Anxiety Predicts Worse Cardiorespiratory Fitness Outcomes in Cardiac Rehabilitation for Lower Socioeconomic Status Patients

Abstract

Purpose:

Patients in cardiac rehabilitation (CR) with lower socioeconomic status (SES) have worse clinical profiles, higher rates of psychiatric difficulties, and show lower cardiorespiratory fitness (CRF) improvements from CR than their counterparts with higher-SES. Improvement in CRF during CR predicts better long-term health outcomes. Research suggests that higher anxiety impairs CRF in structured exercise regimes and is overrepresented among patients with lower SES. However, no study has determined whether this relationship holds true in CR.

Methods:

This study is a secondary analysis of a randomized controlled trial to improve CR attendance among patients with lower SES. Anxiety (ASEBA ASR; Achenbach System of Empirically Based Assessment, Adult Self Report), and CRF measures (metabolic equivalents of task [METs_peak ]) were collected prior to CR enrollment and 4 mo later. Regression was used to examine the association of anxiety with CRF at 4 mo while controlling for other demographic and clinical characteristics.

Results:

Eight-eight participants were included in the analyses, 31% of whom had clinically significant levels of anxiety (T ≥ 63). Higher anxiety significantly predicted lower exit CRF when controlling for baseline CRF, age, sex, qualifying diagnosis, and number of CR sessions attended (β = −.05, P = .04). Patients with clinically significant levels of anxiety could be expected to lose over .65 METs_peak in improvement.

Conclusions:

The results from this study suggest anxiety, which is overrepresented in populations with lower-SES, is associated with less CRF improvement across the duration of CR. The effect size was clinically meaningful and calls for future research on addressing psychological factor in CR.

CONDENSED ABSTRACT

Patients in cardiac rehabilitation (CR) with anxiety have worse clinical profiles, lower rates of attendance, and higher rates of smoking. We demonstrate that anxiety is associated with lower cardiorespiratory fitness improvement over the duration of a CR prescription. The results of this study highlight the need for psychological care in CR alongside traditional care.

For patients who have experienced a cardiac event, such as a myocardial infarction, or undergone a cardiac-related procedure, such as bypass surgery, the most efficacious approach to reduce future morbidity and mortality is to attend Phase II cardiac rehabilitation (CR). Cardiac ehabilitation is an outpatient, secondary prevention program that targets modifiable risk factors of cardiovascular disease (CVD). Cardiac rehabilitation is an evidence-based treatment regime that has received the highest level of recommendation (Class 1A) from the American Heart Association due to overwhelming evidence of its efficacy.^1,2

Many of the benefits of CR are conferred through individualized, progressive exercise regimes that improve cardiorespiratory fitness (CRF) of patients. Cardiorespiratory fitness improvement in cardiac rehabilitation is often measured in Metabolic Equivalent of Task scores (METs_peak) taken during exercise tolerance tests. The METs_peak scores are a relative unit of energy expenditure, where 1 MET is representative of sitting at rest, 5 METs represents moderate activity, and 7 METs is equivalent to vigorous activity.³ Improvements in CRF during CR are highly correlated with future morbidity and mortality. For instance, an improvement of 1 MET during exercise tolerance tests from the beginning to end of CR is associated with up to a 42% reduction in all-cause mortality rates in patients with heart failure.⁴

Unfortunately, not all patients experience CRF improvement from CR, for example females and patients with lower socioeconomic status (SES) show markedly lower CRF improvement than males⁵ and those with higher SES (as measured by educational attainment).⁶ One possible driver for these differences in CRF improvements are psychosocial factors, such as anxiety and depression, which can impact engagement with CR and an are often overrepresented in females and those of lower SES.^6,7,8

Several psychosocial variables deserve more scrutiny due to their prevalence in CR, severity among patients with CVD, and for sharing potential bidirectional relationships with CRF. Among CR patients, clinically important levels of depression (32%),⁹ and anxiety (28%),¹⁰ are higher than population norms. Depression is perhaps the best studied psychiatric problem among CR patients, several studies have identified strong associations between depression and CRF in CR patients.^9,11

Unlike depression, the effects of anxiety on CRF improvement are understudied. Anxiety is a trait or state emotion, characterized by elevated levels of arousal, feelings of dread, fear, worry, overreaction, and somatic discomfort.¹² Fears of precipitating a cardiac event through exercise have been reported in patients with high levels of anxiety and are hypothesized to result in exercising at lower intensities or even discontinuing CR entirely.¹³ Further, heightened sensitivity to physical discomfort due to anxiety may also decrease performance during exercise within CR.¹³ Accordingly, several pathways exist through which anxiety could impede CRF improvement.

