Bringing doctors together

or

A free, private network for verified physicians

Leading voices

Dr. Bruno Bezerra Lima

Advanced Cardiac Imaging · VUMC

View

Dr. Sara Ersozlu

Advanced Cardiac Imaging · MGH

View

Dr. Christopher Kramer

Advanced Cardiac Imaging · University of Virginia

View

Dr. Filipe Moura

Advanced Cardiac Imaging · Yale College

View

Dr. Robert Hendel

Advanced Cardiac Imaging · Tulane University

View

Dr. Venkatesh Murthy

Advanced Cardiac Imaging · UMHS

View

Dr. Dorairaj Prabhakaran

General Cardiology

View

Dr. Pugazhendhi Vijayaraman

EP · Geisinger

View

Dr. Robert Schaller

EP · HUP

View

Dr. Trey Woods

General Cardiology · Emory Healthcare

View
  • Discover your community

  • Stay current

  • A profile that represents you

  • Find your people

About RoonPrivacyTerms

    Overview

    Clinical Impact of Routine Assessment of Patient-Reported Health Status in Heart Failure Clinic: The PRO-HF Trial.

    Apr 7, 2024DOI: 10.1161/circulationaha.124.069624Journal ArticleAHA

    Written by: Alexander T SandhuJamie CalmaMegan SkyeNeil KalwaniJimmy ZhengJessica SchirmerNatasha DinCati Brown JohnsonAnshal GuptaRoy LanBrian YuJohn A SpertusPaul A Heidenreich

