A Longitudinal Comparison of Health-Related Quality of Life in Rural and Urban Recipients of Left Ventricular Assist Devices

Abstract

Left ventricular assist devices (LVAD) are a common treatment for advanced heart failure (HF) to improve ventricular function, symptoms, and health-related quality of life (HRQOL). Many LVAD recipients travel long distances from rural areas for LVAD implantation and follow-up care. Individuals with HF in rural settings who have not undergone LVAD implantation have reported poor HRQOL. However, to date, no studies have compared HF-specific or generic HRQOL in rural and urban LVAD recipients. The purpose of this study was to compare generic and HF-specific HRQOL longitudinally from pre-implantation to 1-, 3-, and 6- months post-implant in a cohort of rural and urban LVAD recipients (N=95; Rural n=32, Urban n=63). We measured generic HRQOL using the European Quality of Life Visual Analog Scale (EQ-VAS) and HF-specific HRQOL with the quality of life domain of the Kansas City Cardiomyopathy Questionnaire (KCCQ). Latent growth curve modeling identified 2 phases of change in generic and HF-specific HRQOL: the initial response to LVAD between pre-implantation and 1-month post-implant and the subsequent change between 1- and 6-months post-implant. Comparable improvements in generic HRQOL were noted in rural and urban LVAD recipients during both phases of change. Urban LVAD recipients had greater initial improvements in HF-specific HRQOL (KCCQ) compared with rural recipients (13.0±5.6, p=0.02), but subsequent improvements were similar among rural and urban recipients. Ongoing assessment of generic and HF-specific HRQOL is necessary during LVAD therapy.

INTRODUCTION

Heart failure (HF) prevalence continues to rise. In the United States, there are over six million individuals with HF and this number is projected to increase to eight million by 2030 (Benjamin et al., 2019; Heidenreich et al., 2013). As the number of individuals with HF grows, the costs and HF-associated mortality increase correspondingly. Approximately nineteen percent of the U.S. population lives in rural areas where the risk of death from heart disease, including HF, is significantly greater when compared to urban areas (Cosby et al., 2019; Knudson, Meit, & Popat, 2014). Individuals in rural areas are at a disadvantage due to limited access to appropriate care for HF. The cardiology workforce is reduced or completely absent in some rural areas, especially in midwestern and western regions of the U.S. (Aneja et al., 2011). Rural HF patients have been shown to be at increased risk of poor HF-related outcomes, including increased mortality and more frequent emergency visits and hospitalizations compared to urban counterparts (Dracup et al., 2014; Teng et al., 2014). Additionally, poor health-related quality of life (HRQOL), has been reported in rural HF populations (Kwon, Lee, Jeon, & Kim, 2017; Nesbitt et al., 2014) These disadvantages and outcome disparities may be even more apparent in those rural-dwelling individuals with advanced HF as end-stage disease requires frequent access to specialty care.

Advanced HF, or HF that is refractory to medical treatment (American Heart Association (AHA) Stage D, New York Heart Association (NYHA) Class IV), is estimated to impact ten percent of the HF population in the U.S. (AbouEzzeddine & Redfield, 2011). Individuals with advanced HF often report poor HRQOL due to the high symptom burden associated with their disease. The gold standard treatment for advanced HF is heart transplantation; however, there are barriers to transplantation. The number of potential organ recipients exceeds the number of organs available; therefore, the waitlist time can be long. Many individuals with advanced HF are ineligible for transplant due to age, and medical or psychosocial limitations (Cowger, 2017; Hsich, 2016). Left ventricular assist device (LVAD) implantation is a common treatment option for individuals who are ineligible for transplant (known as destination therapy) and those that require circulatory support as a bridge to transplantation (Kirklin et al., 2015).

