Patterns and predictors of dyspnoea following left ventricular assist device implantation 

Abstract

Aims

Dyspnoea is a common symptom of heart failure (HF) that often prompts patients to seek treatment. Implantation of a left ventricular assist device (LVAD) has been associated with reduced dyspnoea but it is unclear if all patients experience similar improvements in dyspnoea over time following LVAD implantation. Our aim was to identify distinct trajectories of dyspnoea symptoms over time following LVAD implantation and predictors of dyspnoea trajectory. We hypothesized that at least two, distinct trajectories of dyspnoea would be observed following LVAD implantation.

Methods and results

This was a secondary analysis of data from the Profiling Biobehavioral Responses to Mechanical Support in Advanced Heart Failure study. In the parent study, sociodemographic and clinical data were collected prior to LVAD implantation and at 1, 3, and 6 months following LVAD implantation from a sample (n = 101) of patients with advanced HF. Latent growth mixture modelling was performed to identify distinct trajectories of dyspnoea symptoms. Backwards stepwise logistic regression was used to identify predictors of dyspnoea trajectory. Two, distinct trajectories of dyspnoea symptoms were identified: sustained improvement and unsustained improvement. Participants who experienced sustained improvement (86.7% of sample) demonstrated large, significant improvement in dyspnoea from pre-implantation to 3 months post-implant followed by smaller, non-significant improvement from 3 to 6 months. Participants who experienced unsustained improvement (13.3% of sample) demonstrated initial improvement from pre-implantation to 3 months post-implantation followed by worsening of dyspnoea from 3 to 6 months. Greater depressive symptoms at baseline and living alone were significant predictors of unsustained improvement.

Conclusion

Patients experience different patterns of dyspnoea over time following LVAD implantation. Clinicians should inquire about living arrangements and depressive symptoms at each visit to determine risk of unsustained improvement in dyspnoea.

Introduction

Advanced heart failure (HF), defined as persistent HF symptoms despite optimized medical treatment, affects nearly half a million Americans and negatively affects quality of life.^1,2 Although heart transplantation is considered the ‘gold standard’ treatment for advanced HF, donor hearts are limited and not all patients with advanced HF are candidates for transplant.^3,4 Consequently, left ventricular assist device (LVAD) implantation has become a common treatment for patients awaiting heart transplantation or, if ineligible for transplant, as long-term LVAD support.^3,5,6 Implantation of an LVAD has been shown to reduce HF-associated symptoms.^7–9 Among the most bothersome of these symptoms is dyspnoea, which is reported by over half of patients with HF and is often the symptom that drives patients to seek urgent treatment.^10,11 In addition to being associated with greater healthcare utilization, dyspnoea is associated with worse outcomes, including higher mortality rates and worse health-related quality of life.^10,12

There is evidence that dyspnoea is reduced following LVAD implantation.⁸ In a sample of 101 patients with advanced HF, Lee et al.⁸ observed a significant decline in dyspnoea in the 6 months following LVAD implantation. However, it is unclear if all LVAD recipients experience similar patterns of reduced dyspnoea over time. According to the Theory of Unpleasant Symptoms, symptom distress is influenced by physiological, psychological, and situational factors.¹³ In the context of advanced HF treatment, patients may describe dyspnoea differently following LVAD insertion depending on differences in physiological factors (comorbid illness burden, sleep disturbances, and complications of LVAD implantation), psychological factors (cognitive status and depression/anxiety), and situational factors (whether one is partnered/not partnered). Each of these factors is highly prevalent in HF and each has been associated with symptom perception individually.^14–20 However, the associations between these factors and unique patterns of dyspnoea over time have not been described. Therefore, we sought to explore the existence of distinct patterns of dyspnoea from pre-implant to 6 months post-implantation in patients with a continuous flow LVAD and to identify clinical and contextual factors that predict patterns of dyspnoea over time.