    Find these authors on Roon

    AG
    Dr. Anshal Gupta
    Advanced Cardiac Imaging · Stanford

    Methods

    The PRO-HF trial was an open-label, pragmatic, patient-level randomized trial comparing routine PRO assessments with usual care in the Stanford HF Clinic. The detailed study design has been published previously18 and is registered on ClinicalTrials.gov (NCT04164004). In brief, adult patients with a scheduled Stanford HF clinic visit from August 30, 2021, to June 30, 2022, regardless of the purpose of that visit, were eligible. The clinic included 13 cardiologists and 4 advanced practice providers who manage patients across the HF spectrum. We excluded patients who potentially would be enrolled in alternate clinical trials that focused on PROs as primary or secondary outcomes. There were no other exclusions in this pragmatic trial.We contacted eligible patients by email approximately one week before their visit. Patients who did not respond to email were contacted by telephone and/or text message. Details regarding enrollment modality have been described separately.20 The Stanford Institutional Review Board approved the study and required informed consent, which was completed electronically. Consented patients were randomized 1:1 via a randomly generated allocation sequence to routine PRO assessment or usual care. Randomization was stratified by the treating clinician with permuted block sizes of 2 and 4. Neither patients nor clinicians were blinded to allocation.De-identified, individual data and a data dictionary will be shared upon reasonable request made to the corresponding author.The Kansas City Cardiomyopathy Questionnaire (KCCQ) is the most widely used HF-specific PRO instrument.17 It summarizes HF’s impact on patients’ symptoms, function, and quality of life. This instrument has demonstrated high reliability,21 is a better marker of clinical change than other clinical or laboratory findings,22 and improves the prediction of subsequent HF clinical events beyond traditional risk factors.23,24This trial used the KCCQ-12 for both routine PRO assessment and as the primary outcome. The KCCQ-12 is a validated, abbreviated version of the KCCQ designed for clinical care.25 It includes 12 questions that query a patient’s perception of the effect of HF on four domains—symptom frequency, physical limitations, social limitations, and quality of life—over a 2-week recall period. The Overall Summary Score (OSS) is the average score across the 4 domains, and the Clinical Summary Score (CSS) is the average of the symptom and physical limitation domains to mirror the concepts quantified by the New York Heart Association (NYHA) classification system. The OSS, CSS, and domain scores range from 0 (severely impaired) to 100 (optimal).
    Participants randomized to routine PRO assessment were instructed to complete the KCCQ-12 prior to each HF clinic visit during the 1 year period post-randomization. We collected the KCCQ-12 via the electronic health record (EHR) patient portal within 3 days of participants’ visits. Clinic staff asked participants who did not complete the KCCQ-12 online to complete the instrument during clinic check-in.
    Individual KCCQ-12 responses and summary scores were available in the EHR in both tabular and graphical form (Figure S1). Clinicians were instructed how the KCCQ-12 responses and summary scores could be embedded in clinical notes. Finally, for each patient’s first clinic visit, the KCCQ-12 results were printed out for in-person visits and emailed to the treating clinician for virtual visits. Clinicians and clinic staff received detailed education regarding the interpretation and potential uses of the KCCQ-12 including educational conferences, individual clinician outreach, and posted informational sheets (Figure S2). Details regarding the education have been reported previously.18,26
    Patients randomized to usual care completed the KCCQ-12 at the time of enrollment on REDCap, a median of 6 days prior to their clinic visit. Their KCCQ-12 results were not visible to their treating clinicians. Patients in usual care were not prompted to repeat KCCQ-12 assessment preceding follow-up clinic visits during the 1-year follow-up period. In both arms, all diagnostic and treatment decisions were made at the discretion of the treating clinician.
    The primary outcome was the patient-reported health status after 1 year of follow-up, assessed via the KCCQ-12 OSS. In both arms, the final KCCQ-12 was assessed at the first HF clinic visit between 12-15 months post-randomization. For patients who did not attend a clinic visit or did not complete the KCCQ-12 during that period, the final KCCQ-12 was assessed via email or telephone. Secondary quality of life outcomes included the KCCQ-12 CSS and the four domain scores of the KCCQ-12.
