Cardiomyopathy Questionnaire in patients with heart failure with preserved and reduced ejection fraction.Circ Heart Fail.2013;6 (6):1139-1146.doi:10.1161/CIRCHEARTFAILURE.113.000359Proportions of Missing Data in Covariates eFigure 1. Flow Chart of Study Cohort Development eTable 2. Characteristics of Patients Included and Excluded From Current Study eTable 3. Patient Characteristics by NYHA Class at Admission eTable 4. Patient Characteristics by KCCQ-OS at Admission eTable 5. Agreement Between NYHA Class and KCCQ-OS at Admission and 1 Month eTable 6. Factors Associated With Specific Directionality of Discordance Between NYHA Class and KCCQ-OS Categories eTable 7. Correlations Among NYHA Class and KCCQ Measurements eTable 8. Association Between Changes in Each Domain of KCCQ-OS and Clinical Outcomes 2. Sensitivity Analysis of the Association Between Change in KCCQ-OS and Clinical Outcomes eFigure 3. Subgroup Analysis of the Association Between Change in NYHA Class and KCCQ-OS With 4-Year All-Cause Mortality