Abstract
How individuals perceive uncertainties in sequencing results may affect their clinical utility. The purpose of this study was to explore perceptions of uncertainties in carrier results and how they relate to psychological well-being and health behavior. Post-reproductive adults (N = 462) were randomized to receive carrier results from sequencing through either a web platform or a genetic counselor. On average, participants received two results. Group differences in affective, evaluative, and clinical uncertainties were assessed from baseline to 1 and 6 months; associations with test-specific distress and communication of results were assessed at 6 months. Reductions in affective uncertainty (∆x̅ = 0.78, 95% CI: 0.53, 1.02) and evaluative uncertainty (∆x̅ = 0.69, 95% CI: 0.51, 0.87) followed receipt of results regardless of randomization arm at 1 month. Participants in the web platform arm reported greater clinical uncertainty than those in the genetic counselor arm at 1 and 6 months; this was corroborated by the 1,230 questions asked of the genetic counselor and residual questions reported by those randomized to the web platform. Evaluative uncertainty was associated with a lower likelihood of communicating results to health care providers. Clinical uncertainty was associated with a lower likelihood of communicating results to children. Learning one’s carrier results may reduce perceptions of uncertainties, though web-based return may lead to less reduction in clinical uncertainty in the short term. These findings warrant reinforcement of clinical implications to minimize residual questions and promote appropriate health behavior (communicating results to at-risk relatives in the case of carrier results), especially when testing alternative delivery models.
Keywords: Uncertainty, Genome sequencing, Carrier testing, Genetic counseling, Web platform
Relative to a genetic counselor, computerized return of results was associated with greater uncertainty about whether results were trustworthy and what to do in response.
Implications.
Practice: Web-based return of results from sequencing may leave patients with greater uncertainty about health implications and indicated health behaviors, which highlights the importance of reinforcing key clinical information.
Policy: Policy makers should attend to the diverse uncertainties inherent in genome sequencing when considering alternative delivery models for return of results.
Research: Future research should address how specific uncertainties manifest following return of carrier results among other populations (e.g., younger adults making reproductive decisions) and return of other types of results (e.g., medically actionable).
Introduction
Health care providers should ensure that patients understand and have appropriate expectations of the prevalence and types of uncertain genomic information. Perceived uncertainty—or the “conscious awareness of ignorance” [1]—has been identified as a key variable to be studied in social and behavioral research in genomics [2] and may have undesirable effects on subsequent use of genomic information. For example, a meta-analysis on illness uncertainty found medium-to-large, direct associations with information avoidance across 6 studies and anxiety across 34 studies [3]. However, it is unclear whether these direct associations with information avoidance and anxiety also pertain to receipt of results from sequencing. Given the extensive variety and breadth of uncertainties within this context, these uncertainties may give rise to unique challenges [4].
Uncertainty in genome sequencing has been described [5] and its dimensions delineated in a conceptual taxonomy [1], but to the best of our knowledge, the ways in which uncertainty about sequencing results affects patient outcomes has not been reported. In this context, uncertainty arises from various specific sources, including probability (i.e., risk), ambiguity (i.e., lack of reliability, credibility, or adequacy of risk information), and complexity (factors that make risk information difficult to understand), each of which may be associated with distinct patient outcomes and thus warrant distinct management strategies [1]. Distinguishing dimensions and types of uncertainty that may be reducible (e.g., personal uncertainties about the implications of sequencing for one’s life and well-being) from those that are not (e.g., penetrance of a pathogenic variant) is an important prerequisite to understanding how distinct elements of uncertainty change following receipt of results. The Perceptions of Uncertainty in Genome Sequencing (PUGS) scale assesses three personal uncertainties that relate to receipt of sequencing information: clinical uncertainty about patient and physician actions in response to sequencing results, affective uncertainty about the implications of sequencing results for one’s psychological well-being, and evaluative uncertainty about the trustworthiness of the information [6]. Understanding how these specific, patient-centered uncertainties change following receipt of results may inform the development of effective strategies to maximize the benefits patients derive from clinical sequencing—which may include communicating results to at-risk relatives—and minimize the consequences (e.g., psychological distress).
A randomized controlled trial (RCT) to evaluate a web platform relative to a genetic counselor for return of carrier results from sequencing reported noninferiority in terms of relevant knowledge, test-specific distress, and decisional conflict about choosing to learn results [7]. Reported herein are further exploratory analyses from that trial to assess participants’ perceptions of clinical, affective, and evaluative uncertainties and their associations with one source of uncertainty (ambiguity) and two clinical outcomes (test-specific distress and communication of results).