    Additional secondary clinical outcomes included Stanford acute care encounters: hospitalizations, HF hospitalizations, and ED visits. Stanford clinical encounters were assessed via EHR data. HF hospitalizations were identified based on a principal diagnosis of HF.27,28 We also captured all-cause hospitalization and ED visits outside Stanford via participant report. In addition, the use of guideline-directed medical therapy for heart failure was compared. This included active prescriptions for the following: evidence-based blockers for HF (carvedilol, metoprolol succinate, bisoprolol), angiotensin-converting enzyme inhibitors (ACEI)/angiotensin receptor blocker (ARB)/angiotensin receptor neprilysin inhibitor (ARNI), ARNI alone, mineralocorticoid receptor antagonist (MRA), and sodium-glucose cotransporter-2 inhibitor (SGLT2i). For beta blocker, RASI, ARNI, and MRA therapy, we also evaluated treatment in the subgroup with EF ≤40%. We also assessed the aggregate count of HF medication uptitrations or initiations and the total number of HF medication orders including discontinuation and downtitration. Finally, we evaluated loop diuretic therapy. We assessed medication therapy using the first EHR medication reconciliation between 12-15 months post-randomization. For participants without medication reconciliation during this period, we reconciled medications via telephone. If unavailable, we used the most recently available EHR medication reconciliation.
    Other secondary outcomes included Stanford testing and outpatient management: echocardiography, natriuretic peptide testing, right heart catheterization, cardiopulmonary exercise testing, HF clinic visits, palliative care clinic visits, and psychiatry clinic visits. These outcomes were collected via structured EHR data. Advanced HF therapy evaluation – for transplant or left ventricular assist device - was collected via institutional records.
    The PRO assessment arm was intended to complete the KCCQ-12 at each HF clinic visit between randomization and 1-year follow-up. To measure the fidelity to the intervention, we determined the proportion of HF clinic visits in which the PRO assessment arm completed the KCCQ-12. However, we were unable to measure if completed KCCQ-12 assessments were reviewed by the treating clinician.
    We planned a sample size of 1,200 participants. Prior work has suggested 4-6 points in the KCCQ-12 OSS is the minimal clinically important difference, although a smaller mean change across a population may be meaningful.17,29 With 1,200 patients and 20% loss to follow-up, we estimated 97% power to detect a mean difference of 6 and 73% power for a mean difference of 4 in the KCCQ-12 OSS using a two-sided α of 0.05. With adjustment for baseline KCCQ-12 OSS, we estimated 81% power for a mean difference of 4 via simulation modeling.18
    All analyses for the primary and secondary outcomes were performed based on the intention-to-treat principle. The primary outcome was evaluated using linear regression with adjustment for pre-defined characteristics: baseline KCCQ-12 OSS and treating clinician. The baseline KCCQ-12 OSS was modeled as a restricted cubic spline to account for potential non-linearity.30 For the primary analysis, missing baseline KCCQ-12 OSS values were imputed via multiple imputation with chained equations (details in Supplemental Methods). Patients with missing final KCCQ-12 OSS scores were excluded from the primary analysis.
    We performed multiple sensitivity analyses for the primary outcome. First, we compared the unadjusted final KCCQ-12 OSS via a two-sample t-test. Second, we adjusted for additional predictors of the KCCQ-12 OSS: age and baseline left ventricular ejection fraction (LVEF). We also performed multiple sensitivity analyses around missing data, including imputing the missing KCCQ-12 OSS for all patients alive at 1-year, imputing a score of 0 for patients who died during the 1-year follow-up, imputing 0 for patients who died and the minimum observed score for all living patients with missing data, imputing scores of 0 for everyone with missing data, and imputing scores 100 for living patients with missing data and 0 for those that died.
    Secondary outcomes were analyzed with adjusted regression models. KCCQ-12 CSS and domain scores were modeled as linear regression models. Medication therapy, right heart catheterization, cardiopulmonary exercise testing, any ED visit, any hospitalization, palliative care and psychiatry clinic evaluations were modeled as binary outcomes with logistic regression. Common recurrent events - medication orders, hospitalizations, ED visits, HF clinic visits, echocardiograms, and NT-BNP testing - were modeled as ordinal data with negative binomial regression. Secondary outcomes were adjusted for pre-specified baseline characteristics. We performed three sensitivity analyses for secondary outcomes. First, we evaluated each secondary outcome among participants alive at 1 year. Second, hospitalizations, ED visits, echocardiograms, and NT-BNP tests were also modeled with logistic regression. Third, given routine PRO assessment may impact treatment intensity based on the baseline health status, we evaluated the interaction between baseline KCCQ-12 OSS and the treatment effect for secondary outcomes.
    We performed subgroup analyses for the primary outcome for pre-specified subgroups: age, sex, baseline KCCQ-12 OSS, Charlson Comorbidity Index, LVEF, and HF/cardiomyopathy diagnosis. We explicitly tested group-by-characteristic interactions to explore potential heterogeneity of benefit by each characteristic. We also performed a post-hoc, exploratory subgroup analysis of the primary outcome by stratifying the cohort based on KCCQ-12 OSS at clinically relevant thresholds using conventional NYHA comparisons (>80 – NYHA Class I; 60-80 – NYHA Class II; <60 - NYHA Class III/IV) to better capture potential heterogeneity in treatment effect across baseline health status.17

