Pediatric Residents' Responses That Discourage Discussion of Psychosocial Problems in Primary Care
http://www.100md.com
《小儿科》
Department of Health Policy and Management, Johns Hopkins Bloomberg School of Public Health, Baltimore, Maryland
Department of Medicine, Sinai Hospital, Baltimore, Maryland
Société Psychanalytique de Paris, Paris, France
ABSTRACT
Objective. Studies spanning nearly 4 decades demonstrate that doctors ignore or dismiss many patient bids for discussion of psychosocial topics. We sought to understand characteristics of doctors, patients, and visits in which this occurs.
Methods. Reanalysis of 167 audiotapes from 2 studies of parent-doctor communication in a pediatric residents' continuity clinic was performed. Tapes included visits by 100 mothers or female guardians to 55 residents who were the children's primary care providers. Coders identified mentions of psychosocial topics and noted the content and the doctor's response. Responses were classified with an adaptation of a previously described, psychoanalytically derived typology of avoidant or discouraging responses.
Results. Discouraging responses occurred in 34 (77%) of 44 discussions that involved corporal punishment and 51 (34%) of 64 discussions that involved other psychosocial topics. The particular topic (parent/family versus routine parenting issue) and how the topic was framed (as a problem versus simply mentioned) were associated with doctors' discouraging responses (OR: 3.07; 95% confidence interval: 1.56–6.05; and OR: 7.57; 95% confidence interval: 3.50–16.44; respectively). Discouraging responses were not related to the doctor's gender, parent's ethnicity, length of the parent-doctor relationship, or doctor's overall interview style (patient-centeredness). Discouraging responses to routine problems tended to be dismissive, but 41% of discouraging responses to parent/family problems were failed attempts to provide advice.
Conclusions. Discouraging responses seem to be related less to doctor or patient characteristics than to the type and acuity of the psychosocial topic. These responses may originate with doctors' discomfort with particular subject areas and thus might be approached with training that combines communication and emotion-handling skills with clinical tools such as Bright Futures in Practice: Mental Health or the International Classification of Diseases, 10th Revision, Primary Care.
Key Words: mental health primary care resident training
One persistent difficulty in pediatric primary care is the contrast between the high prevalence of parents' concerns about child behavior or development and the limited extent to which such concerns are discussed with doctors. The analysis by Horwitz et al 1 of pediatric visits in the New Haven, Connecticut, area found that 80% of parents of 4- to 8-year-old children thought that they had a psychosocial concern appropriate to discuss with their pediatricians but only approximately one half actually did so. In adult primary care visits, patients often approach psychosocial issues tentatively, starting with hints or feelers that they hope doctors will recognize. 2
Doctors may contribute to these hesitations consciously or unconsciously. Studies in both adult and pediatric primary care, spanning several decades, found that doctors ignore or dismiss the majority of patients' hints and disclosures of emotional distress. 2–9 Some patients may take these responses as a message that, in the future, the doctor would rather not hear about similar issues. 10,11
Many things may lead doctors to avoid discussion of patients' distress. Doctors may be troubled by the emotions involved and by feeling that they lack the skills, time, or referral resources required to manage emotional problems. 12–16 Clinicians also may chose to distance themselves from patients whose needs they see as going beyond what is expected or reasonable for their specialty or type of interaction. 17
Although previous studies documented the extent to which primary care providers avoid discussion of patient psychosocial concerns or engage in discussions that fail to provide advice, we lack an understanding of the kinds of concerns that evoke avoidance and of which doctors are most likely to be avoidant. Understanding when and why clinicians become avoidant could facilitate development of training, structural alterations to practice settings, and incentives that might promote better care of emotional issues. To that end, Hadjiisky et al 18 interviewed clinicians from a variety of backgrounds (social workers, day care center workers and directors, nurses, and physicians) who had encountered a situation widely acknowledged to promote discomfort among professionals, ie, interaction with a family for which there was suspicion of child abuse. Those investigators asked clinicians to describe their feelings and responses and then used psychoanalytic theory to classify and to interpret the reactions the clinicians reported. Psychoanalytic theory posits that clinicians develop emotional responses to patients and their life situations, just as patients develop emotional responses to their clinicians and to what is discussed during treatment. When these responses are uncomfortable, clinicians, like their patients, can manage them in ways that are more or less adaptive. 19 Hadjiisky et al 18 identified 3 main mechanisms clinicians used to distance themselves from the unpleasant emotions associated with cases of suspected abuse. First, they might reframe or reexplain what they had heard, to make it seem as if it should not be labeled as abuse or to make it seem less severe ("doubt" or denial of reality). Second, they might acknowledge the severity of the problem but propose an overly simplistic solution, as if their recommendation was sufficient to make everything better ("all-powerfulness"). Finally, they might rationalize inaction or silence by saying to themselves that the problem was insoluble ("helplessness").