Methods

Setting and sample

This was a secondary analysis of data collected as part of the Profiling Biobehavioral Responses to Mechanical Support in Advanced Heart Failure (PREMISE) study.²¹ All women and men >21 years of age with advanced HF who were attending an advanced HF clinic in the Northwest USA and were scheduled to receive an LVAD were asked to participate in the PREMISE study.^8,21 Individuals who were unable to understand 5th grade English or Spanish, had a history of major uncorrected hearing or vision impairment, or were diagnosed with cognitive dysfunction or major psychiatric illness (e.g. psychosis) were excluded.^8,21 Patients with previous/prior mechanical circulatory support or heart transplantation, or who were unable to participate for other reasons (e.g. terminal illness) also were excluded.^8,21 Eligibility was confirmed via electronic medical records review.^8,21 All participants recruited for the PREMISE study were included in this secondary analysis. The study was approved by the institutional review board at Oregon Health & Science University (IRB #7907) and all participants provided written informed consent upon enrolment in the PREMISE study.^8,21 The investigation conforms with the principles outlined in the Declaration of Helsinki.²²

Sample characteristics

Sociodemographic characteristics, including age, biological sex, race, and marital status were collected at baseline using a researcher-developed questionnaire designed to capture these data.²¹ Data on HF-associated indices, including New York Heart Association (NYHA) functional class, duration of HF, and left ventricular ejection fraction (LVEF), were collected from the medical record pre-implantation and updated at 1, 3, and 6 months post-implant.²¹

Dyspnoea

The dyspnoea subscale of the Heart Failure Somatic Perception Scale (HFSPS) was used to capture how bothered patients were by dyspnoea pre-implantation and at 1, 3, and 6 months post-implant.²³ There is evidence that dyspnoea improves in the first month following LVAD implantation,²⁴ so data collection points were selected to capture initial improvements in dyspnoea and changes over the next few months. The dyspnoea subscale has six questions to determine the burden of dyspnoea in the preceding week.²³ Responses range from 1 (had the symptom but was not bothered by it) to 5 (perceived the symptom as ‘extremely bothersome’).²³ A response of 0 indicates the participant did not have dyspnoea.²³ The total score is the sum of the scores on the six items with higher scores indicating greater dyspnoea burden.²³ The HFSPS was validated in the HF population,²³ demonstrated good internal consistency in the current analysis (Cronbach’s α = 0.89) and has been shown to be associated independently with clinical event risk in HF.

Predictors of dyspnoea

The burden of comorbid illness at pre-implantation was quantified using the Charlson Comorbidity Index. This weighted index accounts for the presence of a comorbid illness and the relative severity of the illness, with higher values indicating greater burden of comorbid illness.²⁵ The validity of the Charlson Comorbidity Index has been established in patients with cardiovascular disease.²⁶ Comorbid illness has been shown by others to be associated with dyspnoea and may influence patterns of dyspnoea following LVAD implantation.^15,27

Cognitive function was evaluated at each time point using serial versions of the Montreal Cognitive Assessment (MoCA). The MoCA evaluates several cognitive domains including memory, attention, and executive function.²⁸ Total score on the MoCA is the sum of the individual domain scores with a corrected maximum score of 30.²⁸ Scores below 26 are indicative of cognitive dysfunction.²⁸ To adjust for low levels of education, participants with fewer than 12 years of education are awarded two additional points.²⁸ The MoCA demonstrated reasonable internal consistency in the current analysis (Cronbach’s α = 0.77) and has been validated in the cardiovascular population.²⁹ In prior work, cognitive function has been associated with dyspnoea in HF.¹⁶