Overall, LVAD implantation has been shown to improve readmission rates, HF-related mortality, and HRQOL over time. However, LVAD implantation requires commitment to an arduous routine of specialty follow up care to avoid adverse events and hospitalizations. Rural LVAD recipients may be at a considerable disadvantage due to poor access to the specialty care required to maintain health with an LVAD. Inability to access specialty care may contribute to complications of LVAD therapy. There is evidence that rural LVAD recipients experience a greater number of hospitalizations in the first year following LVAD implantation (Alonso, Hupcey, Kitko, Pozehl, & Kupzyk, 2019). The higher rate of complications of LVAD therapy observed among rural LVAD recipients could adversely impact HRQOL. There also is evidence that HRQOL is worse among individuals with HF in rural settings compared to their urban counterparts in the absence of LVAD implantation. However, no literature was found that reported data on HRQOL in rural HF patients who were LVAD recipients, despite the potential for rurality to negatively influence HRQOL in the context of LVAD therapy. Therefore, the purpose of this descriptive study was to compare HF-specific and generic HRQOL in the first six months following LVAD implantation in a sample of rural and urban LVAD recipients. We hypothesized that rural LVAD recipients would report poorer HRQOL prior to implantation and would experience smaller improvements in HRQOL in the post-implantation period compared to urban counterparts.

METHODS

Design

This descriptive study is a secondary analysis of data collected during the “Profiling Biobehavioral Responses to Mechanical Circulatory Support in Advanced Heart Failure” (PREMISE) study (R01NR013492). The PREMISE study was a prospective cohort study designed to evaluate patient-reported outcomes in the first six months following LVAD implantation with the goal of gaining insight into the heterogeneity in recipient responses to LVAD therapy. A detailed study protocol for the PREMISE study was previously reported (Lee et al., 2014). Following approval from the center’s institutional review board, subjects were consented and enrolled at a single academic medical center in the northwestern United States between May 2012 and August 2016. To be considered for enrollment, subjects were 21 years of age or older and an Interagency Registry of Mechanically Assisted Circulatory Support profile of 1 to 4 (Lee et al., 2014). Subjects were excluded if they had previously received a heart transplant or LVAD, were undergoing treatment for a concurrent life-limiting illness that would preclude study completion, or were diagnosed with a medically documented, serious cognitive impairment, such as Alzheimer’s disease.

Data for this secondary analysis were collected at four time points in the first 6 months following LVAD implantation including: baseline (pre-implantation - median of 5 days prior to implantation), and 1-, 3-, and 6-months post-implantation. Demographic information (age, gender, race/ethnicity, educational level, employment, and zip code of primary residence) were obtained by subject self-report. Additional subject characteristics were abstracted from the medical record including HF etiology and duration, ejection fraction, NYHA HF class, and implant strategy (bridge-to-transplant vs. destination therapy).

Framework

The conceptual framework for this study was influenced by Ronald Andersen’s Behavioral Model of Health Service Use (behavioral model) (Andersen, 1995). The behavioral model suggests that contextual and individual factors influence health behaviors and subsequent outcomes. In Andersen’s model, contextual and individual factors are sub-divided into predisposing, enabling and need based constructs. Predisposing factors are preexisting contextual or individual characteristics that influence outcomes. Need-based factors are related to acuity. For example, in the context of HF, individuals given a NYHA functional class of IV (indicates severely limited function related to shortness of breath at rest and extreme activity intolerance) are considered a higher need than those with a functional class of III (indicates limited function related to symptom burden, only able to walk short distances, only comfortable at rest) (Caraballo et al., 2019). In this study, we test rural vs. urban settings as a predisposing contextual factor. In our model, we also consider individual predisposing factors including demographics, clinical factors such as co-occurring chronic conditions, NYHA functional class, and LVAD implantation indication (bridge-to-transplant (BTT) vs. destination therapy) and individual-level evaluated need factors such as adverse events and hospitalizations. HRQOL is at the center of our model as the primary outcome as shown in Figure 1.

Figure 1. Conceptual Framework.

Rural and urban settings act as predisposing contextual factors to directly influence longitudinal health-related quality of life in individuals with left ventricular assist devices. Over time, health-related quality of life is also influenced by individual predisposing and need factors including demographics, clinical factors, hospitalizations, and adverse events.

Abbreviations: Health-related quality of life, HRQOL; left ventricular assist device, LVAD

Measures

Rural and urban.

Rural and urban residency were determined based on zip code of primary residence. Zip codes were used to identify county of residence which was then translated to rural or urban based on definitions from the United States Office of Management and Budget (OMB). The OMB divides counties into metropolitan, micropolitan, or neither based on population density and the presence of an urbanized area or urban core within the county. Metropolitan counties have a core area with a population ≥ 50,000 whereas, micropolitan counties have a population ≥ 10,000 but < 50,000 people (Health Resources and Services Administration, 2017). Rural was defined as any county not considered metropolitan or micropolitan.

