Abstract
Background. Newborn screening saves lives, but the way in which parents learn of a positive screening test is also important for adherence with treatment plans and avoidance of psychosocial complications. The first messages provided to parents may be particularly important for understanding, especially when the infant is found to be a heterozygous carrier for sickle cell hemoglobinopathy (SCH) or cystic fibrosis (CF). This study investigated the prevalence of “initially misleading” communication, defined as the inclusion of 1 of 55 “bad-news” content items (eg, the screening test is positive) before any of 39 “good-news” content items (eg, the infant is healthy, normal, a carrier, or otherwise without problems).
Methods. As part of a larger study of the content of counseling after newborn genetic screening, we used a quantitative, explicit-criteria method to abstract 59 transcribed conversations between pediatric residents and standardized parents of an “infant” who was found through newborn screening to carry either CF or SCH.
Results. Of 59 transcripts, 41 were found to be misleading (at least 1 bad-news content statement before the first good-news content statement). There were significantly more misleading likely-CF-carrier than SCH-carrier transcripts (89.7% vs 50%). Among the misleading transcripts, the mean number of misleading statements was 5.5. The mean distance between the first bad-news and first good-news statements was 28.1 statements (20.5% of the total duration of counseling).
Discussion. The high prevalence of misleading content and the time lag before clarification does not bode well for parental understanding of infant carrier status. Future projects should improve curricula for training programs and develop quality-assurance efforts for community clinicians both to improve parental understanding and help assuage society's fears about the safety of genetic screening technologies.
The purpose of newborn screening is to identify metabolic and genetic diseases early so that affected infants may receive treatment and avoid death or disability.1 Despite recent technologic advances,2,3 however, it remains true that one of the most important events in newborn screening is the way in which parents learn of positive results. Effective communication is necessary for the sharing of accurate information between parents and clinicians and also promotes adherence with follow-up testing and treatment regimens.4–6 Communication is even more important for families of infants identified by newborn screening as heterozygous “carriers” for sickle cell hemoglobinopathy (SCH) or cystic fibrosis (CF). For families of infants found to be carriers, effective communication may be the only way to “ensure more good than harm,”7 especially because the biomedical benefit of newborn screening for carriers is small compared with their risk for harm from anxiety8,9 or psychosocial complications such as impaired parent-child bonding,9 negative selfperception,10 or stigmatization.10–12 Historically, psychosocial complications have often resulted from misunderstandings about what it means to be a carrier, and misunderstandings have often followed ineffective communication with families and the community.11 In fact, ethical concerns about carrier families' experiences with psychosocial complications are at the root of society's doubts about expansions in genetic screening.13,14
Just as communication is important to newborn screening, the first few messages presented are important to parental understanding. This is in part because of the primacy effect, a term for the well-studied observation that communication recipients recall the first message presented in counseling better than messages presented later.15 For example, when newborn screening identifies an infant as a carrier, the ideal clinician behavior for communicating carrier results may be to begin with a clear positive message that will be most likely to lead to understanding of the infant's essentially benign situation. For example the clinician could begin for an SCH carrier with: “It is important for you to know that your infant is healthy now and should continue to be healthy.” This good-news-first approach contrasts with what may be the more intuitive chronological approach of beginning with basic information first (eg, “Your infant's sickle cell screening test was positive”) and then expanding to related messages such as an explanation about newborn screening and SCH, clarifying that the infant is actually a carrier, explaining what a carrier is and possibly about the reproductive implications of carrier status, and then finally emphasizing that the infant is healthy. We called this building-up approach “initially misleading,” because it presents background information in such a way that it could be misconstrued by parents as suggesting that something is wrong with their infant. We decided to use data from our larger study of pediatric residents and communication “quality indicators” after newborn genetic screening16 to determine the prevalence of initially misleading communication (bad news before good news) versus nonmisleading communication (good news before bad news). Our hypothesis was that initially misleading counseling would be found in a majority of the transcripts.
METHODS
This study used data from a larger project designed to develop and refine a system for assessing the content domain of the quality of communication after newborn genetic screening.16 The project used an explicit-criteria instrument to abstract transcripts of conversations between pediatrics residents and standardized parents (SPs) of a fictitious infant with a positive newborn-screening test for the SCH trait or single-allele CF status. Methods and materials were approved by the institutional review boards at Yale University, Northwestern University, and the Medical College of Wisconsin.