We developed a system to identify apparent manifestations of the distress responses described by Hadjiisky et al 18 in audiotape recordings of primary care visits between pediatric residents and their patients' mothers. We hypothesized that these responses would be more common when mothers presented residents with psychosocial issues that were outside the expected content of a pediatric visit or that were presented as problems, that is, associated with potential or overt emotional distress. We also wanted to explore the influence of the length of the mother-doctor relationship, the doctor's gender and "patient-centered" style of communicating, and the mother's ethnicity (Fig 1). Female physicians are noted to be more concerned with psychosocial issues, 20 and some studies find that longer patient-doctor relationships increase trust and satisfaction, 21 as does doctors' use of a patient-centered style of communication. 22 In the United States, black patients report difficulty communicating with doctors of different ethnicities. 23,24 Finally, we hoped that the classification of responses described by Hadjiisky et al 18 would provide us with clues about the mechanisms behind residents' responses and thus lead us to hypotheses about possible interventions.
METHODS
Clinical Setting
Data came from 2 studies performed 8 years apart in the same pediatric residents' continuity clinic. 11,25 Both studies were approved by the Joint Committee on Clinical Investigation of the Johns Hopkins Medical Institutions, and the subsequent reanalysis was approved by the Committee of Human Research of the Johns Hopkins Bloomberg School of Public Health. Both studies involved asking parents and residents for permission to audiotape scheduled well-child and follow-up visits.
Initial Coding
Tapes were initially coded with the Roter Interaction Analysis System, 26 a scheme for classifying talk in clinical conversations that identifies the purpose of the talk (eg, question or information giving), the speaker, and the intended recipient. Doctors' patient-centeredness was calculated as the sum of all doctor talk classified as giving information (medical and psychosocial), asking psychosocial questions, showing empathy, giving reassurance, and building a partnership. In our previous work, this formulation of patient centeredness was associated with higher levels of parents' psychosocial talk. 27
Visits Included in the Analysis
In the first study, 25 coders identified 44 of 234 visits as including discussion of physical punishment. In 26 of the 44 visits, the parents stated that they used physical punishment; they mentioned its use as a possibility in 13 visits, and there was some other discussion of discipline that raised concerns about its use in the remaining 5 visits. Twenty-nine doctors provided care in these visits; 20 (69%) were female, 27 were white, 1 was black, and 1 was Asian American. Eleven doctors were in their third year of training, 10 were in their second year, and 8 were at the end of their first year. In these visits, the majority of children (33 of 44 children [75%]) were 3 years of age, and the oldest was 10 years of age. Twenty (45%) of the 44 mothers had a high school education or some college, and all except 4 were black. The mothers' average age was 26.6 years (range: 16–68 years). Mothers were asked to estimate how many times they had seen the same doctor previously. The average estimate was 11, with a range of 0 to 16.
