Study
Patients were enrolled in the multi-center IMPACT study [17, 18], investigating CAD patients with multiple comorbidities. Patients were recruited at the cardiology departments of the Amsterdam University Medical Centers (Amsterdam UMC): Academic Medical Center (AMC) and VU Medical Center (VUmc). The overall objective of the IMPACT study was to improve the conceptualization of HRQoL and to enhance the sensitivity and comprehensiveness of its measurement. A subsample of these patients was enrolled in an add-on study on momentary assessments.
Patients
Patients were scheduled for cardiac revascularization procedures after being discussed in the multidisciplinary “heart teams”. Patients were eligible if they were 18 years or older, had stable CAD and were scheduled for elective coronary artery bypass graft (CABG) or elective percutaneous coronary intervention (PCI). Patients had to have at least one somatic comorbidity (e.g., diabetes mellitus, obesity, joint disease). Additional inclusion criteria for participation in the current add-on study on momentary assessments were being experienced smart phone users (indicated at their own discretion), and having a functional Wi-Fi connection at home to enable daily transfer of data to a central server to avoid data loss. Patients with cognitive impairments due to brain haemorrhage, cerebral infarction, mental retardation, dementia, Alzheimer’s disease, or patients who were unable to complete questionnaires due language problems were excluded.
Procedure
Patients completed momentary and retrospective questionnaires at baseline (up to 1 week prior to PCI and CABG) and at two follow-up time points. These latter time points differed per vascularization type due to expected difference in recovery, i.e., 2 weeks (for PCI), 3 months (for PCI and CABG), and 6 months (for CABG) following revascularization (see Fig. 1). For this study, we used the data of two time points: baseline and 2 weeks for PCI and baseline and 3 months for CABG. If these follow-up data were missing, data collected at 3 months (for PCI) and 6 months (for CABG) were used (see dotted line in Fig. 1). Momentary assessments were conducted over the course of 7 days. Patients received an iPod for the duration of the assessment period with the PsyMate™ application installed (www.psymate.eu). PsyMate™ was programmed to give nine beeps during daytime, at random moments within predefined time slots (maximally 2 h apart). After each beep, a set of items assessing HRQoL was presented. If patients did not respond within 15 min, the application was programmed to close, making response to that particular beep impossible. Hence, the maximum number of completed momentary questionnaires per time period is 63 (seven beeps × 9 days). We did not adopt a minimum completion rate for the momentary questionnaires. Retrospective questionnaires of HRQoL were administered 1 day after completion of the momentary assessments. Since these questions employed a one-week time frame, the period of momentary assessments coincided with that week. Patients had the choice between completing the retrospective questionnaires on paper or online. Criterion measures of HRQoL change were collected together with the retrospective questionnaires at baseline and follow-up. Demographic information was collected at baseline. As the central ethics committee decided that the Medical Research Involving Human Subjects Act did not apply, the study was exempted from further ethical assessment. Written informed consent was obtained from all patients.
HRQoL measures
Momentary items
Patients rated a total of 14 items measuring five dimensions of HRQoL. Items were based on an earlier version developed by Maes and colleagues [9] for patients with tinnitus. We adapted this version to make it more suitable for patients with CAD. We particularly replaced the symptoms of tinnitus by the symptoms CAD patients may have. The Positive mood dimension was measured with four items (i.e., “I feel...‘cheerful, ‘relaxed’, ‘energetic, and ‘happy”’). Negative mood was measured with four items (i.e., “I feel … ‘anxious’, ‘sad, ‘irritated, and ‘worried”). CAD symptoms were measured with five items (i.e., “I feel … ‘tired’, ‘shortness of breath’, ‘pain on my chest’, ‘tightness on my chest’, and ‘an oppressive feeling on my chest”’). Fatigue was measured with two items (i.e., “I feel … ‘tired’ and ‘energetic’”). Pain was measured with two items (i.e., “I feel … ‘pain’, and ‘pain on my chest”’). All items were rated on a 7-point scale, ranging from 1 (“not at all”) to 7 (“very much”).
Retrospective items
Patients rated the same 14 items that were administered as momentary items, now phrased in the past tense, referring to the past week. For example, “past week I felt...‘energetic’, ‘relaxed’, ‘cheerful’, and ‘happy”’. These items were also rated on a 7-point scale, ranging from 1 (“not at all”) to 7 (“very much”). The items were again combined to form the same five dimensions as for EMA: positive mood, negative mood, CAD symptoms, fatigue and pain.