Relationships between anxiety and CRF improvement have not yet been studied extensively within CR. In exercise studies more generally, it has been suggested that anxiety is a predictor of drop out from exercise programs and that the comorbidity of anxiety along with depression may attenuate the benefits of exercise .^14,15 Specifically within CR, one study¹⁶ found evidence indicating that lower CRF improvements occur in patients who have high levels of fear associated with exercise. Precedence also exists for an association between anxiety and lower CRF improvement during CR.¹⁶ Research on exercise in other clinical areas, such as sports injury rehabilitation, indicate that high levels of anxiety may impair performance during exercise tasks and thus reduce CRF gains during structured exercise.^15,16,17,18 Fear of exercising on an injured body-part, analogous to inducing stress on cardiovascular systems, may be the primary mechanism for reduced performance. In CR, a similar effect may impact the intensity patients are encouraged to exercise at, as > 60% of practitioners reported in one study that they would reduce exercise intensity below recommendations in response to patient fears.¹³ These links warrant further examination among CR patients.

Issues of higher anxiety and less improvement in CRF are more prevalent in certain populations of patients. Patients with lower-SES status are one of these groups. These individuals have higher rates of psychological symptoms such as depression and anxiety,^,19,20 and may have lower levels of CRF improvement following completion of CR.⁶ These characteristics make a sample of individuals with lower-SES status ideal for examining potential links between anxiety and CRF improvement.

Accordingly, the current study examined the relationship between anxiety and CRF improvement in a sample of patients with lower-SES, as we suspected that any potential relationships shared between anxiety and CRF improvement would be especially potent in this population. We hypothesized that higher anxiety would predict lower CRF improvement during CR.

METHODS

Data for analyses were taken from a controlled randomized clinical trial to improve CR participation among patients with lower-SES. Detailed methods and primary results of this trial are presented elsewhere.²⁰ Briefly, the primary objective of the parent study was to determine whether providing financial incentives contingent on attendance increased participation in, and continued attendance at, CR within a sample of patients with lower-SES, defined by insurance type.

Patients with a CR-eligible diagnosis (myocardial infarction, coronary bypass surgery, coronary stent placement, chronic heart failure with reduced ejection fraction <35%) who were enrolled in Medicaid or similar state-supported insurance program were recruited shortly after diagnosis and prior to CR enrollment from inpatient and outpatient settings. Participants (n=130) were randomized 1:1 to usual care or to the intervention (Figure 1). Participants randomized to the intervention received incentives, in the form of cash, for completion of individual CR sessions on an escalating schedule. Participants in both conditions completed assessments prior to CR participation and again 4 mo later. Assessments consisted of an exercise tolerance test, detailed sociodemographic questions, medical history, mood disorder questionnaires, and an executive function and cognitive battery.

Figure 1.

Cohort Diagram

No significant differences across demographics (age, race/ethnicity, sex, anxiety, depression, entrance CRF, exit CRF, education level, diagnosis or medical condition) were detected between intervention and non-intervention groups. Patients randomized to the incentive condition completed more sessions (22.4 vs. 14.7, respectively; P= .013) and were almost twice as likely to complete CR (55.4 vs. 29.2%, respectively; P = .002) as controls. Patients who received incentives were also more likely to experience improvements in executive function (P < .001), although there were no significant effects on other secondary outcomes. This study did not assess the effects of anxiety on attendance or CRF improvement.