The 9-item Patient Health Questionnaire (PHQ-9) was used to capture depressive symptoms at pre-implantation and 1, 3, and 6 months post-implantation.³⁰ The PHQ-9 has 9 items, wherein respondents rate the severity of each symptom over the preceding two weeks on a scale ranging from 0 to 3.³⁰ Total score ranges from 0 to 27, with higher scores indicating more severe depressive symptoms.³⁰ Internal consistency reliability was high in this sample (Cronbach’s α = 0.81) and validity has been demonstrated in HF patients.³¹ Depressive symptoms have been shown by others to be associated with dyspnoea in HF and was therefore included as a candidate predictor.^15,32

Severity of anxiety was captured using the 6-item anxiety subscale of the Brief Symptom Inventory (BSI) pre-implantation and 1, 3, and 6 months post-implantation.³³ Respondents are asked to rate feelings of anxiety during the previous week on a scale of 0 to 4.³³ The summary scale (calculated by summing the ratings and dividing by the number of items in the subscale) ranges from 0 to 4 with higher scores indicating greater anxiety.³³ In this sample, the BSI anxiety subscale demonstrated good internal consistency reliability (Cronbach’s α = 0.84). Validity also has been demonstrated in the HF population.³⁴ Anxiety has been shown by others to be associated with dyspnoea in adults with HF and may influence patterns of dyspnoea over time.^15,32

The Epworth Sleepiness Scale (ESS) was used to capture daytime sleepiness at pre-implantation and 1, 3, and 6 months post-implantation.³⁵ The ESS is an 8-item measure in which participants are asked to rate how likely they are to fall asleep in a given situation with response options ranging from 0 (indicating they would never doze in that situation) to 3 (indicating high likelihood of dozing).³⁵ Scores on the eight items are summed to create the total score (ranging from 0 to 24) with higher scores indicating greater daytime sleepiness.³⁵ The ESS has been shown to be a valid measure of sleepiness³⁶ and Cronbach’s α for the ESS in this analysis was 0.87, indicating good internal consistency. Previous researchers have demonstrated associations between dyspnoea and daytime sleepiness in HF.³⁷

Statistical analyses

Means and standard deviations were calculated for continuous variables. Counts and percentages were calculated for categorical variables. Latent growth mixture modelling was performed to identify distinct patterns of change over time. Latent growth mixture modelling enables researchers to identify different patterns of change over time in unobserved (latent) groups and to evaluate differences in baseline values (intercepts) and patterns of change (slopes) over time between the unobserved groups.^38,39

Broadly, patients who receive an LVAD can be categorized as ‘responders’ (those who benefit from the therapy) and ‘non-responders’ (those who do not benefit from the therapy), so we hypothesized that at least two patterns of change would be observed. Several different patterns of change over time were considered; including linear, quadratic, and piecewise patterns. Piecewise patterns consisting of two, distinct phases of change have been reported previously following LVAD implantation, so evaluation of piecewise patterns of change was justified.^8,17 Two different piecewise models were evaluated to reflect the two possible piecewise trajectories over 6 months that have been demonstrated previously. In the first piecewise model, the first phase of change was from pre-implantation to 1 month post-implant and the second phase was from 1 month post-implant to 6 months post-implant. In the second piecewise model, the first phase of change was from pre-implantation to 3 months post-implant and the second phase was from 3 months to 6 months post-implant.

An initial model was generated for each trajectory pattern under consideration (linear, quadratic, and piecewise) to determine if the collected data were more likely to be representative of a single pattern of change or two distinct patterns of change. Subsequent models were generated by increasing the number of possible patterns of change sequentially. Parameters (intercepts and slopes) were generated for each pattern of change. Maximum likelihood estimation with robust standard errors was used to account for missing data. Fit statistics, including likelihood ratio tests, were generated with each iteration to compare the fit of each model to the fit of a model with one fewer pattern of change.