Criterion variables
Subjective change in HRQoL
Subjective change in HRQoL was measured with the Subjective Significance Questionnaire (SSQ) [21]. The SSQ consists of six Likert items measuring subjective change in HRQoL since the cardiac revascularization. From the SSQ we selected three items which provide a criterion measure for the change in positive and negative mood (e.g., “To what extent did your emotional state change since the cardiac revascularization?”), change in fatigue (e.g., “To what extent did your fatigue change since the cardiac revascularization?”), and change in pain (e.g., “To what extent did your pain change since the cardiac revascularization?”). All items were rated on a 7-point scale, ranging from 1 (“much worse”), to 7 (“much better”). Scores of 1 to 3 represent a decline in HRQoL, 4 no change, and 5 to 7 an improvement in HRQoL since the cardiac revascularization.
NYHA class
The NYHA class was measured by the patient-based version of the New York Heart Association (NYHA) functioning classification system [19, 20]. The NYHA classifies functional limitations due to CAD symptoms. NYHA consists of one item assessing limitations during physical activity (I = “not limited in physical activities”, II = “somewhat limited in physical activities”, III = “fairly limited in physical activities”, IV = “not capable of physical activities”). Change in NYHA class provides a criterion measure for the change in CAD symptoms.
For assessing changes in pain and fatigue, both the SSQ and NYHA class were used as criterion measures. The separate criterion measures for fatigue (SSQ fatigue) and pain (SSQ pain) coincide conceptually with the pain and fatigue scales. Furthermore, the criterion measure for CAD symptoms (NYHA class) coincide partially with the pain and fatigue scale; i.e., ‘pain on the chest’ and ‘tired’ are considered CAD symptoms. We therefore wanted to examine how these two items relate to the NYHA class.
Analysis
To enable the comparison of the momentary data with the retrospective data the momentary HRQoL data were combined into an aggregated mean score per assessment period, i.e., an aggregated mean score at baseline and at follow-up. We thereby brought the nested momentary data at the same person level as the retrospective data. This enabled us to analyse the associations with the criterion variables using the same regression model.
Scale structures of momentary and retrospective HRQoL measures
Exploratory factor analysis (EFA) was applied to momentary and retrospective items (at baseline and follow-up) to examine the scale structure (positive mood, negative mood, CAD symptoms, fatigue and pain), and whether the scale structure remained stable over time. In all cases we extracted a five factor solution using the minimum residual and varimax rotation method. Furthermore, we calculated the Cronbach’s alpha of each scale (momentary and retrospective) at baseline and follow-up. If the expected scale structure was not observed, and/or the reliabilities were suboptimal (Cronbach’s alpha < 0.70), analysis would be performed at item-level.
Change in momentary, retrospective HRQoL and criterion measures
For each patient, we first calculated change scores (from baseline to follow-up) for momentary, retrospective and criterion measures (only for change in NYHA class).
Change in momentary and retrospective HRQoL
Momentary change scores were calculated by subtracting the average item scores at follow-up from the average item scores at baseline. Retrospective change scores were calculated by subtracting item scores at follow-up from item scores at baseline. A positive change score on the positive mood items (i.e., cheerful, relaxed, energetic and happy) indicates more positive mood. Change scores of the negative mood and CAD symptoms items were reversed, such that a positive change score on negative mood, pain, fatigue, and CAD symptoms also indicates better functioning, i.e., less negative mood and symptomatology.
Change in NYHA class
was calculated by subtracting the NYHA class of the follow-up from the NYHA class of the baseline. A positive change score indicates improved physical functioning (less limited in physical activity due to CAD symptoms).
Association momentary and retrospective change with criterion measures of change
For each HRQoL outcome, we describe mean momentary and retrospective change for patients who reported declined, unchanged or improved HRQoL on the corresponding criterion measure of change to obtain more insight in the direction and magnitude of change.
For each HRQoL outcome we fitted a separate regression model with momentary and retrospective change as independent variables, and the corresponding criterion measure of change as dependent variable to determine the relative strength of the association of momentary and retrospective change with their corresponding criterion measure of change. For each model we expected a positive association between momentary and retrospective change and the criterion measures of change. Additionally, hierarchical regression analysis was used to determine whether momentary change was significantly more related to the criterion measure than retrospective change. For each HRQoL measure we first fitted a simple regression model with only retrospective change as independent variable. Next, we compared the model fit, using the chi-squared statistic, of the simple model with the full regression model (both momentary and retrospective as independent variables). If the full model fitted significantly better than the simple model, momentary change was significantly more related to the criterion than retrospective change. All regression models were adjusted for baseline momentary (mean) and retrospective scores. Results were considered statistically significant with a 2-sided p-value of < 0.05.
Software
All analyses were performed in R version 3.6.1 [22].