The primary aim of this study was to explore how the three types of personal uncertainties assessed (clinical, affective, and evaluative) changed following receipt of results. A secondary, more applied, aim of this study was to assess whether perceptions of uncertainty differed depending on how the results were returned (genetic counselor or web platform). Because they would have more opportunity to clarify uncertainties by asking questions of the genetic counselor, we hypothesized that (a) participants randomized to receive results in-person from the genetic counselor would report lower perceptions of uncertainty. We also hypothesized that (b) participants’ anticipated ambiguity of sequencing results would be directly associated with perceptions of uncertainty; and (c) perceptions of uncertainty would be directly associated with test-specific distress and inversely associated with communication of results to children.
MATERIALS AND METHODS
Participants
The ClinSeq cohort is a longitudinal sequencing study of early adopters of technology at the National Institutes of Health (NIH) [8,9] registered on ClinicalTrials.gov (identifier: NCT00410241). Members of this adult cohort were eligible to participate in this trial if they (a) were heterozygous (carriers) for a variant in a gene associated with a disorder inherited in an autosomal recessive pattern and (b) had not previously received carrier results. A list of more than 1,500 autosomal recessive genes for inclusion was compiled using several sources [10,11] and expert opinion. Variants associated with disorders for which there was evidence of a heterozygote phenotype were excluded. Clinical Laboratory Improvement Amendments (CLIA)-validated variants classified as pathogenic, likely pathogenic, or variant of uncertain significance (VUS) were returned to participants. The most common results related to carrier status for hemochromatosis, biotinidase deficiency, and α-1 antitrypsin deficiency.
Trial design
To evaluate the impact of web-based disclosure of results from sequencing relative to standard disclosure by a genetic counselor, participants were randomly allocated in a one-to-one ratio to be educated by and receive personal carrier results from either a genetic counselor or the web platform. Separately, the participants were also randomly allocated in a one-to-one ratio to receive genetic counseling or not following education disclosure sessions; the counseling sessions included invitations to explore participants’ thoughts and feelings about their results and communication of results. The analyses and findings described herein derive exclusively from comparison of the education randomization arms as there were no significant differences in uncertainty based on the counseling randomization. As reported previously [12], there were no significant differences in primary outcomes based on counseling randomization, but those who received genetic counseling were more satisfied. The randomization sequences were generated with an online resource (Research Randomizer, version 4.0, https://www.randomizer.org/). Allocation was concealed until participants arrived at the NIH to receive results, but blinding was not possible due to the nature of the interventions. This trial was approved by the National Human Genome Research Institute Institutional Review Board; it began in February 2014 and was concluded in December 2016 based on power calculations and response rates at follow-ups.
Genetic counselor arm
As reported elsewhere [7], professional practice guidelines were used to create the content for educating participants about sequencing and its limitations, recessive inheritance, and carrier status. Individualized CLIA reports were reviewed with each participant, as were descriptions of the respective genetic conditions and risks for the participant to have an affected child or grandchild. The same genetic counselor (K.L.L.) conducted all of the education disclosure sessions within the genetic counselor arm. Only participant questions directly related to the educational content of the sessions were answered by the genetic counselor; other questions (e.g., “When might I have another opportunity to learn results from sequencing?”) were deferred to counseling sessions (as appropriate) or after trial conclusion. Fidelity to the intervention was assessed through analysis of 106 transcripts (47%) of education disclosure sessions to ensure that major topics were covered and additional topics were not introduced.
Web platform arm
The web platform for returning results paralleled the content of the education disclosure session provided by the genetic counselor. The site was divided into three sections: “Review of ClinSeq,” “Genetics Overview,” and “Results.” The first two sections were identical for all participants. The “Results” section, however, was customized for each participant and included each of the individual’s variant results followed by descriptions of the conditions and risks to children and grandchildren.
The web platform was designed to minimize its influence on the processing of content. The content was formatted in such a way that individual concepts were presented on separate pages for easy understanding. Animated graphics, graphical cues, and rollovers were added to elaborate on content where needed. Special attention was placed on the design of “Results” page to accommodate a variable number of genetic variants per participant. To ensure that results were returned with the same content flow in both education arms, users were not able to jump ahead in the material. However, the web platform did allow users to review previous pages at any time.
Specific data on any gene listed on one or more reports, including descriptions of each associated genetic condition and risks to have an affected child and/or grandchild, were assessed, organized, and stored in a database. As a participant signed into the web platform, the site dynamically generated the necessary customized results content based on the individual’s genetic results. Participants were only able to view the web platform on computers at the NIH Clinical Center in a private setting monitored by a research assistant. Access to results was controlled by providing each participant a unique username and password. The only personally identifiable information in the web platform was the lab report page that was specific to the user.