Quality of life.

HRQOL was measured using two previously validated measures: the European Quality of Life Visual Analog scale (EQ-VAS) to measure generic HRQOL and Kansas City Cardiomyopathy Questionnaire (KCCQ) quality of life subscale to measure HF-specific HRQOL. These tools have been used in prior studies to compare generic and HF-specific HRQOL in the LVAD population (Nassif et al., 2017). The EQ-VAS is a single-item measure that allows subjects to rate their perceived generic health status on a scale of 0 (worst imaginable health state) to 100 (best imaginable health state) while thinking of the previous 7 days (Herdman et al., 2011; Rabin & de Charro, 2001). The EQ-VAS has been used previously in large LVAD clinical trials, including the well-known Risk Assessment and Comparative Effectiveness of Left Ventricular Assist Device and Medical Management in Ambulatory Heart Failure Patients (ROADMAP) study (Estep, Trachtenberg, Loza, & Bruckner, 2015; Starling et al., 2017). The objective of ROADMAP was to compare survival and HRQOL in individuals with HF who received an LVAD to those who received optimal medical management. The EQ-VAS is straightforward and can be administered in less than one minute. These strengths make the measure easily adapted to the LVAD clinic setting. However, patients are asked to consider only the last seven days as they complete this instrument which may lead to over- or under-inflated perceived health status. In this study, the EQ-VAS was measured at regular intervals to alleviate some of this concern.

The KCCQ is a 23-item, HF-specific, questionnaire that assesses seven domains. The 3-item quality of life domain is used in this analysis (Spertus & Jones, 2015). Higher scores indicate better perceived quality of life. The KCCQ has demonstrated good internal consistency (Cronbach’s α=0.95 summary score and α=0.78 for quality of life domain), is sensitive to changes in HRQoL, and has been used extensively to examine HRQOL in populations with HF, including LVADs (Garin et al., 2009; Green, Porter, Bresnahan, & Spertus, 2000; Masterson Creber, Polomano, Farrar, & Riegel, 2012). The I-MACS LVAD registry also currently collects the KCCQ (Kirklin et al., 2018). The strengths of the KCCQ include the tool’s widespread use in the heart failure research, strong association with disease severity measures like NYHA functional class and clinical outcomes, and ease of use (Stehlik et al., 2017; Stehlik et al., 2020). In the current sample, Cronbach’s alpha was 0.90 for the summary score and 0.63 for the quality of life domain. We recognize the quality of life domain achieved a lower internal consistency than would generally be acceptable; however, as Nunnally and Bernstein (1994) point out, values slightly below the threshold are acceptable for domains with few items.

Statistical analyses

Latent growth curve modeling was used to evaluate changes in the EQ-VAS and KCCQ score over the four time points, from baseline (pre-implantation) to 6 months post-implantation. Latent growth curve models assess within-person change and between-person variability over time (Duncan, Duncan, & Strycker, 2011; Preacher, Wichman, MacCallum, & Briggs, 2008). Initial models estimated pre-implantation values and change in HRQOL over time in each group (rural vs. urban). Multiphase growth modeling was incorporated to capture the two main phases of change in patient-reported outcomes following LVAD implantation that have been previously described as early and sustained (Kohli & Harring, 2013; Lindenfeld et al., 2010). The first phase (Δ₁) ran from pre-implantation evaluation to 1-month post implantation. The second phase (Δ₂) ran from 1-month post implantation to 6-months post implantation. Maximum likelihood estimation was used to estimate model parameters. Graphs of sample means and estimated means through the four time points were compared to identify the model that most closely approximated the data. Fit indices, including model chi-square (χ²), root mean square error of approximation (RMSEA), comparative fit index (CFI) and standardized room mean square residual (RMSR) also were evaluated to verify good model fit. Published guidelines directed determination of good model fit (non-significant χ², RMSEA<0.06, CFI>0.95 and SRMR<0.08) (Hu & Bentler, 1999). Values of HRQOL at each time point are reported as means, standard errors, and 95% confidence intervals (CI). Rates of change are presented as mean slope and standard error of the slope. Pre-implantation values and rates of change among urban and rural individuals were compared by evaluating random effects between pre-implantation values and rates of change in EQ-VAS and KCCQ scores in a subsequent analysis. Random effects analysis assumes that pre-implantation values and rates of change are not constant, but instead may vary between groups and that these effects can be evaluated by including the grouping variable (rural vs. urban) in the models. Results of this comparison are presented as t-tests and p-values. Hedges’ g statistics were calculated to identify the effect of LVAD implantation on HRQOL over each phase of change and the overall effect from baseline to six months. Consistent with published standards, 0.2=small effect, 0.5=moderate effect, and 0.8=large effect.(Cohen, 1992). Analyses were performed using SAS™ software version 9.4 (Copyright, SAS Institute Inc. SAS and all other SAS Institute Inc. product or service names are registered trademarks or trademarks of SAS Institute Inc., Cary, NC, USA.), and MPlus version 8 (Muthen & Muthen, 2017). SAS was used to compute descriptive statistics including frequencies, means, t-tests and Chi-squares where appropriate and regression modeling. MPlus was used to model longitudinal data through latent growth curve models.