Participants and Setting
The participants were all Yale University pediatrics residents beyond their first year who were scheduled in groups of 4 to attend a workshop on communicating heterozygous results of newborn-screening tests for CF or SCH. Each workshop consisted of (1) a 10-minute review of newborn screening, CF, SCH, and autosomal recessive inheritance (any advice about how to discuss results was excluded), (2) 2 interviews with 1 SP each as described below, and (3) an interactive session on the brief communication of newborn-screening results. The workshops were part of the official curriculum, but residents gave informed consent and were free to decline the use of their counseling tapes in research.
Procedures
Counseling Task
The counseling task was to communicate the results of the infant's screening test to the SP. A handout provided the resident with the test result and some social background data but did not prompt them on how to discuss the result. In the “SCH-carrier” scenarios, the handout reported a screening result of hemoglobin F, A, and S, a result that had been presented in the review session as definitely indicating that an infant is an SCH carrier. In the “likely CF-carrier” scenarios, the handout reported an elevated screening immunoreactive trypsinogen level and the presence of 1 ΔF508 mutation with no multiallele follow-up screening. The review sessions had presented such a result as suggesting that the infant was probably a carrier but still had up to a 5% to 10% chance of having the disease because of an undetected allele.3,17
SPs
Six different SPs worked on the project; all were women and chosen to plausibly depict the age and ethnicity of a mother of an infant with CF or SCH. Each participant was audiotaped in randomized, double-crossover fashion in 2 of 4 possible SP encounters varied by the 2 diseases and 2 script types. The 2 script types were included for a separate validation study of the new script type, the Brief Standardized Communication Assessment (BSCA). The BSCA is a streamlined version of a standardized patient script that simplifies SP role instructions, requires less training, discourages improvisation, and encourages counseling to continue as long as possible with nonleading questions such as: “Is there anything else I should know?” The comparison script type was designed to resemble the methods used in most educational settings, which seek to make the encounter realistic by asking the patient to use acting skills and improvise a 2-way dialogue.18 To help comparison, all SPs in both scripts were coached to not appear anxious and to avoid leading questions of the sort that “helpful” SPs often ask. They were also asked to pretend to understand whatever the resident said, helping us to focus on information provided rather than the resident's ability to reclarify (a critical skill, but not the subject of this study).
Data Handling
For the larger project we audiotaped 32 residents in 2 interviews per resident. The audiotaping equipment failed in at least part of 5 interviews, and so the final sample consisted of the remaining 59 transcripts. The tapes were transcribed verbatim and proofread for accuracy by a board-certified pediatrician (M.H.F.). To provide a unit of distance within transcripts we used a sentence-diagramming process to divide the transcripts into individual concepts or ideas (“statements”), each with 1 subject and 1 predicate. To standardize comparison of distance across transcripts, a “percent-statement” number was calculated by dividing the raw statement number by the total number of statements in that interview. For example, statement 20 in a 200-statement interview would be referred to as a percent-statement number of 10% or 10% of the way into the interview. The transcripts then were abstracted statement by statement by 2 reviewers (M.H.F. and A.L.) using an explicit-criteria coding system patterned after the chart-abstraction technique used in quality-improvement studies19 except that instead of a medical record the reviewers abstracted data from a communication transcript. All transcripts were reviewed and abstracted independently by both authors. Every third transcript was discussed after abstraction to ensure quality control and consistency, following the suggestion by Feinstein.20 Interabstractor reliability, calculated by using an adapted version of Cohen's method,21 found a κ coefficient of .84 for the abstractors' identification of a content statement versus a noncontent statement and .65 for perfectly identifying individual content codes.16
The data dictionary for the entire content project contained 503 content-of-communication codes organized hierarchically by subject. To adapt the resulting data set for the current analysis, the data dictionary was condensed down to “bad-news” codes (content that parents could have easily misinterpreted to mean that the infant is sick or that there is something wrong) and “good-news” codes (content naming the infant to be a carrier, healthy, normal, or otherwise without problems). This process resulted in 39 good-news content codes and 55 bad-news content codes (see Appendix). These codes then were used as filtering criteria in the existing database to identify the first good-news content code for each transcript. The final data set for this article therefore consisted of content codes up to and including the first good-news content statement, in order according to percent-statement number.