Measures
A mixed-methods approach was used. Quantitative surveys were administered to all participants at three time points: at baseline (after consent but before return of results), 1 month after the education disclosure session, and 6 months after the education disclosure session. Qualitative data on personal and practical uncertainties were also collected, consisting of questions participants asked of the genetic counselor during education disclosure sessions and residual questions reported by participants on surveys following both interventions. Quantitative surveys included the following psychometric scales:
Perceived uncertainty was assessed at all time points using the PUGS scale [6], which contained eight items to assess perceptions of three distinct types of personal uncertainties in sequencing results. Clinical uncertainty was assessed using three items (e.g., “what actions I need to take based on my test results”; Cronbach α = 0.76–0.80), affective uncertainty was assessed using three items (e.g., “whether I am worried or concerned about my test results”; Cronbach α = 0.80–0.91), and evaluative uncertainty was assessed using two items (e.g., “whether I can trust my test results”; Cronbach α = 0.90–0.96). Each item was ranked using a 5-point, Likert-type response scale ranging from “very certain” (1) to “very uncertain” (5). Scores ranged from 1 to 5 with higher ones indicating greater perceptions of uncertainty for each subscale.
Anticipated ambiguity of sequencing results was measured at baseline with five items assessing perceptions of limitations in the reliability, credibility, or accuracy of sequencing results—a source of uncertainty defined in the behavioral decision making literature as ambiguity (e.g., “I don’t think my sequencing results will give clear answers about my future health”). Each item was ranked using a five-point, Likert response scale ranging from “strongly disagree” (1) to “strongly agree” (5). Averaged scores ranged from 1 to 5 with higher scores indicating greater anticipated ambiguity (Cronbach α = 0.73).
Test-specific distress was assessed at 6 months using the distress subscale of the Multidimensional Impact of Cancer Risk Assessment (MICRA) [13] modified for genetic risk assessment. Six items (e.g., “feeling upset about your ClinSeq result(s)”) were ranked using a 4-point response: “never” (0), “rarely” (1), “sometimes” (3), and “often” (5). Averaged scores ranged from 0 to 5 with higher scores indicating greater test-specific distress (Cronbach α = 0.84).
Communication of results to children was assessed at 6 months by asking participants how many sons and daughters they told about their carrier results. Responses were dichotomized for analyses into either having told at least one child or none, and percentages were calculated as proportions of participants with at least one child. Communication of results to health care providers was assessed at 6 months by asking participants if they had disclosed their results to a health care provider.
Quantitative data analysis
Chi-square and one-way analysis of variance (ANOVA) were used to test for differences between education arms at baseline in dichotomous variables and continuous variables, respectively. Linear regression was used to test anticipated ambiguity as a moderator of the change in uncertainty from baseline to follow-ups. Two-way (split-plot) ANOVA was used to simultaneously test within-group differences in perceived personal uncertainties over time and between-group differences in perceived personal uncertainties across education arms; a lower-bound correction was applied when Mauchly’s test indicated that the sphericity assumption had been violated. Logistic regression was used to test associations with dichotomous outcome variables. Sensitivity analyses included comparison of results across education arms and across those who did and did not receive a VUS. The analysis was based on available data at each time point, and both intention-to-treat and per-protocol analyses were applied. Statistical analyses were conducted using SPSS (IBM Corp; Macintosh, version 20.0).
Power analysis
Calculations for two-tailed hypothesis tests assumed 80% power and a 0.05 alpha level criterion. The minimum detectable effect for associations between ambiguity and each domain of personal uncertainty was 3.1% of variance. The minimum detectable effect for the within-subjects comparisons of each domain of personal uncertainty was f = 0.13 at 1 month and f = 0.14 at 6 months. The minimum detectable effect for uncertainty as a predictor of distress was an odds ratio (OR) = 1.75, as a predictor of disclosure to children was an OR = 0.57, and as a predictor of communication of results to health care providers was an OR = 0.47. Power analyses were conducted using G*Power (Macintosh, version 3.1).
Qualitative data analysis
Participants in the genetic counselor arm asked content-related questions during the education disclosure session that were addressed by the genetic counselor. In contrast, participants in the web platform arm had no opportunity to ask questions. Education disclosure sessions with the genetic counselor were audiotaped and transcribed verbatim. A codebook was created after an initial review of questions asked by participants, and 107 (47%) randomly selected transcripts were coded by content and frequencies. Half of the questions were coded by a second investigator (reliability was high: 91% agreement). In addition, an open-ended question was asked immediately following the education disclosure sessions regardless of randomization: “What further questions do you have about your result?” These questions were also first- and second-coded for content and frequencies following the development of a codebook (reliability was high: 95% agreement). Qualitative analyses were conducted using NVivo 11 (QSR International).
