To our knowledge, this is the first comprehensive systematic review that evaluated PROMs as prognostic indicators in HT patients. We reported that PROMs demonstrated statistically significant prognostic prediction in four of the five included studies [23,24,25,26]. Farmer et al. and White-Williams et al. studied the same cohort of patients as part of a prospective study of health-related quality of life. While Farmer et al. described a statistically significant association between PROMs and survival, this association was not observed by White-Williams et al. The remaining three studies that showed statistically significant prognostic value presented high heterogeneity, which impeded a comparison between studies. Therefore, the results of this systematic review are limited to support the value of PROMs as a prognostic indicator in HT patients.
Previous systematic reviews have evaluated the value of PROMs as prognostic indicators, especially in cancer populations for overall and disease-free survival [28]. Moss et al. identified several domains of QoL as potential prognostic indicators for oncological outcomes in tumors of the pelvic abdominal cavity [28]. Unlike Moss et al.’s study, our systematic review lacks the rigorous evidence to make these assumptions and we are unable to extrapolate these results to heart transplant patients. However, these strong conclusions should be weighted due to the high heterogeneity that Moss et al.’s faced in the design and methodology of their included studies, which impeded the performance of a meta-analysis, as in our study.
Regarding studies similar to ours in the cardiovascular field, Kelkar et al.’s systematic review was focused on disease-specific PROM in heart failure in clinical care, identifying only two out of nine PROMs with prognosis value: the KCCQ and Minnesota Living with Heart Failure Questionnaire (MLHFQ) [19]. Indeed, the KCCQ and the MLHFQ are two disease-specific PROMs that have been demonstrated as prognostic indicators of heart-failure readmissions and deaths [30,31,32]. Moreover, PROMs may have more prognostic value than classifications used in daily clinical practice in heart failure patients. Greene et al. recently demonstrated a greater than 50% discordance between the New York Heart Association (NYHA) class and the KCCQ overall score in heart-failure patients. The authors demonstrated that changes in the KCCQ overall score might have a better prognostic value than the NYHA class, stating that an improvement of five or more points in the KCCQ overall score was independently associated with decreased mortality (hazard ratio = 0.59, 95% CI [0.44, 0.80], p < 0.001) [30]. Despite the temptation to generalize these results to heart transplant patients, our study suggests that the existing disease-specific PROMs in HT are insufficient to be used with prognostic value.
Why PROMs are not widely implemented in patients undergoing HT?
There is a lack of routine and structured use of PROMs in patients undergoing HT, despite the fact they are recommended for systematic use in cardiovascular diseases [31] and their use as endpoints for trials has grown in the last decade [19, 32]. A systematic review of the role of PROMs in contemporary randomized controlled trials in cardiology showed that PROMs were used in 16% of randomized controlled trials published in major cardiology and general-medical journals between 2005 and 2008 [32]. The limited use of PROMs in specific cardiovascular diseases was highlighted by Chen et al., showing that only 18% (43 of 237) of randomized clinical trials reported using PROMs used PROMs in cardiac catheter ablation for treating symptomatic arrhythmias [33].
A potential explanation for the limited use of PROMs could be the potential delays and disruptions of clinical workflows produced by them, as well as the lack of infrastructure and integration in clinical workflows. However, PROMs have been established as a standard of care in HT clinical practices after demonstrating a reduction in the duration of medical visits without affecting the quality of care and being accepted by patients and healthcare providers [29, 34].
Another concern that may have impeded the routine implementation of PROMs is their lack of capacity to weigh the influence of financial resources, education, and the burden on family caregivers on outcomes. PROMs should systematically capture socioeconomic factors, health care access, and different practice patterns that may have associations with disease outcomes while facilitating informed clinical decision-making. For example, in cardiovascular trials PROMs were found to be crucial for informed clinical decision making in almost a quarter of 413 trials but their role was uncertain in a fifth and irrelevant in 5% of them [32].