In the second study, 11 coders identified 164 visit segments in 123 visits (of a total of 692 visits) containing discussion of a psychosocial topic (Table 1). Coders grouped the topics into categories (routine parenting, parents' relationships with partners or parents, stresses related to a child's medical condition, parents' concerns about their own schooling, job, or child care arrangements, and parents' substance abuse or legal problems) and noted whether the topic was simply mentioned or was framed as a problem for the parent. Simple mentions included parents' passive responses to doctors' questions and explicit statements that there was no problem. An example of an interchange with a passive parent answer is as follows.
Doctor: And things with you are going OK
Parent: Mm.
Doctor: Good, glad to hear it.
An explicit statement of "no problem" is as follows.
Doctor: And how are you guys getting along [now that the infant is here]
Mother: Good.
Doctor: You getting help from dad
Mother: [pause] Yeah.
Framing as a problem included hints that there might be a problem, statements of parents' increased vigilance, explicit disclosures of parents' distress or concern, and explicit requests for help by parents. A hint from a parent might be as follows.
Doctor: Are you having fun with him
Parent: He's a handful.
Doctor: What's that
Parent: He's a handful.
Doctor: Yeah, he is.
A parent's disclosure of distress or a problem would be as follows.
Parent: I'm just having a lot of problems 'cause I graduated from truck driver school and got a job offer.
Doctor: Oh, really
Parent: But I don't have anybody to leave the kids with.
An example of an explicit request for help is as follows.
Parent: Is there a family counselor here Because I have this problem. Right With her.
We approached reliability of coding for topic and problem type by having initial discussions about groupings and definitions, openly discussing several examples to develop a coding manual, and then having 2 investigators (S.L. and L.S.W.) independently code 25 parent-doctor psychosocial interchanges. Agreement between the 2 raters ranged from 88% to 100% for the 7 topic categories and from 84% to 100% for the 6 problem/distress categories.
All of the children in visits in the second study were infants (<1 year of age). Because this was a longitudinal study, the 123 visits involved only 56 mothers, of whom 50 (89%) were black and 6 were white. Mothers ranged in age from 17 to 64 years (mean: 26.1 years; range: 17–64 years). Most (39 of 56 [70%]) had finished high school or had taken some college-level courses. Twenty-six doctors were represented; 23 (88%) were white, 3 identified themselves as Asian, and 20 (76%) were female. Fifteen doctors were in their first year of training, and 11 were in their second year. In this study, the original goal was to recruit parent-doctor dyads as close to the first newborn visit as possible and then to record subsequent visits over the course of a year. Therefore, the visits in this sample ranged from the first to the eighth meeting between the doctor and patient (median: second visit).
Coding of Doctors' Discouraging Responses to Psychosocial Topics
We developed an initial set of codes and definitions by transcribing excerpts from visits with discussions of physical punishment. We chose these excerpts because of their similarity to the situations in the original work by Hadjiisky et al 18 and because physical punishment is a subject toward which attitudes are known to vary among doctors. 28 Two of the authors (E.H. and L.S.W.) reviewed the transcripts independently and then together, matching clinicians' responses to the broad categories of reactions noted in the interviews with clinicians by Hadjiisky et al. 18 On the basis of these discussions, we developed 9 categories of responses, 4 discouraging psychosocial discussion and 5 encouraging discussion (Table 2). The 4 discouraging responses corresponded to the 3 broad categories described by Hadjiisky et al 18 plus a category of seemingly ignored parent remarks. The 5 categories of responses that seemed to encourage disclosure particularly well (openly seeking help from other sources, building alliance with empathy, reducing the crisis without minimizing, bolstering credibility without omnipotence, and seeking additional information from the parent before giving advice) corresponded to elements of doctors' problem-identification and emotion-handling skills that were shown to help reduce psychosocial distress among adult primary care patients. 29
Another pair of authors (S.L. and L.S.W.) then applied these codes to excerpts from audiotapes. They assigned codes independently and then met to discuss disagreements and to modify category definitions. When definitions were finalized, one of the authors (S.L.) coded psychosocial discussion segments from study 2, with the other author duplicate-coding a 15% sample; agreement was >90% for each of the doctor response categories. A third author (J.A.) learned the system later. When she and another author reached 100% agreement coding independently, the third author coded visits from study 1.