RESULTS

Sample characteristics

Of the 101 subjects in the parent study, six were excluded due to missing HRQOL data beyond the pre-implantation collection time. Four were lost to follow up and two died prior to the one-month post-implantation data collection time point (one rural, one urban). Baseline demographic and clinical characteristics of the sample (n=95) are shown in Table 1 by rural or urban designation. The sample was predominantly White (83.1%), male (80%) with a pre-implantation NYHA Class of III or IV (93.6%). Rural subjects had significantly higher comorbidity burden including history of myocardial infarction, hypertension, and higher mean body mass index (BMI). Rural subjects were a median distance of 137.3 (IQR 194.9) miles from the implanting center which was significantly greater than urban counterparts (t=4.34, p<0.001). In simple and multiple regression models, a history of myocardial infarction, hypertension, higher BMI, or miles from the implanting center did not predict baseline generic (EQ-VAS) or HF-specific (KCCQ) HRQOL scores.

Table 1.

Baseline Sample Characteristics by Rural or Urban Designation

	Rural (n=32) mean ± SD or n (%)	Urban (n=63) mean ± SD or n (%)
Age (years)	51.5 ± 15.4	53.5 ± 13.5
Sex (male)	26 (81.2%)	50 (79.4%)
Race (Caucasian/White)	25 (78.1%)	54 (87.1%)
Highest level of Education (High school or greater)	27 (84.4%)	58 (92.1%)
NYHA Class (III/IV)	31 (96.7%)	58 (95.2%)
History of MI*	17 (53.1%)	19 (30.1%)
HTN*	22 (68.8%)	25 (39.7%)
BMI*	31 ± 6.4	28.1 ± 4.8
Device Strategy (BTT)	20 (62.5%)	44 (68.8%)
Miles from implanting center* †	137.3 (194.9)	17.9 (35.9)
Alive without adverse events at 6 months	2 (6.3%)	1 (1.6%)

^*p <.05

^†reported as median and IQR

Abbreviations: New York Heart Association Class, NYHA; Myocardial infarction, MI; Hypertension, HTN; Body-Mass Index, BMI; Bridge-to-transplant, BTT; Interquartile Range, IQR

Adverse events: major infection, bleeding, device thrombosis, and neurological dysfunction

Seven subjects received donor hearts prior to the six-month data collection time point – four were urban and three were rural. Nearly all subjects (97%) experienced at least one adverse event during the six-month study period. Two of those surviving without events lived in a rural county and one lived in an urban county. Major infection was the most common adverse event with 57 occurrences in the 6-month study period. Of these, thirteen (23%) were reported in rural subjects. Major infection was followed in prevalence by bleeding (35 events), device thrombosis (26 events), and neurologic dysfunction (25 events). There were no statistically significant differences found between rural and urban subjects on these outcomes. These comparisons are reported in Table 2.

Table 2.