Anxiety was measured through a self-report assessment, the Achenbach System of Empirically Based Assessment’s Adult Self Report (ASR), and the corresponding scale for those ≥ 60 yr, the Older Adult Self Report (OASR).^21, The ASR/OASR are standardized assessments with subscales for anxiety and depression. Scores were normed by sex and age and calculated as T-scores, where a score of ≥ 63 can be considered impaired.²²

Peak METswas used as the measure of CRF in this study and was assessed through symptom-limited treadmill exercise tolerance tests (ETT). We assessed METs_peak prior to entry and upon 4 mo after entry into CR. Patients performed a symptom-limited treadmill ETT until volitional exhaustion, or medical findings that warranted termination of the test. The METs_peak was defined as the workload performed on the last completed stage of ETT. ²⁴ In the parent study, exit METs_peak scores were acquired for patients whenever possible, regardless of whether they had dropped out of CR.²⁰

Statistics

Variables were checked for skew and kurtosis, but no distribution warranted an adjustment in analyses. Data were presented as mean ±SD.

Linear regression analyses, with ASR/OASR T score as independent variable, were used to test for statistically significant effects of anxiety on CRF at exit, with METs_peak at exit as the dependent variable. Clinically important variables which were considered relevant to be evaluated as potential covariates in analyses included CRF at entry, age, sex, diagnosis, CR sessions attended, and smoking status at entry. Continuous variables were checked for Normality, and none were found to be significantly non-Normal.

Univariate linear regression analyses were conducted to test for significant associations between each potential covariate and ASR/OASR T score and METs_peak at exit. Variables with significant (P < .05) associations with ASR/OASR T score or METs_peak at exit were included, along with ASR/OASR T score, as independent variables in an initial multivariable linear regression model, with METs_peak at exit as the dependent variable. A backward stepwise elimination procedure was then used to produce the final multivariable model with all independent variables having a significant (P < .05) association with METs_peak at exit, in order to determine which variables were significant, independent predictors of CRF at exit.

RESULTS

Correlation matrices were drawn to establish covariates, shown in Table 1. Higher CRF at exit was significantly associated with higher entrance CRF (r = 0.81, P<.01), lower anxiety scores (r = −0.23, P<.05), lower age (r = −0.41, P<.01), male sex (male M = 7.06, female M = 5.59, P<.05), and having a nonsurgical diagnosis (surgical = 5.03, nonsurgical = 6.28, P<.05). Higher anxiety scores were also associated with being female (female M = 62, male M = 58, P<.05). Attendance was not associated with anxiety in the control or trial arms.

Table 1.

Participant Characteristics

Sex
	Male	55 (62.5)
	Female	33 (37.5)
Procedure
	Surgical	16 (18.2)
	Nonsurgical	72 (81.8)
Smoking Status
	Current	28 (32.2)
	Non/Former	59 (67.8)
Anxiety Symptom score(ASR/OASR)		59.5±8.6
Anxiety Symptom Category(ASR/OASR)
	Normal <63	61 (69.3)
	Clinically Significant 63+	27 (30.7)
Diagnosis
	MI	40 (45.0)
	PCI/Stent	53 (60.0)
	CABG	11 (12.4)
	Valve Repair/Replacement	11 (12.4)
	CHF	7 (8.0)
	CAD	6 (6.7)
	Stable Angina	1 (1.1)
Race/Ethnicity
	American Indian	2 (2.2)
	Asian or Pacific Islander	1 (1.1)
	Black or African American	2 (2.2)
	Hispanic	0 (0.0)
	White	86 (96.6)
Highest Level of Education
	Did not finish High School/No GED	0 (0.0)
	GED	11 (12.4)
	High School Diploma	16 (18.0)
	Some College, No Degree	7 (7.9)
	2yr Degree	5 (5.6)
	4yr Degree	3 (3.4)
	Advanced Degree	4 (4.5)
Age, yr		58.7 ± 8.9
Entrance Cardiorespiratory Fitness, METspeak		5.1± 2.8
Exit Cardiorespiratory Fitness, METspeak		6.5± 3.2
CR Sessions Completed		24± 14.6

Data presented as n (%) or mean ± SD.

Abbreviations: ASR/OASR, Adult Self Report/Older Adult Self Report; CABG, ; CAD, ; CHF, ; CR, cardiac rehabilitation; GED, ; METs_peak, peak metabolic equivalent of tasks score; MI, ; PCI .

Thus, CRF at entry, sex, age, and diagnosis were included in both the analyses of anxiety on CRF outcomes. Sessions attended was also included, as number of sessions attended is a standard covariate in all similar literature within CR.