    RESULTS

    All 17 clinicians in the HF clinic agreed to participate. Among the 5,133 eligible patients with scheduled visits between August 30, 2021, and June 30, 2022, 1,248 (24.3%) were enrolled with 624 randomized to routine PRO assessment and 624 to usual care (Figure 1).
    Median participant age was 63.9 years with 38.9% being women (Table 1). Among participants, 87.3% had a documented history of HF or cardiomyopathy. Of the 1,245 participants with an LVEF available, 28.0% had an LVEF ≤40%, 17.1% with an LVEF 41-50%, and 54.9% with LVEF >50%. The median baseline KCCQ-12 OSS was 82.3 (interquartile range
    Among the 624 patients in the PRO assessment arm, 614 patients (98.4%) completed at least one KCCQ-12. The KCCQ-12 was collected in the EHR for 87.2% (1,331/1,526) of clinic visits over the 1-year follow-up period. Over 1-year follow-up, mortality was 3.7% in the PRO assessment arm and 3.4% in the usual care arm (p=0.76). Among those living, final KCCQ-12 assessment was completed among 87.9% in the PRO arm and 85.1% in usual care (p=0.16). Missing data are detailed in Tables S1-S3.
    The median final KCCQ-12 OSS was 87.5 (IQR 68.8-96.9) in the PRO arm and 87.6 (IQR 69.7-96.9) in the usual care arm. Figure 2 illustrates the baseline KCCQ-12 OSS, the final KCCQ-12 OSS, and the change in KCCQ-12 OSS across arms. The KCCQ-12 OSS improved by ≥ 5 points among 35.9% of participants (185/516) in the PRO assessment arm versus 35.4% participants (179/506) in usual care (p=0.87). The KCCQ-12 OSS worsened by ≤5 or more points among 23.8% of participants (123/516) in the PRO arm and 23.1% of participants (117/506) in usual care (p=0.79). The baseline-adjusted mean difference in the KCCQ-12 OSS was 0.2 (95% CI: −1.7 to 2.0; p=0.85) (Table 2). The results were consistent across imputation approaches (Table S4).
    There was no significant heterogeneity across pre-specified subgroups (Figure 2 Panel B). However, there was significant heterogeneity across conventional NYHA-related stratifications of baseline KCCQ-12 OSS in a post-hoc exploratory analysis. The effect of PRO assessment was greater among participants with baseline KCCQ-12 OSS of 60-80 (5.0; 95% CI: 0.0 to 10.0) than those with KCCQ-12 OSS of <60 (−2.8; 95% CI: −8.7 to 3.1) or >80 (−0.2; 95% CI: −1.8 to 1.4) (p=0.01 and p=0.04 for interactions, respectively).
    Table 2 shows the mean baseline and follow-up levels of the KCCQ-12 CSS and the four domain scores. Each domain score improved over follow-up, with no significant differences in the baseline-adjusted mean between the PRO arm and usual care (Table 2).
    The proportion of patients hospitalized over the 1-year follow-up period was 17.6% in the PRO assessment arm and 19.7% in the usual care arm (Table 3). The odds of at least 1 hospitalization were not significantly different across arms (OR 0.87; 95% CI: 0.68-1.12; p=0.28). The rate of Stanford HF hospitalizations was 9.0 per 100 patient-years (SE 0.4) in the PRO arm compared with 12.0 per 100 patient-years (SE 0.6) in the usual care arm with a rate ratio of 0.75 (95% CI: 0.48-1.17; p=0.20). The rate of ED visits was also similar across arms (Table 3).
    Among the 348 patients with EF≤40%, baseline therapy was the following: beta-blocker 87.6%, RASI 76.7%, MRA 64.7%, and SGLT2i 33.6%. Among the 897 patients with EF>40%, 60.9% were on beta-blockers, 47.2% RASI, 27.6% MRAs, and 8.0% SGLT2i. Treatment with beta-blocker, RASI, and MRA therapy was overall similar at the end of the trial (Table 3). SGLT2i therapy rates increased to 55.5% and 16.3% among those with EF≤40% and EF>40%, respectively.
    There was no significant difference in follow-up HF medication therapy across arms (Table 3). There were also no significant differences among the 1,089 patients with baseline HF/cardiomyopathy. There was also no significant difference in the frequency of HF medication initiation/uptitration across the two arms (rate ratio 1.02; 95% CI: 0.89-1.17; p=0.74).
    The mean number of outpatient HF clinic visits per patient was 2.45 (SE 1.97) in the PRO assessment arm compared with 2.43 (SE 1.86) in the usual care arm (rate ratio: 1.00; 95% CI: 0.90-1.12; p=0.94). Results were similar among those who survived (Table S5). There was no significant interaction between baseline KCCQ-12 and the effect on HF clinic visit frequency (p=0.15) (Table S6). There were also similar rates of outpatient palliative care and psychiatry evaluation (Table 3).
    The frequency of diagnostic testing was similar across treatment arms (Table 3). In the PRO assessment arm, 1.8% of patients started a new advanced HF therapy evaluation compared with 1.4% of patients in the usual care arm (OR: 1.35; 95% CI 0.75-2.42, p=0.31).