Several fit indices were used to determine model fit, including the percentage of participants assigned to each pattern of change, posterior probabilities, entropy, Bayesian Information Criteria, Vuong-Lo-Mendell-Rubin likelihood ratio, and the parametric bootstrapped likelihood ratio test.^38,39 Consistent with published criteria, models were deemed to have good fit if each pattern of change accounted for at least 5% of the participants, posterior probabilities were closest to 1.0, entropy was greater than 0.80 (with values closer to 1.0 indicating better fit), Bayesian Information Criteria was low, and likelihood ratio tests were statistically significant.^38,39 Effect sizes in the form of Hedges’ g were calculated to quantify the effect of LVAD implantation on dyspnoea over time.⁴⁰ Effect size calculations were based on changes in HFSPS dyspnoea scores in the best-fitting model. Consistent with convention: 0.2 = small effect, 0.5 = moderate effect, and 0.8 = large effect.⁴¹

Once the distinct patterns of change were determined, we identified potential predictors of a specific pattern of change. First, the pre-implantation characteristics of participants assigned to each pattern of change were compared using Fisher’s exact tests or Student’s t-tests to identify differences in baseline characteristics between those who demonstrated different patterns of change in dyspnoea over time. Sociodemographic characteristics (age, biological sex, and partnered/not partnered), disease-specific measures (NYHA functional class, LVEF, and goal of LVAD therapy), and comorbid conditions (Charlson Comorbidity Index, depression, anxiety, and daytime sleepiness) that previously have been associated with greater HF severity or abnormal symptom perception were considered.^17,42,43 Variables were selected for predictive models if the characteristic differed between groups at the pre-implantation time point at a significance level of P ≤ 0.20.^17,42,43 Additionally, the occurrence of neurological (mental status change/stroke), haematological (thrombosis/haemorrhage), or suction events (reduced filling of the pump) following LVAD implantation could influence LVAD effectiveness and patterns of dyspnoea over time, so each was included in the model if prevalence of the event differed between the groups at any time point at a significance level of P ≤ 0.20 or less.²⁰

Next, we used backwards stepwise logistic regression to identify significant predictors of the patterns of change over time and to generate a parsimonious model that was not saturated with non-significant factors. Candidate predictors were systematically removed from the model if associations with patterns of dyspnoea were not significant at the P ≤ 0.10 level or lower. Patient age was retained in all models as older age is theorized to be associated with lower dyspnoea.⁴² Odds ratios were generated comparing the odds of experiencing unsustained improvement in dyspnoea compared to the odds of sustained improvement as the reference category. Analyses were conducted using MPlus version 8 (Los Angeles, CA, USA) and Stata/MP version 15.1 (StataCorp, College Station, TX, USA). The corresponding author will provide data to qualified researchers for the purpose of replication upon reasonable request.

Results

Pre-implantation characteristics of the sample (n = 101) are provided in Table 1. The sample was primarily middle-aged, male, White, and partnered. Mean ejection fraction was low and 96% of the sample were classified as NYHA class III or IV.

The piecewise model that reflected two distinct patterns of change in dyspnoea, one from pre-implantation to 3 months post-implant and the second from 3 to 6 months post-implantation, had the best fit with these data (Table 2; Figure 1). Participants in the first pattern of change, that we labelled ‘sustained improvement’ (86.7% of the sample), experienced large and significant improvements in dyspnoea between pre-implant and 3 months following LVAD implantation (Hedges’ g = 1.61) followed by a smaller, non-significant rate of improvement between 3 and 6 months post-implant (Hedges’ g = 0.20). Participants in the second pattern of dyspnoea, that we labelled ‘unsustained improvement’ (13.3% of the sample), experienced large initial improvements in dyspnoea in the first 3 months following LVAD implantation (Hedges’ g = 1.21), but this was followed by a significant worsening of dyspnoea between 3 and 6 months after LVAD implantation (Hedges’ g = 1.24). Mean scores on the HFSPS dyspnoea subscale were not significantly different between the two groups at baseline (P = 0.07), however participants who demonstrated unstained improvement had significantly higher HFSPS dyspnoea scores at all other time points (P < 0.01).

