Do patient decision aids help people who are facing decisions about solid organ transplantation? A systematic review

Abstract Background Decisions about solid organ transplantation are complex. Patient decision aids (PDAs) enhance traditional education, by improving knowledge and supporting patients to align their values with treatments. There are increasing numbers of transplantation PDAs, however, it is unclear whether these are effective. We conducted a systematic review of studies assessing the impact of PDA use in transplantation. Methods We searched the Cochrane Register of Controlled Trials, CINAHL, EMBASE, MEDLINE, and PsycINFO databases from database inception to October 26, 2020. We included primary studies of solid organ transplantation PDAs defined by the International Patient Decision Aids Standards. All comparators and reported outcomes were included. Mean difference in knowledge (before vs. after) was standardized on a 100‐point scale. Pooled‐effect for PDAs was calculated and compared to the standard of care for randomized controlled trials (RCTs) and meta‐analyzed using random effects. Analysis of all other outcomes was limited due to heterogeneity (PROSPERO registration, CRD42020215940). Results Seven thousand four hundred and sixty‐three studies were screened, 163 underwent full‐text review, and 15 studies with 4278 participants were included. Nine studies were RCTs. Seven RCTs assessed knowledge; all demonstrated increased knowledge with PDA use (mean difference, 8.01;95%CI 4.69–11.34, p < .00001). There were many other outcomes, including behavior and acceptability, but these were too heterogenous and infrequently assessed for meaningful synthesis. Conclusions This review found that PDAs increase knowledge compared to standard education, though the effect size is small. PDAs are mostly considered acceptable; however, it is difficult to determine whether they improve other decision‐making components due to the limited evidence about non‐knowledge‐based outcomes.


INTRODUCTION
Solid organ transplantation is the best treatment option for most people with solid organ failure. [1][2][3][4][5][6][7] For some organs, such as heart, lung, or liver transplants, the choice is often between transplantation and conservative care. The decisions for kidney and pancreas transplants are more nuanced as there are other life-sustaining treatments beyond transplantation. For most patients, transplantation offers a survival benefit and improved quality of life. [8][9][10][11][12] However, the degree of benefit for a transplant recipient varies. 2,4,[13][14][15][16] Additionally, any benefit, either for survival or quality of life, must be balanced against the risk of transplant-associated harms. 17,18 Therefore, these decisions can be difficult [19][20][21] and are different for everyone. To decide which option is best for them, patients must have adequate knowledge about treatment options and align their values with their risk-benefit profile.
Patient education is the process of knowledge transfer, to allow recipients to make an informed decision about health treatments. 22 Transplantation education must impart knowledge and explore the risk-benefit profiles of different options in an individualized way. 20,21 Treatment decisions require balancing of rational and emotional assessments of benefits and risks, thus decision-making needs to address knowledge as well as individual patient concerns. 21,23,24 Patients with organ failure are not as informed as they would like about transplantation, despite being motivated to be involved in treatment decisions. [25][26][27][28][29] Patient decision aids (PDAs) are tools to communicate evidencebased information about the benefits and harms of different healthcare options. 30 Their purpose is to meet the two key components of shared decision-making by imparting information plus aligning patient values with potential treatments. PDAs are tools that can be used to supplement patient-provider discussions and may assist with both components of decision-making. In other healthcare fields, a large Cochrane review demonstrated PDAs increase knowledge and improve congruence with patients' values. 31 Despite this, the effectiveness of PDAs for transplantation has not been systematically analyzed, thus it is unclear whether they are effective in this field. There are an increasing number of PDAs for organ transplantation including some in current use. [32][33][34][35][36][37][38][39][40][41][42][43][44][45][46][47] It is therefore necessary to assess whether PDAs are effective for knowledge and decisions about organ transplantation.

MATERIALS AND METHODS
We performed a systematic review of all studies of PDAs in solid organ transplantation evaluating all outcomes, including a metaanalysis of randomized controlled trials (RCTs) assessing knowledge.
The study was conducted based on the Cochrane Handbook for Systematic Reviews on Interventions. 48

Eligibility criteria
We included any studies of PDA use in solid organ transplantation. For comprehensiveness, we assessed any PDA used in any setting, in any country, with any adult population.
We defined PDAs based on the IPDAS guidelines 50 : • "The decision that is being considered is explicitly stated; • The PDA provides evidence-based information about a health condition, particularly the options, benefits and harms, probabilities, and uncertainties; • The PDA helps patients to recognize that the decision is value sensitive and to clarify the values they place on the harms and benefits." The inclusion criteria were adults involved in decisions about solid organ transplantation, including living donors, recipients (living and deceased organs), carers, or clinicians. We included all comparators to the PDA. We included pre-test/post-test, nonrandomized, RCT, and pilot studies. There were no publication date, language, or publicationstatus restrictions. Exclusion criteria were reviews and studies lacking a comparison, intervention, or outcome assessment. We excluded any studies for interventions that did not meet the definition of PDAs based on the IPDAS criteria during full-text review. 50 We contacted authors to access the PDA, if it was not freely available or described in enough detail to assess if it met this definition.