    DISCUSSION

    To better support a more patient-centered healthcare system, there have been increasing calls to directly integrate PROs into routine care. In the first randomized trial of integrating clinic-based PRO assessments in a HF clinic, we did not find a significant effect of PRO assessment on patient-reported quality of life or other processes or outcomes of care. Despite prior findings showing that routine PRO assessment improved the accuracy of clinicians’ assessments of patients’ disease severity,19 this did not translate into differences in care or outcomes. These data highlight that PRO implementation requires more than the mechanics of making the results available and likely requires additional training, experience, and support to redesign the processes of outpatient care. This is supported by our qualitative evaluation of this PRO implementation in HF clinic, which found that clinicians broadly supported the process but did not consistently believe the PRO data impacted their treatment decisions.26
    These findings contrast with prior trials evaluating the clinical impact of routine PRO assessment in oncology clinics.31–33 Basch and colleagues evaluated a web-based assessment of patient-reported symptoms among patients undergoing chemotherapy in two trials and demonstrated that PRO assessment improved patient-reported quality of life and 5-year survival. A few noteworthy differences between the oncology PRO trials and our study may explain the difference in results. First, Basch and colleagues monitored patient-reported symptoms at regular intervals with home monitoring between clinic visits, whereas this trial only collected PROs at the time of scheduled HF clinic visits. This may be sub-optimal for two reasons. First, there were only a mean 2.4 clinic visits per patient across the 1-year follow-up period, and it is possible that PRO assessments may be more useful in detecting clinical worsening between scheduled clinical encounters. For example, prior analyses demonstrated the KCCQ score worsens substantially in the month prior to HF hospitalization.34 Home monitoring of PROs with alerts for symptomatic worsening may facilitate earlier intervention and prevent morbidity. The use of PRO measures in digital HF monitoring tools are ongoing.35 Second, Basch and colleagues delivered alerts to clinic nurses when meaningful worsening of patient symptoms was detected, whereas the PRO-HF trial passively provided the results without flagging worsening health status. While PRO results may be meaningful for clinician-patient communication irrespective of whether symptoms are worsening,26 emphasizing symptomatic worsening and involving the entire care team may enhance the impact of routine PRO monitoring on clinical decision-making and outcomes.
    Other design elements may also have limited the potential of PRO assessment to improve outcomes. The PRO-HF population had a high baseline mean KCCQ-12 score compared with other HF cohorts.36,37 To maximize pragmatism and generalizability, we recruited all patients in the HF clinic, with 13% of participants not having HF or cardiomyopathy and for whom a disease-specific HF PRO may not be relevant. This may have contributed to the neutral findings. Supporting this possibility, a post-hoc exploratory analysis found an interaction between the intervention and baseline health status, with no benefit in those with high (>80) or low (<60) baseline KCCQ-OS scores and more benefit in those with KCCQ-OS scores of 60-80, consistent with NYHA Class II. While this may be a spurious finding, it is also plausible that routine PRO assessment is most valuable in those with intermediate health status impairment. It is difficult to improve the health status of patients who are already minimally symptomatic, and the incremental benefits of formal PRO assessments may be less important in those with severe symptoms that are more easily recognized by clinicians.
    Another potential explanation for the neutral findings of this trial is that it may require more education and experience for clinicians to understand and use PRO results, which may have also been facilitated by tying the results to protocolized clinical actions. PROs are similar to other diagnostic interventions, like ambulatory pulmonary artery pressure monitoring, that require accurate interpretation and consistent response to improve outcomes. If PROs are not reviewed, are misinterpreted, or there is no downstream clinical response, any potential benefit would be undermined. Future studies may consider including a longer duration of exposure to PROs for clinicians and protocolized interventions in response to worsening health status.
    Randomization at the patient level may have also influenced our results in two different ways. First, patient-level randomization may have led to contamination with clinicians changing their health status assessment among patients in the usual care arm. Second, clinicians only had access to PRO assessment in a subset of their patients, which may have been a barrier to incorporating these data into their standard practice. Future studies may consider clinician- or clinic-level cluster randomization to reduce contamination and help clinicians develop experience in consistently using these tools during routine care.38
    The lack of impact on clinical outcomes should be balanced against other benefits of PRO implementation. PROs provide clinicians and the health system with a patient-centered measure of their population’s health status. As with mortality, health status is an important measure of the impact of health care. Furthermore, PROs improved the accuracy of clinician assessment and improved patient perception of their clinician’s understanding of their health status.19 Clinicians and patients at sites that have implemented routine PRO assessment have been supportive of its value.26,39 Clinicians have noted routine PRO assessment improves their understanding of longitudinal disease changes, helps the patient better understand their condition, and improves visit efficiency.26,40 However, the benefits should also be weighed against the costs of implementation, patient time for PRO completion, and clinician time to review the data.41 Patients have emphasized the importance of the care team using and discussing PRO data.39 Improving the ease with which clinicians can incorporate PROs into their practice is important for increasing their utilization and impact.
    There are important potential limitations of this trial that should be considered. First, this trial was performed at a single academic center with specialized HF clinicians. Nationally, most patients with HF are managed by general cardiologists and primary care clinicians. There may be greater discordance between clinician assessment and patient-reported HF health status in less specialized practices that may increase any potential benefit of routine PRO assessment. Second, the trial only included 17 clinicians. The impact of PRO assessment is highly dependent on how a clinician implements the data, so the results may vary across a broader set of clinicians. Future secondary analyses of this and future trials should assess the fidelity with which PRO data were used by individual clinicians. Third, this was an open-label trial with non-blinded outcomes assessment. However, outcomes were assessed using patient report and structured EHR data. Finally, the impact of using PROs in routine care was not assessed from patients’ perspectives; prior qualitative studies revealed that patients felt more attention should be given to PROs during clinic visits.39
    Improving the patient-centeredness of care is an important goal. While routine PROs improve the accuracy of clinician assessment in HF clinic, they did not translate to changes in quality of life or clinical decision-making in this study. These results emphasize the need to continue assessing different PRO implementation strategies and clinician education to maximize the potential impact of patient-reported data.
    AG

    Dr. Anshal Gupta

    Cardiology · Advanced Cardiac Imaging
    Stanford Health Care
    anshal