Adverse Events in 6-month Study Period by Rural or Urban Designation

Event	Rural	Urban	Χ2	p-value
Major Infection	13	44	0.14	0.93
Bleeding	9	23	0.48	0.79
Device Thrombosis	7	19	2.76	0.43
Neurologic Dysfunction	5	20	2.25	0.52

Major Infection: driveline, sternal wounds, and/or pump pocket

Bleeding: gastrointestinal, epistaxis, intracranial, and/or subconjunctival hemorrhage

Device Thrombosis: confirmed or suspected pump thrombosis

Neurologic Dysfunction: confusion, cognitive changes, tremor, stroke

A total of 107 hospital admissions were recorded among 58 subjects. Median length of stay for hospitalizations, was 4 days (range 0–109 days). Thirty-seven subjects were not admitted during the study. Of the 58 subjects requiring one or more hospitalization, eighteen were rural. Differences between rural and urban subjects requiring hospitalization were not statistically significant in this sample.

Sixty-seven visits to the emergency department (ED) were recorded in the study cohort. These visits were attended by 40/95 (42%) of the study cohort. Of these, half (20) were rural and the other half were urban. Rural subjects visited the ED a mean of 1.28 times during the study compared to urban subjects who visited a mean of 0.41 times during the study (p<.001).

HRQOL

Generic (EQ-VAS).

Mean EQ-VAS scores improved from baseline through each follow-up time point for both groups. Mean EQ-VAS scores for rural and urban subjects were not significantly different at any time point, suggesting that generic HRQOL was not significantly different between the two groups in this sample (pre-implantation 39.5 [95% CI 31.7–47.3] (rural) vs. 34.5 [95% CI 28.7–40.4] (urban), t=0.95, p=0.35; one month post-implantation 57.6 [95% CI 51.3–63.9] (rural) vs. 52.9 [95% CI 46.9–59.0] (urban), t=1.05, p=0.29; three months post-implantation 61.5 [95% CI 54.9–68.1] (rural) vs. 62.4 [95% CI 57.2–67.6] (urban), t=−0.2, p=0.84; and six months post-implantation 62.7 [95% CI 56.0–69.4] (rural) vs. 65.3 [95% CI 60.7–69.8] (urban), t=−0.62, p=0.54).

Findings from the EQ-VAS multiphase growth model are depicted in Figure 2. Two phases of change were evident: baseline to one-month post-implantation and one month to six months post-implantation. The two-phase growth model fit the data well for both urban (χ²=2.78, p=0.10, RMSEA=0.17, CFI=0.94, SRMR=0.06) and rural (χ²=1.97, p=0.16, RMSEA=0.17, CFI=0.92, SRMR=0.05) participants (Hu & Bentler, 1999). Univariate higher-order moment descriptive statistics were evaluated with no evidence of influential skewing or outliers. Both urban and rural participants demonstrated significant improvements in generic HRQOL as measured by the EQ-VAS in the first month following LVAD implantation. From one month to six months post-implantation, generic HRQOL continued to improve at a smaller but significant rate for urban subjects. Improvements in generic HRQOL from one month to six months post implantation were not statistically significant for rural subjects. The rate of improvement (reported in the row marked diff in Figure 2) was not statistically different for rural and urban subjects in both phases of change (shown in the row marked sig). Implantation of an LVAD had a large, significant effect on generic HRQOL in the first month among rural subjects (Hedges’ g=0.88) and urban subjects (Hedges’ g=0.78). The effect of LVAD implantation on generic HRQOL from baseline to 6 months post implantation was also large and significant for both rural subjects (Hedges’ g=1.12) and urban subjects (Hedges’ g=1.48).

Figure 2.

Changes in generic HRQOL following LVAD implantation in rural and urban subjects measured using the European Quality of Life Visual Analog Scale (EQ-VAS). Changes in generic HRQOL, as measured by the EQ-VAS, are depicted comparing pre-implantation assessment with measures taken at 1, 3 and 6 months after LVAD implantation; the mean and 95% confidence interval is represented by column height and the high and low whisker bars, respectively. Solid lines and darker bar graphs represent subjects from rural environments. Dashed lines and lighter bars represent subjects from urban environments. Two phases of change in generic HRQOL are depicted: the initial response to LVAD between pre-implant and 1-month post-implant (Δ₁) and the subsequent change between 1- and 6-months post-implant (Δ₂). The difference (diff) and significance thereof (sig) of generic HRQOL pre-implant and during each phase of change between patients from urban and rural settings are presented below the figure. Sig is the p-value associated with the independent group t-test that assessed differences between rural and urban groups at each time point.