Most baseline characteristics were similar between those who demonstrated sustained improvement and those who demonstrated unsustained improvement in dyspnoea (Table 1); however, history of chronic respiratory disease was more prevalent among those who demonstrated unsustained improvement in dyspnoea (46.2%) than among those who demonstrated sustained improvement (17.7%, P = 0.03). Also, neurological events were more common between 3 and 6 months post-implant among those who demonstrated unsustained improvement in dyspnoea (30.8%) when compared with those with sustained improvement (6.4%, P = 0.02). In addition to these significant findings, age, marital status, history of hypertension, PHQ-9 score, and prevalence of suction events in the first month following LVAD implantation were below the pre-specified model selection criteria of P = 0.20, therefore these variables also were included in prediction models.

In our backwards stepwise regression model (Table 3), participants living alone were more likely to demonstrate unsustained improvement in dyspnoea following LVAD implantation when compared to those living with a spouse or partner. Also, greater burden of depressive symptoms was associated with greater odds of unsustained improvement in dyspnoea. Specifically, the odds of being in the unsustained improvement group increased by 18% for each one-unit increase in PHQ-9 score. No significant associations between dyspnoea profile and age or other clinical events were observed.

Discussion

Two distinct patterns of dyspnoea burden were observed following LVAD implantation in this analysis. Most patients experienced sustained improvement in dyspnoea, characterized by initial improvement in dyspnoea followed by smaller, sustained improvement in dyspnoea following LVAD implantation. A smaller percentage of LVAD recipients experienced unsustained improvement, characterized by initial improvement in dyspnoea in the first 3 months after LVAD implant followed by substantial worsening in dyspnoea from 3 to 6 months after implant. The existence of different patterns of dyspnoea symptoms following LVAD implantation has not been explored previously. Therefore, the findings of the current study are novel and build on previous research by demonstrating that patterns of change in dyspnoea are not consistent among all HF patients. Furthermore, these findings support our hypothesis and are reflective of the two broad categories of ‘responders’ and ‘non-responders’ to LVAD therapy.

Improvements in dyspnoea for those who experienced sustained improvement follow similar patterns as other symptoms of HF in response to LVAD implantation. Lee et al.⁸ reported initial improvement in dyspnoea, depression, anxiety, and sleep disturbances in the first month following LVAD implantation followed by smaller but sustained improvement from 1 to 6 months after implant. These improvements in dyspnoea most likely represent the immediate unloading of the left ventricle and reduction of pulmonary vascular pressures following LVAD implantation.²⁴ The stability in dyspnoea noted after 3 months for most patients may reflect achievement of haemodynamic stability, but additional research is needed to evaluate this possibility.

More than 1 in 10 of the participants experienced unsustained improvement in dyspnoea following LVAD implantation. It is unclear why these individuals had worsening dyspnoea after the first 3 months post-implantation. There was a higher prevalence of chronic obstructive pulmonary disease (COPD) and asthma in the group who experienced unsustained improvement in dyspnoea, so the increase in dyspnoea may have been due to chronic respiratory disease rather than HF. However, the history of COPD or asthma was not a significant predictor of dyspnoea trajectory in backwards stepwise models.

Participants who were living alone were more likely to experience unsustained improvement when compared with those cohabitating with a spouse or partner. There is evidence that the presence of a care partner who participates in post-implantation LVAD care contributes to better outcomes, including fewer hospitalizations and lower mortality.⁴⁴ Therefore, the absence of a care partner who assists with daily care may have contributed to inadequate post-implantation self-care and may have contributed to worse dyspnoea over time. Additional research is needed to evaluate this possibility.