Search
The following databases were searched on the October 26, 2020.
• Cochrane Central Register of Controlled Trials

Selection
All titles and abstracts were independently screened by two authors (G.I., A.W.). Full-texts of relevant studies were reviewed for eligibility. Any disagreements were resolved by third-reviewer consensus or discussion (P.C., J.H.). When multiple reports of the same study were found, the information extracted was collated and treated as one study.

Data collection process
Data were extracted from each study using the data collection forms via COVIDENCE 61

Study risk-of-bias assessment
Different risk-of-bias assessments were performed depending on the study type. Only one tool was used per study type. All RCTs were assessed using the ROB-2 tool. 62 Non-randomized studies were assessed using the ROBINS-I tool. 63 Pre-test/post-test intervention studies were assessed for risk-of-bias using the National Institutes of Health Quality Assessment tool for before-after interventions. 64 This score has been used in other pre-test/post-test risk-of-bias assessments. 65,66 Questions 11-12 in the National Institutes of Health Quality Assessment tool for before-after interventions were completed but not reported as they did not apply to this intervention.
All risk-of-bias assessments were undertaken by two reviewers with disagreements resolved by consensus. Risk-of-bias was done using software: COVIDENCE and ROB-2 Excel macro. Risk-of-bias graphics were presented using the ROBVIS tool. 67

Mean difference in knowledge for RCT
Knowledge was assessed differently depending on the type of study.
For the RCT studies, the mean difference in patient knowledge before and after PDA use was compared to the mean difference in patient knowledge before and after standard of care (traditional education used at the transplant centers). All knowledge tests were developed by the primary authors and based on information within the PDA, so differed for every study.

Outcome measurement
All studies that measured knowledge did so shortly after the use of the PDA. The mean difference in knowledge, between baseline and shortly after either PDA or standard education use, was calculated. If these raw data were not given, then they were estimated from the graphs from the published studies. To compare the different studies, the proportion of accurate responses was scaled to be a standardized score from 0 (no knowledge) to 100 (perfect knowledge). This technique has been employed in other systematic reviews of PDAs. 31 If no standard deviations (SD) were given but confidence intervals or p-values were available, then the SDs were derived. 48

Synthesis methods
The mean difference in knowledge was combined across the RCT studies using a random-effects model because of the likelihood of differences in treatment effect due to intervention variability of the studies. The inverse variance method was used for meta-analysis using RevMan. 68 One study Waterman 2019 46 had two intervention arms to one control group; to prevent counting the control group twice (unitof-error analysis) the control arm was split in half so the control arm contributed to both interventions. 69 The robustness of the results was assessed using sub-group sensitivity analysis of different organs, different PDA formats (paper, web-based) and excluding high risk-of-bias.
Reporting bias was assessed by funnel plot.

Certainty assessment
We used the GRADE approach for certainty assessment which is considered best practice for assessing synthesized findings for systematic reviews. 70 Only the mean knowledge difference assessed by RCT was suitable for GRADE assessment. Other outcomes were not able to be assessed using the GRADE guidelines as there were too few studies using the outcomes and they were assessed in different ways.

Knowledge assessment: Pre-test/post-test and non-randomized studies
For pre-test/post-test studies and non-randomized studies, metaanalysis is not advisable as there is no control group to compare the outcome to. The summary of the effect estimate was performed using Cochrane methodology. Difference in mean knowledge before/after PDA use and statistical significance was documented in tabular-form.
For the non-randomized study of knowledge, the outcome was tabled.

2.11
Other outcomes For comprehensiveness, all outcomes in any study included were assessed as part of this review. The outcomes reported were; acceptability, accuracy of risk perception, adverse effects, behavior, choice made, communication, decisional conflict, durability of a decision, feeling informed, readiness, self-efficacy, and value congruence. These were mapped to the IPDAS criteria. 71 The definition and methods for synthesis are described in the Supplementary Appendix. All of these outcomes were too heterogenous to allow for quantitative synthesis so qualitative summary synthesis was used. Meta-analysis was not feasible for any outcomes apart from knowledge due to differing tools for assessment and outcomes not having an RCT control arm.