Abbreviations: Diff, difference; EQ-VAS, European Quality of Life Visual Analog Scale; HRQOL, Health-related quality of life; LVAD, left ventricular assist device; Sig, significance; t, t test

HF-specific (KCCQ).

KCCQ quality of life scores were significantly higher among rural individuals (39.8 [95% CI 31.7–48.0]) compared with urban subjects (27.6 [95% CI 23.0–32.1]) at baseline (pre-implantation) (t= 2.7, p= 0.008), however, differences in mean KCCQ scores were not statistically significant at any post-implantation time: one-month post-implantation 53.4 [95% CI 45.0–61.7] (rural) vs. 53.1 [95% CI 47.3–58.8] (urban), t=0.06, p=.95; three months post-implantation 57.8 [95% CI 49.2–66.4] (rural) vs. 61.9 [95% CI 57.0–66.9] (urban), t=−0.88, p=.38 and six months post-implantation 64.0 [95% CI 57.7–70.3] (rural) vs. 64.6 [95% CI 59.0–70.2] (urban), t=−0.13, p=.90).

Findings from the KCCQ multiphase growth model are shown in Figure 3. As with the trajectories for the EQ-VAS, the two-phase model demonstrated good fit with the data for urban (χ²=2.58, p=0.11, RMSEA=0.15, CFI=0.93, SRMR=0.06) and rural (χ²=0.09, p=0.76, RMSEA=0.00, CFI=01.00, SRMR=0.01) participants (Hu & Bentler, 1999). Review of univariate higher-order moment descriptive statistics revealed no evidence of outliers or significant influential skewing. Following LVAD implantation, there were large, significant improvements in mean KCCQ scores for both groups. Implantation of an LVAD had a larger effect on HF-specific HRQOL in the first month following LVAD implantation among urban subjects (Hedges’ g=1.21) than rural subjects (Hedges’ g=0.56). The overall effect of LVAD implantation on HF-specific HRQOL from baseline to six months was large and significant for both rural (Hedges’ g=1.14) and urban subjects (Hedges’ g=1.83).

Figure 3.

Changes in HF-specific HRQOL following LVAD implantation in rural and urban subjects measured using the Kansas City Cardiomyopathy Questionnaire (KCCQ).

Changes in HF-specific HRQOL, as measured by the KCCQ, are depicted comparing pre-implantation assessment with measure taken at 1, 3 and 6 months after LVAD implantation; the mean and 95% confidence interval is represented by column height and the high and low whisker bars, respectively. Solid lines and darker bar graphs represent subjects from rural environments. Dashed lines and lighter bars represent subjects from urban environments. Two phases of change in HF-specific HRQOL are depicted: the initial response to LVAD between pre-implant and 1-month post-implant (Δ₁) and the subsequent change between 1- and 6-months post-implant (Δ₂). The difference (diff) and significance thereof (sig) of HF-specific HRQOL pre-implant and during each phase of change between patients from urban and rural settings are presented below the figure. Sig is the p-value associated with the independent group t-test that assessed differences between rural and urban groups at each time point.

Abbreviations: Diff, difference; HF, heart failure; HRQOL, Health-related quality of life; KCCQ, Kansas City Cardiomyopathy Questionnaire; LVAD, left ventricular assist device; Sig, significance; t, t test

History of myocardial infarction, history of hypertension, BMI, miles from the implanting center, and proportion of patients who had a visit to the emergency department were included as covariates in subsequent growth models due to baseline differences among urban and rural participants. Each covariate was included in a separate model to evaluate associations between the covariate and baseline values and rates of change over time. None of these covariates influenced generic or HF-specific HRQOL at baseline or over time. The effect of baseline HRQOL on each phase of change also was evaluated but no significant effect was noted.

DISCUSSION

On average, the LVAD recipients in our sample reported improvements in HF-specific and generic HRQOL from pre-implantation to 6 months post-implantation. This coincides with previous reports of improved HRQOL in LVAD recipients (RW.ERROR - Unable to find reference:doc:5c51d97de4b04cd870343dd2; Grady et al., 2016; Grady, 2017; Rogers et al., 2010). LVAD implantation often provides relief from the burdensome symptoms of advanced HF which has been documented to correspond with improvements in HRQOL (Arnold et al., 2016; Delgado & Radovancevic, 2007).