Main Outcomes

Of the 130 patients in the original study, 90 patients completed exit CRF assessments. Anxiety measures were missing for 2 patients, resulting in 88 patients in the final sample (Figure 1). Characteristics of the 88 patients with complete CRF and anxiety data is presented in Table 1. On average, patients in the lower-SES sample were young (59 yr) and achieved low METs_peak at entrance (5.1) and at exit (6.5). Patients had high rates of smoking (32%) and roughly two-thirds of the sample was male. Thirty-one percent of patients had problematic anxiety scores by the ASR/OASR anxiety subscale.

Higher CRF at exit was significantly associated with higher entrance CRF (r = 0.81, P<.01), lower anxiety scores (r = −0.23, P<.05), lower age (r = −0.41, P<.01), male sex (male M = 7.06, female M = 5.59, P<.05), and having a nonsurgical diagnosis (surgical = 5.03, nonsurgical = 6.28, P<.05). Higher anxiety scores were associated with being female (female M = 62, male M = 58, P<.05). There was no relationship between anxiety and patient attendance, as a whole, or within the experimental and control conditions.

Initial independent linear regression with each potential covariate on anxiety and CRF exit scores are shown in Table 2. Sex was the only variable significantly related to ASR/OASR scores at entry. Age, sex, diagnosis and CRF at baseline were all significantly related to METs_peak at exit. These variables were entered into a multivariable linear model, and selected via backwards stepwise elimination.

Table 2.

Univariate Regression for METspeak at Exit and ASR/OASR at Entry

Predictor	Exit METspeak Std. Beta	Entry ASR/OASR Std. Beta
Age	−.406b	−.061
Sexa	−.218a	.198a
Diagnosisb	.209a	−.043
Sessions attended	.191	−.02
Smoking statusc	.047	−.094
METspeak at Entry	.810b	−.084
Incentive Conditiond	−.032	−.063
Exit METspeak	NA	−.228a
ASR/OASR	−.228a	NA

^aP < .05

^bP < .01

Regression Standardized Beta coefficient.

Abbreviations: ASR, Adult Self Report; CR, cardiac rehabilitation; METs_peak, peak metabolic equivalent of tasks score; OASR, Older Adult Self Report.

^cCoded Male = 0, Female = 1

^dCoded Nonsurgical intervention = 0, Surgical intervention = 1

^eCoded No smoking = 0, Smoking = 1

^fCoded No intervention = 0, Intervention = 1

The initial multivariate model revealed no significant associations between METs_peak at exit and any of our predictor variables aside from METs_peak at entry (β = 0.89, t(86) = 11.03, P < .01), potentially due to significant multicollinearity. The overall model was significant and accounted for 69% of the variability in METs_peak exit scores. Backward, stepwise selection was then used to narrow down the model using all the above covariates. This procedure resulted in the removal of surgical diagnosis, sex and age from the model due to nonsignificant change in R². This resulted in a model where only ASR/OASR scores and baseline METs_peak scores remained significant.

The final model included only METs_peak at entry and ASR/OASR scores (Table 3.). The model was significant (R² = .67, ΔF(2,86) = 87.69, P < .001), and accounted for 67% of the variability in METs_peak scores at exit. Both variables were significantly related to exit CRF. A one-point increase in METs_peak at baseline was associated with a .94 increase in METs_peak at exit (β = 0.94, t(86) = 12.72, P < .001). Every increase in an ASR/OASR score was associated with a decrease of .048 METs, (β = 0.048, t(86) = −2.00, P < .05).

Table 3.

Anxiety as a Predictor of Exit Fitness

Predictor	Coefficient	S.E.	t	P
ASR/OASR	−0.05	0.022	−2.04	.045
Entry METspeak	0.94	0.074	12.72	.001

Abbreviations: ASR/OASR, Achenbach System of Empirically Based Self Assessment Adult/Older Adult Self Report; METs_peak, peak metabolic efficiency of tasks score.

We conducted further tests of anxiety improvement to determine whether sample characteristics may interact with improvement, and to determine whether anxiety improved significantly in both the general population and amongst those with clinically concerning levels of anxiety (ASR/OASR >62) by the end of CR. None of our covariates were significantly associated with ASR/OASR improvement, including sessions attended or treatment condition. We also found no significant improvement in ASR/OASR scores in the general sample, however among those with clinically concerning levels of anxiety upon entry, the average patient improved 4.3±8.49 points on the screener by the end of CR. No significant associations between anxiety improvement and CRF improvement were found.