Statistical Analysis
Statistical analysis was performed by using Excel (Microsoft Corp, Redmond, WA) and JMP software (SAS Institute, Cary, NC). Data were analyzed by using 1-way analysis of variance for continuous responses for categorical variables and the χ2 test for grouped categorical responses.
RESULTS
Descriptive data on the participants (Table 1) were similar to those of the population of Yale University pediatric residents at the time of the study. Interviews lasted an average of 9.8 minutes (SD: 4.2; range: 5–20) and contained an average of 166 separate concepts or statements by the pediatric resident (SD: 81; range: 65–401). Using the percent-statement system for comparison across transcripts, the average position of the first good-news statement was 24.8% of the way into counseling, leaving an average of 21.5 statements in place before the first good news that the infant is healthy, normal, a carrier, or otherwise without problems. This distribution varied widely across transcripts, with a moderate skew to the greater (skewness: 1.6).
Characteristics of Participating Pediatrics Residents
Misleading Transcripts
All but 1 transcript contained both good- and bad-news content items, but 41 of 59 transcripts (69.5%; P < .005 [χ2]) included at least 1 bad-news content statement before their first good-news content statement, the condition necessary for our “misleading” designation. As illustrated in Fig 1, there were significantly more misleading likely CF-carrier transcripts (26 of 29 [89.7%]) than misleading SCH-carrier transcripts (15 of 30 [50%]; P < .001 [χ2]). No statistical difference was observed in the number of misleading transcripts by age or year in residency, but there was a nonsignificant trend for gender; male residents were more often misleading (81.3%) than female residents (63.4%) (P = .18 [χ2]).
Initially misleading and nonmisleading transcripts by infant's genetic status.
Misleading Statements
Only 10 of the 39 good-news content items that we had identified a priori (Appendix) were observed in the first good-news position (Table 2); only 11 of the 55 bad-news content items were observed before the first good-news content item (Table 3). Most misleading statements concerned the screening-test result itself, although some residents also commented about what might happen in the infant's future (Table 3). The first misleading statement in 10 of 41 transcripts (30.8% of likely CF-carrier transcripts and 13.3% of SCH-carrier transcripts) implied that the infant had a positive test for a specific disease without any mention of carrier status (Table 4). For example:
“One of the tests that came back was for sickle cell disease . . . [followed 8 statements by] So, based on Natalie's blood work, um, it looks like she does not have sickle cell disease but that she does carry 1 of the genes or the components of it.”
Misleading-Content Items
First Good-News Content Items
Bad-News Content Items Appearing First in Misleading Transcripts
Misleading and Good-News Content Code Names
and
“On Nathan's screen what came positive was the test for cystic fibrosis … [followed 35 statements later by] … what is possible is that he is a carrier and not actually someone who is going to be sick.”
Other initially misleading-content items were less specific, including 8 transcripts beginning with a statement that the newborn-screening test was positive (or in 1 case “concerning”). The remaining 23 misleading transcripts (56.1%) began with a simple, nonspecific statement that there are test results to discuss (Table 4), as in the following example:
“There are some test results that came back … [followed 23 statements later by] But, um, the one that came back positive for Natalie is for sickle cell trait.”
A separate analysis was performed of the total number of content statements (good, bad, or otherwise) between the initial misleading statement and the first good-news statement. This was done to ensure that the nonspecific “there is a test result (not otherwise specified)” statement was not being immediately followed by good news, which would have led to an overestimate of the number of misleading transcripts. This analysis found a mean lag of 29 content statements of any type in between the initially misleading statement and the first good-news statement. With the percent-statement–numbering system, this corresponded with a mean relative distance of 20.5% of the transcript (SD: 18.6%); that is, there was an average time lag of one fifth of the total number of statements in the transcript between the first misleading/bad-news statement and the first good-news statement. No significant difference was observed for the type of initial misleading content; the mean lag for “there is a test result (not otherwise specified)” of 30 statements was only slightly >29 statements for the “positive test result” and 28 statements for the disease-specific content.