Prior to LVAD implantation rural LVAD recipients reported significantly better HF-specific HRQOL (KCCQ) compared to urban counterparts. We found no evidence in our data to indicate what may have driven this difference. To our knowledge, HF-specific HRQOL has not previously been compared between rural and urban patients. Chronically ill rural adults have been described as living with “quiet pride” and expressed high value in being perceived as “healthy” (Davis & Magilvy, 2000). It is possible that at baseline rural subjects were hesitant to admit how much their HF was impacting their HRQOL resulting in artificially high KCCQ QOL scores. Generic HRQOL measured with the EQ-VAS was also higher in our rural cohort, but this difference was not significant.

Both rural and urban subjects in this sample demonstrated statistically significant improvement in generic and HF-specific HRQOL from baseline (pre-implantation) to the first month following LVAD implantation. However, urban subjects reported a greater increase in HF-specific HRQOL compared to rural counterparts. A similar phenomenon was reported by Nassif and colleagues (2017) where subjects that reported higher HRQOL preimplantation were less likely to see the vast improvements common to the post-LVAD implantation assessments. Urban subjects in this study reported poorer HRQOL than rural subjects at baseline and had more to gain by the insertion of an LVAD. This corresponded to a significantly greater change in HF-specific HRQOL 1-month post-implantation compared to rural counterparts.

Although change in HF-specific HRQOL in the first month following implantation was statistically greater among urban subjects than rural subjects using the KCCQ, improvements in generic HRQOL were not statistically different when measured using the EQ-VAS. The lack of significant difference on the EQ-VAS may be attributable to differences in the sensitivity of the instruments. The EQ-VAS is a single-item, generic measure and is not specific to heart failure. Questions on the KCCQ are specific to how HF affects quality of life, and therefore may be more sensitive to HRQOL in the HF population. Prior to LVAD implantation, patients with advanced HF experience a high symptom burden. LVAD recipients frequently experience an improvement in these severe HF symptoms that would likely correspond to an increase in HF-specific HRQOL. However, this increase may be offset by a reduction in generic HRQOL that results from the increased burden associated with LVAD management. Future research should explore how these instruments differ in their sensitivity to HRQOL in specific conditions. At the time of this study, there were no tools to measure HRQOL in the LVAD population specifically. Since that time, two LVAD-specific HRQOL tools have been under development that target challenges to HRQOL that may uniquely impact those implanted with an LVAD (Grady, 2017; Sandau et al., 2018). Future research will further our understanding of the impact of the LVAD on HRQOL.

Post- LVAD implantation, neither HF-specific or generic HRQOL scores were statistically different between rural and urban subjects at any time point. These findings did not support our hypothesis that rural LVAD recipients would report lower HRQOL. Our study identified HRQOL differences between rural and urban LVAD recipients at baseline, prior to LVAD implantation. What is not clear from our findings is why urban subjects reported lower HF-specific HRQOL at baseline and why generic HRQOL was similar between groups. Concurrent assessment of HRQOL using quantitative instruments and qualitative interviews may help to identify the details that may be missing in a quantitative assessment. Additional longitudinal data collection points also would be helpful to determine if HRQOL in rural LVAD recipients stabilizes or continues to decline. Further exploration of HRQOL and LVAD management in the post-implantation period may help to explain these findings. Patient interviews and qualitative content analyses have been used in prior work with patients with HF and LVAD recipients to identify post-implantation needs and barriers to successful therapy, and to explore pre- and post-implantation expectations in LVAD recipients (Hupcey, Kitko, & Alonso, 2016; Kitko, Hupcey, Pinto, & Palese, 2015; Kitko, Hupcey, Birriel, & Alonso, 2016). LVAD recipients expressed disappointment when the device implantation did not meet their expectations or when they were doing well post-implantation and their improved health status left them stagnant on the transplant list (Kitko et al., 2016). Some patients that were doing well expressed gratitude for being alive and were pleased with their progress while their more ill counterparts expressed doubt if they would put themselves through implantation a second time (Kitko et al., 2016).