Depressive symptoms also were associated with the unsustained improvement group. Congruent with the theory of unpleasant symptoms, physiological and situational factors underpin the association between depression and dyspnoea. For example, dysregulation of inflammatory pathways and social isolation increase risk for both depression and HF.⁴⁵ There is evidence from previous research that greater depression is associated with greater symptom burden and worse dyspnoea in HF.^15,46 In a sample of patients with HF (n = 347), Haedtke et al.⁴⁷ reported greater symptom burden among participants with higher depression scores. Worse depressive symptoms are associated with higher reporting of physical symptoms in chronic disease.⁴⁷ Patients with depression often are more focused on internal signals than on outside stimuli and therefore more likely to report physical symptoms, such as dyspnoea.⁴⁸ However, additional research is necessary to determine if unsustained improvement in dyspnoea burden following LVAD implantation is due to increased awareness or if another mechanism is responsible for the association.

Suction events and neurological events were more prevalent following LVAD implantation among those who demonstrated unsustained improvement in dyspnoea when compared with those who demonstrated sustained improvement. There is evidence from previous research that complications of LVAD therapy can interfere with LVAD effectiveness and increase dyspnoea.⁴⁹ However, neither suction events nor neurological events were significant predictors of dyspnoea trajectory. Numbers of adverse events were very small; however, therefore low numbers of events may have contributed to the lack of significant findings. Similar evaluation should be conducted on larger samples to identify associations.

Interestingly, indicators of HF severity (LVEF, NYHA functional class, and INTERMACS category) were not significantly different among those who demonstrated sustained improvement in dyspnoea when compared with those who demonstrated unsustained improvement. Similarly, the prevalence of medical conditions that could influence dyspnoea (renal disease, pulmonary, and hypertension) and the number of comorbid illnesses were not significantly different among members of both trajectories. As each of these factors is known to influence dyspnoea, it is surprising that none were predictors of dyspnoea trajectory. Additional research is needed to explore why most patients respond well to LVAD therapy whereas others do not.

There are several limitations of this secondary analysis. The sample was recruited from a single academic medical centre in the Pacific Northwest, therefore generalizability is limited. Furthermore, the sample was racially homogenous with a predominance of men and may not be representative of the population of patients with HF. Lastly, the group with unsustained improvement was quite small so this may have influenced the ability to detect differences between the groups.

Implications for practice

While an LVAD has the potential to improve symptoms for patients with HF, clinicians should appreciate that changes in symptoms may not be consistent for all patients. To ensure that the burden of dyspnoea is being captured, clinicians should incorporate evaluation of dyspnoea at each visit. Furthermore, to appreciate the severity of dyspnoea and recognize changes in dyspnoea over time, objective measures such as the HFSPS dyspnoea subscale or a simple visual analogue scale would be better than simply documenting the presence or absence of dyspnoea. Objective measures such as these can be used as a guide when teaching patients about when they should contact their provider. Also, to understand factors that trigger episodes of dyspnoea, dyspnoea burden should be evaluated at rest and with exertion.

Depressive symptoms are common in HF. This secondary analysis demonstrated an association between depressive symptoms and greater odds of experiencing unsustained improvement in dyspnoea burden following LVAD implantation. To identify patients at greater risk for unsustained improvement, incorporating regular screening of depressive symptoms into the plan of care is recommended.

This analysis provides evidence that living with a spouse or partner is associated with sustained improvement in dyspnoea following LVAD implantation. Care partners are essential members of the care team and often assist with post-implant self-care. Clinicians need to monitor the living situation of their patients and recognize that patients who do not live with a spouse or partner may be more likely to experience unsustained improvement in dyspnoea following implantation.

Conclusion

While most patients are likely to experience sustained improvement in dyspnoea following LVAD implantation, some patients may not. Clinicians should be aware that patients may experience different patterns of dyspnoea following LVAD implantation and should advise LVAD candidates accordingly. Furthermore, clinicians should monitor patients with greater depressive symptoms and who live alone following LVAD implant, as they are more likely to experience unsustained improvement in dyspnoea.

Funding

The parent study was supported by a grant from the National Institute of Nursing Research (1R01NR013492).

References

Author notes