RESULTS
Of the 571 members of the ClinSeq cohort meeting eligibility criteria, 462 consented verbally to participation in this trial and were randomized to receive results from either a web platform or a genetic counselor. All participants had completed baseline questionnaires following enrollment in the ClinSeq cohort, but the time elapsed from baseline to the 1-month follow-up ranged from 4 months to 4 years (x̅ = 1.9 years, SD = 0.7 years) due to ongoing enrollment. This time duration was approximately normally distributed and not significantly associated with any outcome variables. All but one participant received their allocated intervention, and 86% (n = 398) returned 1-month questionnaires and 85% (n = 392) returned 6-month questionnaires. Participant flow is shown in Fig. 1.
Fig. 1.
Trial flowchart. Participant (N = 462) flow by education arm from baseline to each of the follow-ups.
Participant demographics and clinical characteristics by education arm are shown in Table 1. This sample was predominantly non-Hispanic whites of high socioeconomic status who were married with at least one child. Participants received a median of two carrier results. Means and SDs of all variables by education arm are shown in Table 2. The randomization was effective in that there were no significant differences between arms in any demographic or variable measured at baseline, other than the education disclosure session being significantly longer in the genetic counselor arm (p < .001). Fidelity to the intervention was high (x̅ = 95%).
Table 1.
Participant demographics and clinical characteristics.
| Genetic Counselor (n=227) | Web Platform (n=234) | ||
|---|---|---|---|
| Characteristic | Dichotomous Classification | n (%) | n (%) |
| Gender | Male | 127 (55.9) | 119 (51.3) |
| Female | 100 (44.1) | 113 (48.7) | |
| Marital Status | Not in a marriage-like partnership | 49 (22.2) | 45 (20.3) |
| In a marriage-like partnership | 172 (77.8) | 177 (79.7) | |
| Household Income | Less than $100,000 per year | 52 (24.0) | 53 (23.9) |
| More than $100,000 per year | 165 (76.0) | 169 (76.1) | |
| Education | Less than a graduate degree | 77 (34.8) | 90 (40.2) |
| Graduate degree | 144 (65.2) | 134 (59.8) | |
| Race | White | 215 (94.7) | 211 (91.3) |
| Non-white | 12 (5.3) | 20 (8.7) | |
| Ethnicity | Hispanic or Latino | 2 (0.9) | 7 (3.0) |
| Not Hispanic or Latino | 224 (99.1) | 223 (97.0) | |
| Parental Status | No children | 52 (24.9) | 46 (22.9) |
| At least one child | 157 (75.1) | 155 (77.1) | |
| Results Returned | At least one “pathogenic” | 175 (77.1) | 174 (75.0) |
| Zero “pathogenic” | 52 (22.9) | 58 (25.0) | |
| At least one “likely pathogenic” | 70 (30.8) | 75 (32.3) | |
| Zero “likely pathogenic” | 157 (69.2) | 157 (67.7) | |
| At least one “VUS” | 122 (53.7) | 106 (45.7) | |
| Zero “VUS” | 105 (46.3) | 126 (54.3) | |
| Continuous Characteristics | mean (SD) | mean (SD) | |
| Total number of results returned | 2.4 (1.2) | 2.3 (1.3) | |
| Education disclosure session length in minutes | 27 (9.3) | 21 (10.6) | |
| Age in years | 63.2 (5.4) | 63.3 (5.7) |
Table 2.
Distributions of key variables.
| Variable | Genetic Counselor Armb | Web Platform Armb |
|---|---|---|
| Mean ± SD | Mean ± SD | |
| Anticipated ambiguity (1–5) | ||
| Baseline | 2.26 ±0.60 (n=223) | 2.16 ±0.58 (n=226) |
| Clinical uncertainty (1–5) | ||
| Baseline | 2.32 ±0.78 (n=223) | 2.35 ±0.77 (n=228) |
| 1 month | 1.83 ±0.89 (n=130) | 2.15 ±1.03 (n=126) |
| 6 months | 2.18 ±1.00 (n=180) | 2.47 ±1.01 (n=195) |
| Affective uncertainty (1–5) | ||
| Baseline | 3.09 ±0.77 (n=217) | 3.08 ±0.76 (n=228) |
| 1 month | 2.19 ±1.34 (n=130) | 2.34 ±1.27 (n=126) |
| 6 months | 2.37 ±1.28 (n=179) | 2.57 ±1.24 (n=192) |
| Evaluative uncertainty (1–5) | ||
| Baseline | 2.30 ±0.75 (n=220) | 2.29 ±0.85 (n=227) |
| 1 month | 1.56 ±0.79 (n=130) | 1.66 ±0.96 (n=126) |
| 6 months | 1.47 ±0.68 (n=179) | 1.63 ±0.82 (n=195) |
| Test-specific distress (0–5) a | ||
| 6 months | 0.11 ±0.30 (n=190) | 0.20 ±0.46 (n=200) |
aA floor effect was observed for test-specific distress at six months (median=0, interquartile range=0).
bSignificant differences observed between those who received results from a genetic counselor and those who received results from a web platform are denoted in bold.