Finally, the development and deployment of PROMs in clinical settings may be constrained by the costs of hiring additional personnel, the acquisition of new equipment, and follow-up costs.
What characteristics should PROMs have to be a standard of care in HT patients?
In order to be established as a standard of care in clinical settings where HT are performed, we propose that PROMs meet three characteristics: cost-effectiveness, scientific validation, and scalability (see Fig. 3). PROMs that meet these characteristics have captured > 95% of baseline PROs, > 95% of disease severity and treatment outcomes, and > 70% of one-year follow-ups across multiple clinical settings [35].
PROMs expand beyond traditional functional status measures and longevity improvement, assessing a broader impact on patients’ health status and potentially influencing clinical decision-making. Cost-effective PROMs should allow data capture using existing software and hardware infrastructure while migrating from manual to automated follow-up methods[36]. In order to ensure scalability in tight-scheduled and resource-constrained clinical practices, PROMs should be easy to administer, allow automated data collection and scoring of the assessed clinical event, and provide understandable results for patients and caregivers.
In order to limit unintended consequences, PROMs have to be scientifically validated and supported by peer-review publications, demonstrating content and construct validity [19]. Content validity warrants that PROMs consider specific outcomes for a specific population and the participation of a specific population in the development of the instrument [19]. Complementary PROMs with construct validity demonstrate correlations and detect changes over time between PROMs’ results and clinically meaningful measures [19]. For example, the disease-specific KCCQ for heart-failure patients demonstrated content validity by involving them in its development, while it demonstrated construct validity by correlating its results with the six-minute walk test and peak-exercise oxygen consumption [37].
By achieving cost-effectiveness, scientific validation, and scalability, PROMs would consolidate as a cross-sectional and longitudinal standard of care to improve clinical decision-making by, for example, modifying follow-up intervals, involving other providers, comparing outcomes after treatments, individualizing risk stratification, and providing tailored treatment (see Fig. 3).
Additional practical considerations to facilitate the implementation of PROMs are the ease of data collection, the level of collaboration among colleagues, the provision of clear guidelines for implementation and data collection, the level of managerial involvement, the availability of training and support, and the use of technology [38]. Finally, from a provider perspective, PROMs should allow to interpretability and validity of the information [38].
Limitations
The major limitation of our study is the inclusion of a small number of observational studies between 1987 and 2015 with PROMs being collected at very different time intervals, time points, and comparing multiple types of outcomes. Observational studies lack sound methodological guidance compared to randomized clinical trials, which reduces the strength of potential recommendations and conclusions made in our study [39]. Most observational studies present different designs, complicating data extraction and outcomes comparison, increasing the probability of errors and affecting the synthesis of evidence. The difference in study designs, collection of PROMs at different time points, and outcomes were evident in our systematic review, where three studies strategized the collection of PROMs before and after HT [23,24,25] and two studies only after HT [26, 27]. In addition, the included studies in our systematic review were held in three different countries (the United States, Spain, and the United Kingdom), which limited the comparison of results among the included studies due to the different delivery, infrastructure and organization of healthcare systems among countries.
Another limitation of observational studies is the complexity of the risk of bias assessment tools compared to the tools used for randomized clinical trials. We used QUIPS, a risk of bias tool validated for prognostic factors [22], which requires assessing six domains (study participation, study attrition, prognostic-factor measurement, outcome measurement, study confounding, and statistical analysis-reporting). The overall risk of bias was moderate to high in four of the five included studies, indicating a high heterogeneity between studies due to different PROMs used to evaluate the prognostic value, different follow-up intervals, and lack of definition of new comorbidities after the HT. Thus, we did not perform a quantitative comparative analysis between the included studies and therefore the external validity of our results is limited.
However these limitations, our study has shed light on the need to conduct randomized clinical trials where generic and disease-specific PROMs are systematically used to compare their prognostic value with usual care. Such research will elucidate PROMs’ prognostic value and lead to their incorporation as permanent tools to aid clinical decision-making.