DISCUSSION

In this sample of patients with lower-SES, higher anxiety was significantly associated with detriments in CRF improvement; an increase of a single ASR/OASR T-score was associated with a decrease of 0.05 METs by the end of CR. While this may seem like a small difference, a patient with a clinically concerning anxiety score (T = 63) would be expected lose out on two thirds of a MET (0.65) in CRF improvement compared to a patient with normal symptoms of anxiety (T = 50). Unfortunately, these levels of anxiety were not uncommon in this sample. Thirty-one percent of patients in this sample exceeded the clinically significant cutoff (63+ ASR/OASR).

Epidemiological studies have shown that even small improvements in CRF are associated with sizeable reductions in all-cause mortality for CR patients.^23,24 One study found that an improvement of a single MET after a 12-wk CR program was associated with a 34% decrease in mortality at 1-yr follow up, 28% after adjusting for age.²⁵ Other studies have found even greater effects. Martin et al., (2013) found a 30% decrease in mortality per MET increase among those with low baseline CRF and Keteyian et al., (2018) found that a 1 MET improvement correlated with a 42% reduction in all-cause mortality and a 38% lower adjusted risk for heart failure hospitalization.²⁶,⁴ From any of these findings, a loss of 0.65 METs in improvement may have significant implications for patients’ risk of future morbidity and mortality.

There are several pathways through which anxiety may impair CRF improvement. Prior literature indicates that anxiety may be impairing performance levels of patients during CR exercise sessions, or may be inhibiting physical activity patients engage in at home.^14,16 Our study showed that anxiety was not associated session attended, and the literature is currently mixed on whether attendance is consistently predicted by anxiety. A more likely mechanism is that high levels of anxiety reduce patient performance during CR exercise sessions. Patients with high levels of anxiety are more vigilant and sensitive to symptoms of discomfort,²⁷ experienced during exercise and may thus be attenuating their performance during CR sessions to alleviate heightened discomfort. This may result in lower overall improvements in CRF after completion of CR. Research by Farris et al., in 2019 provided corroborating evidence, over 60% of clinicians reported reducing exercise intensity in CR in response to patients exercise related anxiety.¹³ Further, individuals with anxiety disorders have lower rates of exercise participation and physical activity compared to non-anxious counterparts,²⁸ so patients with high levels of anxiety may not be appropriately supplementing CR exercise with physical activity at home.

Literature is currently mixed on the effects of CR on anxiety, our results may highlight some nuance. For example, the average patient did not improve on anxiety but highly anxious patients displayed significant improvement. It should be noted that this may be a regression to the mean.

Future research should be conducted to examine relationships between anxiety and CRF improvement. Additionally, randomized controlled trials should be conducted to assess whether ameliorating anxiety can lead to improvements in CRF. For example, therapies such as interoceptive exposure have shown success in reducing state-based anxiety experienced during panic attacks,^29,30,31 and may be effective in reducing exercise related anxiety.

Some limitations of this study merit mention. This study was a secondary analysis of a clinical trial and thus was not designed to test associations between symptoms of anxiety and CRF improvements. Similarly, not all patients in the original sample could be included in the current analyses, generally due to missing CRF data. Further, this sample was restricted to patients with lower-SES which may limit the generalizability of our findings. While this population is in need of special attention, further research should be done to see if psychological factors impact CRF improvement in the general population. Current recommendations for assessing CRF improvement in CR include using multiple measures of CRF to reduce non-responder bias. The use of the ASEBA ASR/OASR may also be a limitation, as it has not been used in exercise related trials. Our performance of multiple pairwise comparisons is also a limitation, however as findings other than our main analyses were exploratory, we did not perform corrections. We utilized METs_peak scores, rather than measures like peak oxygen uptake and heart rate due to limited number of observations available from the study drawn. We highly recommend further investigation using multiple measures of CRF improvement, including balance and strength measures alongside multiple CRF measures.

KEY PERSPECTIVES.

• This was the first investigation into whether anxiety impairs cardiorespiratory fitness improvement among patients in cardiac rehabilitation with lower socioeconomic status.

• This research warrants future investigation into brief therapies and treatment for anxiety in cardiac rehabilitation programs.