Anticipated ambiguity of sequencing results
Participants reported moderate perceptions of anticipated ambiguity (i.e., limitations in the reliability, credibility, or accuracy) of sequencing results at baseline (x̅ = 2.21, SD = 0.59, n = 449). Anticipated ambiguity was significantly associated with all three types of personal uncertainty, accounting for 8% of variance in clinical uncertainty, 2% of variance in affective uncertainty, and 12% of variance in evaluative uncertainty at baseline.
Perceived uncertainty
Participants reported modest perceptions of all three types of uncertainty at baseline: clinical uncertainty (x̅ = 2.34, SD = 0.77, n = 451), affective uncertainty (x̅ = 3.08, SD = 0.76, n = 445), and evaluative uncertainty (x̅ = 2.29, SD = 0.80, n = 447). None differed significantly between education arms at baseline. Owing to the addition of the PUGS scale to 1-month questionnaires after the trial had commenced, ANOVA results that follow are limited to the last 256 participants to complete the trial. The pairwise comparisons from the split-plot ANOVA are shown in Fig. 2. Results were consistent across those who did and did not receive a VUS.
Fig. 2.
Split-plot analysis of variance. Differences in adjusted means by education arm for clinical uncertainty, affective uncertainty, and evaluative uncertainty at baseline, 1 month after education, and 6 months after education (n = 256). Error bars denote standard error; significance for the within-subjects comparisons is shown with respect to the aggregate sample, the genetic counselor arm (GC), or the web platform arm (WP) if p < .10 (^), p < .05 (*), p < .01 (**), or p < .001 (***).
Clinical uncertainty
A significant interaction between education arm and time was observed F(2,217) = 3.66 (p < .027, η 2 = 0.033), suggesting a small-to-medium effect size. Simple main effects analysis indicated a significant reduction in clinical uncertainty in the genetic counselor arm from baseline to one month (∆x̅ = 0.53, 95% CI: 0.29, 0.78), but not from baseline to six months (∆x̅ = 0.24, 95% CI: −0.02, 0.50). In contrast, there was no significant change in clinical uncertainty observed in the web platform arm from baseline to 1 month (∆x̅ = 0.16, 95% CI: −0.08, 0.41) or to 6 months (∆x̅ = −0.11, 95% CI: −0.37, 0.15). As such, participants who received results via the genetic counselor perceived less clinical uncertainty 1 month after the education disclosure session, but participants who received results via the web platform did not reduce their perceptions. Importantly, clinical uncertainty was significantly higher in the web platform arm than the genetic counselor arm at both 1 month F(1,218) = 9.59 (p < .002) and 6 months F(1,218) = 6.98 (p < .009). In addition, anticipated ambiguity moderated the association between results disclosure and clinical uncertainty (β = −.134, p < .033), suggesting that those who anticipated greater ambiguity of sequencing results at baseline experienced a greater reduction in clinical uncertainty from baseline to 1 month.
Affective uncertainty
The interaction between education arm and time was not significant F(1,214) = 1.01 (p < .317), and neither was the main effect of education arm F(1,214) = 1.82 (p < .179). The main effect of time was significant F(1,214) = 39.46 (p < .001, η 2 = 0.156), suggesting a large effect size. Specifically, a significant reduction in affective uncertainty was observed in the aggregate sample from baseline to 1 month (∆x̅ = 0.78, 95% CI: 0.53, 1.02) and from baseline to 6 months (∆x̅ = 0.69, 95% CI: 0.46, 0.92). As such, participants perceived less affective uncertainty at both follow-ups, an outcome that did not depend on mode of result disclosure. The association between results disclosure and affective uncertainty was not moderated by anticipated ambiguity of sequencing results.