Comorbidity, particularly co-occurring depression, has been shown to negatively influence HRQOL in individuals with chronic HF and LVAD implantations (Faller et al., 2007; Kiernan et al., 2016). In this sample, rural subjects experienced a heavier comorbidity burden, with a higher prevalence of hypertension, myocardial infarction, and obesity. Despite the greater prevalence of these conditions, the rural subjects reported better HF-specific HRQOL on the KCCQ at baseline. Rural subjects also visited the emergency department significantly more times in the post-implantation study period compared to their urban counterparts. The tendency for rural patients with HF to seek emergency care more often than urban counterparts has been reported previously (Gamble et al., 2011; Nesbitt et al., 2014). Previous studies have attributed this difference to the smaller number of cardiovascular specialists readily available in rural areas (Aneja et al., 2011). This lack of specialty care may ultimately lead to a greater number of rural patients with HF seeking care in the ED.

A recent report noted a relationship between high social support, decreased stress, and better HRQOL in LVAD recipients (Abshire et al., 2018). There is evidence that, although rural areas may be limited in professional support and access to health services, rural individuals report high levels of social connectedness and support from the communities (Davis & Magilvy, 2000; Rowles & Ohta, 2013). These social factors may contribute to better HRQOL, but additional research is needed to evaluate this association. Importantly, we did not find statistically significant differences in HRQOL between rural and urban LVAD recipients at any time point after LVAD implantation. These findings provide initial evidence that proximity to health services is not the only factor contributing to generic and HF-specific HRQOL following LVAD implantation. Additional research is necessary to uncover the factors that contribute to HRQOL in patients following LVAD implantation.

Limitations

This study has several limitations we would like to acknowledge. First, we recognize that our findings may have limited generalizability given the small, homogenous (white, male) sample that was collected in only one region of the United States and over-represented rural LVAD recipients (33.7%) compared to the general population (19%). The most recent report from the Interagency Registry of Mechanically Assisted Circulatory Support, which includes the implantations of all mechanical circulatory support approved by the U.S. Food and Drug Administration, indicated that the majority of those implanted were white males (Kirklin et al., 2017). The defining characteristics of “rural” in one region may be quite different in another region even within the same country. There are several methods to differentiate rural from urban. We selected our differentiation based on the region. A larger, more regionally, racially, and sex diverse cohort of both rural and urban subjects would be necessary to confirm our findings. Second, as a secondary data analysis, we were not able to account for every covariate that may have influenced HRQOL in rural and urban LVAD recipients. For example, previous reports have identified a connection between post-LVAD HRQOL and sleep quality (Casida, Brewer, Smith, & Davis, 2012). Sleep quality was not evaluated in this study. In addition, we were not able to examine social connectedness, access to transportation, or the presence of a caregiver, which may have contributed to initial higher HRQOL in rural subjects. Third, we would like to acknowledge that the EQ-VAS is a single-item instrument and we elected to only analyze the quality of life domain of the KCCQ as this domain was most appropriate to meet our purpose, to compare HF-specific and generic HRQOL in the first six months following LVAD implantation in a sample of rural and urban LVAD recipients. Finally, the rural sample was small, so the ability to estimate parameters and evaluate random effects may be limited. A larger sample of participants from rural environments might strengthen estimates.

Clinical Implications

There are several important implications to consider in light of our findings. Rural LVAD recipients reported significantly greater pre-implantation HF-specific HRQOL compared to urban counterparts. This finding highlights the importance of ongoing monitoring of HRQOL in patients with advanced heart failure prior to LVAD implantation, particularly for those in urban settings. LVAD recipients in both rural and urban settings reported increases in their generic and HF-specific HRQOL from pre-implantation to post-implantation. Despite the potential influence LVAD implantation may have on HRQOL, HRQOL is unassessed at regular intervals outside the context of research and measurements are largely missing from national databases(Grady et al., 2017). Nurses and LVAD nurse coordinators are in an optimal position to gain an understanding of the nuances of living with an LVAD and how context influences HRQOL. Lastly, rural LVAD recipients visited the emergency department significantly more frequently than urban counterparts. Many rural areas in the U.S. are served by critical access hospitals that may have less experience with LVAD recipients. Ongoing training opportunities including practical experiences in LVAD management and care should be provided for all rural clinicians on a regular basis.

Conclusions

Rural patients with advanced HF reported higher HF-specific HRQOL prior to LVAD implantation. Further, both rural and urban LVAD recipients experienced two phases of improvement in generic and HF-specific HRQOL post-implantation. Ongoing monitoring of HRQOL during LVAD therapy may lead to improved patient outcomes for both rural and urban LVAD recipients.