The density of misleading statements was also of relevance, because we were stipulating that the misleading statements would have a dose-response effect on parental misunderstanding. Among the misleading transcripts, the mean number of misleading statements was 5.5 (SD: 4.1; maximum: 18). There was substantial variation in the number of misleading statements (Fig 2); the likely CF-carrier transcripts contained significantly more such statements than the SCH-carrier transcripts (1-way analysis of variance: P < .0001). No significant differences were detected in the number of misleading statements by gender, age of the resident, or year in residency; power was too low to assess equivalence, but differences between groups were slight enough that for 90% power a least-significant number of 1500 observations would have been required.
Number of misleading statements by disease scenario.
Ratio of Bad News Versus Good News Over the Entire Transcript
We added this final analysis based on the consideration that a large amount of good-news content could compensate for initially misleading counseling (a “damage-control” strategy). We therefore re-tallied all good-news and bad-news content codes regardless of where they appeared in the transcript. Excluding the single transcript with no good-news statements, we identified 25 transcripts (41%) with more bad-news than good-news content statements; the average ratio for bad news to good news was 1.7/1 (SD: 2.3/1). Finally, we examined the ratio of bad news before versus bad news after the first good-news content statement; excluding 6 transcripts that included all the bad news before the good news, this average ratio was 0.9/1 (SD: 1.3/1).
DISCUSSION
The aim of this study was to investigate the frequency of initially misleading counseling by pediatric residents after newborn genetic screening for CF and SCH. Our most disconcerting discovery was that more than half of the SPs in this sample were provided initially misleading information, much of which implied that there was something wrong or that the infant had a disease. In all but 1 of the transcripts the resident corrected the initial misimpression, but it took an average of 28.1 statements (20.5% of the way into counseling) before the parent was told that the infant would most likely be fine. Misunderstanding among parents of carrier infants could be tempered by a damage-control strategy of emphasizing good news later, but in this sample almost half of the misleading transcripts included more bad-news content than good news. Of course, parents of carrier infants are free to seek out more information or to make their own interpretations, but it seems likely to us that many parents who are given misleading information could develop the incorrect idea that their infant has the actual disease.
Our focus on good news, bad news, and misleading statements may be unfamiliar to some pediatricians or genetic counselors, but it is consistent with well-established communication theory, extensive prior research such as Ley's studies of the primacy effect,15 and with the widespread truism that “first impressions count.” Seen another way, initial messages are important to effective communication because they form a foundation for comprehension of subsequent messages, an effect that has been modeled successfully in schema theory.22 Schema theory speculates that recipients construct complex mental structures of ideas and relationships (“schemas”) in response to incoming messages.23 Graphically, schemas often are portrayed as a series of nodes and associations or links, rather like drawings of the atoms and chemical bonds that make up a molecule. For counseling of carriers after newborn screening, each new idea is incorporated as a new node and is best recalled if it can be associated with the rest of the schema.22 If early messages lead to an inaccurate schema (eg, “my infant has CF”), a poor foundation is laid for subsequent schema building so that understanding may be impaired. Misunderstandings may then contribute to greater anxiety, guilt, and misinformed self-images and decision-making.11 Indeed, some studies have found that parental feelings of fear, worry, and shock can be attributed to the way in which they were told about their newborns' screening result.24–27
Comparing results for different diseases, we found significantly more misleading likely CF-carrier transcripts than misleading SCH-carrier transcripts (89.7% vs 50%; P < .001 [χ2]). Part of this difference may be because of the complexity and ambiguity of the single-allele result on an immunoreactive trypsinogen/DNA-screening test, which suggests that the infant is probably a carrier but will need a sweat test to verify the lack of an unmeasured allele.3 Because in such uncertain circumstances the clinician cannot truthfully say that the infant is definitely free of disease, we included in our list of good-news content items several codes for provisional statements such as: “Infant may be a CF carrier.” In fact, these codes were observed later in several transcripts but were never seen early enough to prevent the misleading designation. There were no apparent differences in the misleading phenomenon by age or year in residency; a small trend for gender in misleading transcripts was not corroborated by the more specific comparison of number of misleading statements. A larger sample may someday demonstrate a significant difference by gender, especially given the preponderance of female residents in our sample.