Evaluative uncertainty
The interaction between education arm and time was not significant F(1,214) = 0.92 (p < .338). However, both the small main effect of education arm F(1,214) = 4.55 (p < .034, η 2 = 0.021) and the large main effect of time F(1,214) = 80.04 (p < .001, η 2 = 0.272) were significant. A significant reduction in evaluative uncertainty was observed in the aggregate sample from baseline to one month (∆x̅ = 0.69, 95% CI: 0.51, 0.87) and from baseline to 6 months (∆x̅ = 0.78, 95% CI: 0.62, 0.93). As such, participants perceived less evaluative uncertainty at both follow-ups; although the significance of this outcome did not depend on mode of result disclosure, differences between education arms were significant at 6 months F(1,214) = 6.11 (p < .014). In addition, anticipated ambiguity moderated the association between results disclosure and evaluative uncertainty (β = −.181, p < .004), suggesting that those who anticipated greater ambiguity of sequencing results at baseline experienced a greater reduction in evaluative uncertainty from baseline to 1 month.
Six-month outcomes
Test-specific distress
A floor effect was observed for distress at the 6-month follow-up: 304 participants (78%) reported zero distress. The proportion of participants reporting zero distress did not differ significantly between education arms at either follow-up. Clinical uncertainty at 6 months was associated with a significantly greater likelihood of reporting distress (OR = 1.65, 95% CI: 1.29, 2.11, p < .001), but affective uncertainty and evaluative uncertainty were not.
Communication of results
Most participants had communicated their results to their children at the 6-month follow-up: 220 parents (75%) reported having communicated results to at least one child. The proportion of parents who had communicated results to children did not differ significantly between education arms. Clinical uncertainty at 6 months was associated with a significantly lesser likelihood of communicating results to children (OR = 0.69, 95% CI: 0.53, 0.91, p < .008), but affective uncertainty and evaluative uncertainty were not. Communication of results to health care providers was lower at the 6-month follow-up: 111 participants (29%) reported having communicated results. The proportion of participants who had communicated results to health care providers did not differ significantly between education arms. Evaluative uncertainty at 6 months was associated with a significantly lesser likelihood of communicating results to health care providers (OR = 0.70, 95% CI: 0.51, 0.98, p < .035), but clinical uncertainty and affective uncertainty were not.
Qualitative findings
There were 1,230 questions asked of the genetic counselor by 107 participants (47%) during education disclosure sessions. Participants asked about the conditions caused by the variants reported, as well as treatments and prognosis for the associated conditions (n = 91). Less often, they also asked about the chances that their siblings and children may also be carriers (n = 61) and about genetic mechanisms (n = 50).
Participants from both education arms responded to an open-ended question immediately following the intervention. Two hundred seven (92%) of those randomized to the genetic counselor and 144 (62%) of those randomized to the web platform reported no residual questions. Among those who reported having residual questions in the web platform arm, the most frequent topic (19%) was about risks to relatives (n = 45): how the condition(s) would manifest in a relative, whether their relatives should undergo genetic testing, and to which relatives they should communicate their results. There were also questions about the specific conditions (17%, n = 40): inquiries into symptoms and treatments, as well as population frequencies. Additional questions pertained to, for example, factors that influence the expression of mutations (9%, n = 20). There was remarkable similarity between the content of questions asked during the genetic counseling session among those randomized to the genetic counselor and the residual questions among those randomized to the web platform.
Discussion
The novelty of this study lies in exploring the roles of clinical, affective, and evaluative uncertainties as key outcomes following the receipt of carrier results from sequencing. All three types of uncertainty were reduced following return of results to this post-reproductive sample, with the exception of clinical uncertainty among those in the web platform arm. As the exception may be explained by lack of an opportunity to clarify residual questions (e.g., with a genetic counselor), this suggests that the sequencing results themselves may reduce perceived uncertainties among recipients. Interestingly, as the reduction in clinical uncertainty among those randomized to the genetic counselor arm was no longer significant at 6 months, it may be that recipients of sequencing results would benefit from reinforcement of clinical information no matter the mode of education and result disclosure.
Results from this study differ from prior research that assesses perceived uncertainty as a monolithic construct [14] by specifying the types of uncertainties that may or may not be reduced following receipt of results from sequencing, which in turn allows for a more nuanced approach to management. For example, with respect to the first hypothesis, no differences in affective uncertainty were observed between education arms. In contrast, differences in clinical and evaluative uncertainties were observed: these types of uncertainty were reduced to a lower extent for participants in the web platform arm than for participants in the genetic counselor arm. These findings indicate that the mode of education and result disclosure may have a greater effect on participants’ uncertainties about indicated health behaviors (clinical) and the trustworthiness of results (evaluative) than their uncertainties about their psychosocial responses toward the results (affective), which may be attributable to the lower health threat of carrier results for post-reproductive adults. This hypothesis should be tested in future research examining web-based return of more potentially distressing results.