Study selection
After the search, 9530 reports were imported for screening (CINAHL = 1333, Cochrane = 442, EMBASE = 5119, Medline = 2272, PsycInfo = 359, Grey Literature = 5). Figure 1 illustrates the study selection process. Reports that detailed the same study were collated into one study. Fifteen studies were included in the final analysis. Some studies reported assessment of resources they described as transplant decision aids but were excluded for failing to meet the IPDAS criteria for being a PDA. As per the IPDAS criteria, a key component that distinguishes a PDA is it "helps patients to recognize that the decision is value sensitive and to clarify the values they place on the harms and benefits. 50

Study characteristics
Of the final 15 studies, there were 4278 participants (

Risk-of-bias assessment
Risk-of-bias was assessed for knowledge. When knowledge was not included, then the primary outcome of the study was used to assess risk-of-bias including actions to pursue transplant (behavior) 34 and choice made. 41 Figures 2 and S1-S4 show the risk-of-bias assessment.
The risk for the RCTs varied from low to high (Figures 2 and S1).
Two studies had problematic randomisation processes and missing outcome data leading to high risk-of-bias. Pritchard et al. 40   the PDA and those who declined were used as the comparison group).

Interventions
There were a variety of formats and modes of delivery used for the PDAs (Table S7). There were also variable environments for PDA usage; some were used by patients alone while others were used within consultations. All the PDAs have been described using the SUNDAE checklist (Table S8).

Outcomes
Several outcomes were assessed in the included studies (Table 1).
These are listed below and described in greater detail within the Supplementary Methods.

Knowledge
Eighty seven percent of studies (n = 13) assessed knowledge. All demonstrated an increase in knowledge with PDA use. Seven RCTs assessed knowledge (Table 2) and six could be combined for metaanalysis ( Figure 3). This favored PDA to control with a mean knowledge difference of 8.01 on the 0-100 scale (95% CI 4.69-11.34, p < .00001).

Heterogeneity
There was moderate statistical heterogeneity with an I 2 value of 75%, however, most of the CIs overlap.

Sensitivity analysis
There was no change in the direction or strength of effects for the meta-analysis with subgroup-analysis restricted to kidney transplants or excluding high-risk-of-bias studies. There was no change when removing PDAs which looked at the complexities of decisions about transplant (i.e., high viral risk donors), rather than whether to have a transplant or not.

Reporting bias
A funnel plot ( Figure S5) suggests no publication bias.

Certainty of evidence
For knowledge for the RCTs the certainty of evidence was low. This was down-graded from high due to risk-of-bias and inconsistency based on the moderate heterogeneity from the I 2 value as per the GRADE methodology by Cochrane. There was one non-randomized trial design that had low certainty evidence. The pre-test/post-test studies ( Table 2) also showed a statistically significant increase in knowledge though there is no comparison for this outcome.

Accuracy of risk perception
Two studies found strong evidence that the PDA improved risk perception. 42,43 This is defined as whether patients could accurately judge the probability of an outcome for an individual with similar characteristics to themselves.

Acceptability
The acceptability of the PDA (whether it helped users make a decision) was examined in nine studies. 32 17% (n = 14) of users found the survival graphs upsetting to view. There were no other reported adverse outcomes for cost or health impacts.

Behavior
Forty seven percent (n = 7) of studies assessed behavior outcomes, defined as whether the intervention led to any behaviour change. 34

Choice made
Two studies assessed whether patients made a choice after PDA use.

Communication
Two studies examined patient-clinician communication. 38

CONCLUSION
In conclusion, our review demonstrates that PDAs increase knowledge and are mostly considered acceptable with few adverse outcomes when making decisions about solid organ transplantation. These results support the ongoing use and development of these tools as they have potential to improve the transplant-related knowledge of patients with organ failure. Further work, however, is needed on the impact of PDAs on other markers of decision-making.

AUTHOR CONTRIBUTIONS
Georgina L. Irish: Substantially contributed to the design of the work, undertook screening, data extraction, risk-of-bias assessment, analysis of the data, drafting of the manuscript, and gave final approval.
Alison Weightman: Undertook screening, data extraction, the risk-ofbias assessment, participated in critical revision of the paper and gave final approval. Jolyn Hersch: Resolved conflicts as a third reviewer, contributed to analysis and interpretation of the data, participated in critical revision of the paper, and gave final approval. P. Toby Coates: Substantially contributed to the interpretation of data for the work, participated in critical revision of the paper, and gave final approval.
Philip A Clayton: Substantially contributed to the conception of the work, resolved any conflicts as a third reviewer, contributed to analysis and interpretation of the data, participated in critical revision of the paper, and gave final approval.