Although this study was limited by its small sample size, we were pleased with the ease and reliability of the explicit-criteria approach to measuring this aspect of communication quality. Qualitative methods could have provided a richer description of the complex interchange between clinician and parent but would have had limited reliability and will be prohibitively expensive for our next goal of measurement on a population scale.16 Future studies with a more focused data dictionary and fewer variables may increase the reliability further and allow more time for separate exploratory research projects. Our data collection was limited to a single institution's residency program and thus may not be representative of other programs or of community pediatricians. On the other hand, we felt that residents would be ideal for this initial study, because they may be near the peak of their knowledge as generalists. SPs may be artificial, but using SP encounters avoids logistic, privacy, and consent difficulties associated with audiotaping real parents. We also feel that the Hawthorne effect (change in subject behavior because of a sense of observation) has been helpful for our study, especially because competence is felt to be a prerequisite of performance.28 Finally, our methods focus on clinician communication processes rather than parental communication outcomes, but our efforts to develop process measures will be critical to future efforts to link processes and outcomes. In the meantime, we maintain that effective communication is necessary (and perhaps, in some cases, sufficient) for parental understanding.
The implications of this study fit well with our efforts to assess and improve communication quality after newborn genetic screening for SCH, CF, and eventually the many tests promised by the Human Genome Project.29 We have chosen to focus on counseling in primary care, because the demand for communication services will likely continue to exceed the supply of genetic counselors,30 and counseling by primary care clinicians has been criticized by families27 and public health officials.31 Physician communication skills are supposed to be addressed during training,32 but after graduation there seem to be few opportunities for continuing education or feedback. Thus, the method developed for this article may be useful as a new “communication quality indicator” for explicit reviewers to abstract from transcribed encounters to determine if clinicians are falling into the misleading-communication trap. As we reported elsewhere,16 such an assessment tool can be an inexpensive and quantitatively reliable way to assess communication quality, target interventions to those who need them the most, and track results after projects are complete. Future projects should attempt to assess communication in other screening situations, including with infants diagnosed with a disease or infants with false-positive results for tests such as phenylketonuria or congenital adrenal hyperplasia.
Several additional comments could be made about counseling by individual clinicians. First, it seems both inefficient and potentially ineffective to begin with misleading bad news when counseling about a carrier result, especially if the good news has to be introduced later as damage control. Beginning with good news instead does not necessarily conflict with recommendations for a “warning shot,” or opening bad news with nonspecific, preparatory statements such as: “I have bad news,”33 but clinicians wishing to avoid misinterpretation may wish to precede or immediately follow the warning shot with the best news possible under the circumstances. Our communication quality indicator for good-news placement could then be used to assess the timing of when good news is introduced to the parent.16 Finally, we should clarify that this study should not be used to support the use of the question, “Do you want to hear the good news or bad news first?” Querying preferences for counseling style is good policy in risk communication,34 but as used in common conversation, the “good news or bad news?” question may be perceived as flippant or disrespectful and thus would not be appropriate for informing parents about a positive newborn-screening test.
CONCLUSIONS
This study found that the initially misleading phenomenon was unfortunately quite common in our resident sample and that the SPs in this sample were not provided clarifying information until, on average, one fifth of the way into counseling. Future studies should be done in larger, community-based populations of pediatricians, but for now this study seems to suggest a need for improved training curricula and efforts to assess communication quality in the community. More families will then have access to clear, effective communication services after newborn screening. Improved communication, in turn, will lessen the risks for anxiety, stress, stigma, and misunderstanding associated with the discovery of carrier status.
Acknowledgments
Dr Farrell is supported in part by National Heart, Lung, and Blood Institute grant 7K01HL072530.
When this project was begun, Ms La Pean was a master's student at the genetic counseling program at Northwestern University in Chicago, Illinois, and Dr Farrell was an assistant professor at the Yale Primary Care Medicine Research Center in Waterbury, Connecticut. We are indebted to Lynnea Ladouceur, MPH, Jeffrey Stein, MD, Laura Farrell, MAT, and Kelly E. Ormond, MS, CGC, for excellent advice about many aspects of this project.
Footnotes
- Accepted April 12, 2005.
- Address correspondence to Michael H. Farrell, MD, Internal Medicine and Pediatrics, Center for Patient Care and Outcomes Research, 8701 Watertown Plank Rd, PO Box 26509, Milwaukee, WI 53226-0509. E-mail: mfarrell{at}mail.mcw.edu
A much earlier version of this analysis was first presented as a student thesis by Ms La Pean at Northwestern University (Chicago, IL).
No conflict of interest declared.
REFERENCES
- Copyright © 2005 by the American Academy of Pediatrics
In this issue