Consistent with previous research adapting the MICRA for carrier results from sequencing [15], test-specific distress was extremely low in this trial. Although the low level of distress minimized power to characterize its relationship with the three types of perceived personal uncertainties assessed, the association of elevated clinical uncertainty with an approximately 65% greater likelihood of reporting distress is consistent with a finding from a recent meta-analysis, which showed that illness-related uncertainty was associated with anxiety [3]. In contrast, affective and evaluative uncertainties were not significantly associated with a differential likelihood of self-reported distress. It seems reasonable that uncertainty about one’s affective response to results and uncertainty about whether one trusts sequencing results may be less distressing than clinical uncertainty about health implications and indicated health behaviors.
It has been reported previously that greater perceptions of anticipated ambiguity about sequencing results account for lower intentions to share results with family [16]. In this trial, anticipated ambiguity of sequencing results was found to moderate the changes in clinical uncertainty and evaluative uncertainty from pre- to post-result disclosure. This finding has not been previously reported to our knowledge, and the underlying mechanisms remain to be elucidated. It is possible that those who anticipated greater ambiguity at baseline regressed to the mean. Alternatively, greater anticipated ambiguity of sequencing results may reflect lower initial expectations of the ability of this technology to produce definitive results, and the receipt of any results might cause participants to perceive excessive certainty regarding their meaning or implications. It was also surprising to find that the level of perceived uncertainty did not differ between those who did and did not receive a VUS, though this is consistent with findings from a recent study applying the PUGS following return of hypothetical results [17]. More research is needed to test explanations of these findings.
Perceived ambiguity is a primary source of uncertainty in genome sequencing [1], and in this study it accounted for variance in all three types of personal uncertainty that were assessed. In turn, clinical uncertainty was associated with a lesser likelihood of communicating results to at-risk relatives, and evaluative uncertainty was associated with a lesser likelihood of sharing results with a health care provider. The latter may indicate a perception of lower utility of results from sequencing. Further research is needed to determine whether these findings generalize to web-based return of results in other populations or contexts and to adequately understand the role of expectations of test-related ambiguity and uncertainty in predicting health behaviors such as communication of results to at-risk relatives and health care providers.
The participants in the web platform arm reported more residual questions immediately after the education disclosure sessions, and the topics were remarkably similar to those asked of the counselor during the education disclosure sessions. Given that scores on knowledge of recessive inheritance were not different across arms [7], participants may have been seeking validation or clarification of the information. Again, although a definitive conclusion is not appropriate, in this case due to the qualitative nature of the analyses, these findings are appreciably consistent with the elevated clinical uncertainty in the web platform arm relative to the genetic counselor arm.
Limitations and future directions
The limitations of the RCT have been reported elsewhere [7]: the majority of the sample is post-reproductive, non-Hispanic whites who are highly educated and of high socioeconomic status, and results may not generalize to other populations. A replication study in a more diverse sample is underway. Specific to this analysis, participant age has been identified as a moderator of the relationship between illness uncertainty and management (i.e., information seeking) [3]. Therefore, younger adults learning carrier status from sequencing may manage perceived personal uncertainties differently whether or not they are making reproductive decisions. In addition, as the ages of participants’ children were not collected, it is possible that some children were not of child-bearing age; this may have affected rates of communication to children. Participants also may have had more residual questions than they reported as the open-ended format required recalling them. The results presented support evaluation of interventions to minimize residual questions and perceptions of uncertainty as an important endeavor for future application of web-based return of results from sequencing. Further reinforcement of information with clinical implications in a web platform (e.g., frequently asked questions) may boost participants’ ability to acknowledge and manage the uncertainties associated with the results of genome sequencing.
Conclusions
This study identified novel relationships between three specific types of perceived personal uncertainties (clinical, affective, and evaluative) and other variables that illustrate their importance in future genomic sequencing studies. This approach challenges previous research that treats uncertainty as a single monolithic variable and thus promotes a more nuanced appreciation of the diverse patient-centered uncertainties inherent in the receipt of results from sequencing. The findings advance our understanding of the role of specific perceived uncertainties as key outcomes. Further examination of uncertainties will be of paramount importance not only in replication and extrapolation of findings but also in understanding and managing the ever more common patient experience of receiving results from genome sequencing.
Acknowledgment
The authors thank Kristen Fishler for coordinating the study visits; Jennifer Johnston for creating the reports of carrier results; Cristofer Price and Niraj Trivedi for statistical consultation; Julia Fekecs for initial development of web platform graphics; Gillian Hooker for input into study design and constructs assessed in ClinSeq; John Patton and Ashlee Hulbert for assistance in the laboratory; William Fix for assistance with piloting the study; William Klein for contributions to the constructs assessed in ClinSeq; David Ng for his participation on the panel that assigned pathogenicity to variants returned. This research was supported by the Intramural Research Program of the National Human Genome Research Institute, National Institutes of Health.
Compliance with Ethical Standards
Primary Data: The findings reported have not been previously published, and the manuscript is not being simultaneously submitted elsewhere. Some of the data have been previously reported (Biesecker et al., 2017 [6], Biesecker et al., 2018 [7], Lewis et al., 2018 [12]). The authors have full control of all primary data and agree to allow the journal to review their data if requested.
Conflict of Interest: L.G.B. receives royalties from Genentech Corp., is an unpaid advisor to Illumina Corp., and receives honoraria from Wiley-Blackwell Inc. K.L.U., P.K.J.H., K.L.L., I.M.M., C.L.H., L.J.T., T.G.W., A.-D.N., M.T.F., G.G., E.T., and B.B.B. declare that they have no conflict of interest. This article does not contain any studies with animals performed by any of the authors.
Ethical Approval: All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Declaration of Helsinki and its later amendments or comparable ethical standards. Informed consent was obtained from all individual participants included in the study, which was approved by the National Human Genome Research Institute Institutional Review Board.
References
- 1. Han PKJ, Umstead KL, Bernhardt BA, et al. A taxonomy of medical uncertainties in clinical genome sequencing. Genet Med. 2017;19(8):918–925. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2. Gray SW, Martins Y, Feuerman LZ, et al. ; CSER Consortium Outcomes and Measures Working Group Social and behavioral research in genomic sequencing: approaches from the Clinical Sequencing Exploratory Research Consortium Outcomes and Measures Working Group. Genet Med. 2014;16(10):727–735. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3. Kuang K, Wilson SR. A meta-analysis of uncertainty and information management in illness contexts. J Commun. 2017;67(3):378–401. [Google Scholar]
- 4. Biesecker LG, Green RC. Diagnostic clinical genome and exome sequencing. N Engl J Med. 2014;370(25):2418–2425. [DOI] [PubMed] [Google Scholar]
- 5. Biesecker BB, Klein W, Lewis KL, et al. How do research participants perceive “uncertainty” in genome sequencing? Genet Med. 2014;16(12):977–980. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. Biesecker BB, Woolford SW, Klein WMP, et al. PUGS: a novel scale to assess perceptions of uncertainties in genome sequencing. Clin Genet. 2017;92(2):172–179. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. Biesecker BB, Lewis KL, Umstead KL, et al. Web platform vs in-person genetic counselor for return of carrier results from exome sequencing: a randomized clinical trial. JAMA Intern Med 2018;178(3):338–346. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8. Biesecker LG, Mullikin JC, Facio FM, et al. ; NISC Comparative Sequencing Program The ClinSeq Project: piloting large-scale genome sequencing for research in genomic medicine. Genome Res. 2009;19(9):1665–1674. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. Lewis KL, Han PK, Hooker GW, Klein WM, Biesecker LG, Biesecker BB. Characterizing participants in the ClinSeq genome sequencing cohort as early adopters of a new health technology. PLoS One. 2015;10(7):e0132690. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10. Bell CJ, Dinwiddie DL, Miller NA, et al. Carrier testing for severe childhood recessive diseases by next-generation sequencing. Sci Transl Med. 2011;3(65):65ra4. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11. Berg JS, Adams M, Nassar N, et al. An informatics approach to analyzing the incidentalome. Genet Med. 2013;15(1):36–44. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. Lewis KL, Umstead KL, Johnston JJ, et al. Outcomes of counseling after education about carrier results: a randomized controlled trial. Am J Hum Genet. 2018;102(4):540–546. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. Cella D, Hughes C, Peterman A, et al. A brief assessment of concerns associated with genetic testing for cancer: the Multidimensional Impact of Cancer Risk Assessment (MICRA) questionnaire. Health Psychol. 2002;21(6):564–572. [PubMed] [Google Scholar]
- 14. Mishel MH. Parents’ perception of uncertainty concerning their hospitalized child. Nurs Res. 1983;32(6):324–330. [PubMed] [Google Scholar]
- 15. Sanderson SC, Linderman MD, Suckiel SA, et al. Psychological and behavioural impact of returning personal results from whole-genome sequencing: the HealthSeq project. Eur J Hum Genet. 2017;25(3):280–292. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16. Taber JM, Klein WM, Ferrer RA, et al. Perceived ambiguity as a barrier to intentions to learn genome sequencing results. J Behav Med. 2015;38(5):715–726. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17. Hellwig LD, Biesecker BB, Lewis KL, Biesecker LG, James CA, Klein WMP. Ability of patients to distinguish among cardiac genomic variant subclassifications. Circ Genom Precis Med. 2018;11(6):e001975. [DOI] [PMC free article] [PubMed] [Google Scholar]