The coding process involved going to the place in a recording where the target discussion took place and listening to it 1 times, paying attention to tone of voice as well as the words spoken. Tone and context were particularly useful in labeling doctor responses that in isolation might have seemed ambiguous. For example, "I wouldn't worry about it," might be labeled as dismissive and discouraging of discussion if it was said curtly and without any accompanying explanation. In contrast, it might be labeled as empathetic and positive if it was said warmly and after an explanation and an opportunity for the parent to make clear her concerns.
Not all responses were classified as discouraging or particularly encouraging. Responses classified as neutral were those that gave appropriate information but did not make additional attempts to explore the parent's feelings or to provide empathy. A neutral response to a disclosure of fatigue related to child sleep problems might be as follows.
Parent: I'm feeling really wiped out because I'm going in to check every time I hear him cry at night.
Doctor: At this age it's OK to let them cry for a few minutes before going in; they need to learn how to wake up and settle themselves down. How's he feeding
Coders were allowed to make their own determination of when a discussion ended and were asked to assign the single response code that seemed to fit best. The final coding process thus involved (1) initial coding of tapes to mark locations of psychosocial discussion, (2) study of the coding manual (descriptions of the codes, with examples), (3) training to agreement with someone already familiar with the system, and (4) independent coding, with quality checks.
Statistical Methods
Issues of statistical independence complicated analysis of the 2 studies. In both studies, doctors had visits by >1 parent; in study 2, there could be >1 visit between unique doctor-parent pairs and >1 psychosocial segment within each visit. We computed final results with generalized estimating equation-based procedures in the statistical software program Stata 6.0 (Stata Corp, College Station, TX). These procedures compute population-averaged statistics that take into account the nonindependence of observations and also require few assumptions about the distributions of the variables explored. 30
RESULTS
Topics Eliciting Discouraging Responses
Among the 208 psychosocial discussions from the 2 studies, 89 (43%) contained discouraging responses to parent psychosocial topics. These responses were more common among visits discussing corporal punishment (34 of 44 visits [77%]) than they were among the sample of segments containing a range of topics (55 of 164 segments [34%]). Among these more varied visits, the odds of a discouraging doctor response was greater (odds ratio [OR]: 3.07; 95% confidence interval [CI]: 1.56–6.05) if the topic discussed deviated from routine parenting (Table 3). The odds of a discouraging response were also increased if the topic was presented by the parent as a problem, rather than in a neutral manner (OR: 7.57; 95% CI: 3.50–16.44). The odds of a discouraging response did not vary according to who initiated the discussion of the topic (parent versus doctor; OR: 1.41; 95% CI: 0.56–3.53).
Parents were much more likely to present nonroutine topics as problems (57 of 73 topics [78%]), compared with routine topics (23 of 91 topics [25%]) (P < .001). Because of this relationship, we generated 4 classes of psychosocial issues, representing the 4 possible combinations of routine versus nonroutine topic and problem versus neutral presentation by the parent (Table 3). Presentation of a topic as a problem had a stronger association with discouraging responses than did the type of topic (nonroutine versus routine). The greatest odds of a discouraging response were for nonroutine topics presented as problems (OR: 16.6), followed by routine topics presented as problems (OR: 10.8) and nonroutine topics presented in a neutral way (OR: 5.5). Stated another way, doctors were relatively unlikely to give discouraging responses as long as the parent did not present a topic as being a problem. Presentation of even routine topics as problems, however, resulted in discouraging responses more than one half of the time.
Types of Discouraging Responses
Among the 89 visits or segments with discouraging responses, ignoring or dropping the subject was the most common (32 segments [36%]), followed by seeming powerless (25 segments [28%]), acting omnipotent (19 segments [21%]), and changing reality (13 segments [15%]). The type of discouraging response was associated with the type of problem mothers presented. "Routine" psychosocial issues, when they elicited discouraging responses, tended to be ignored, reframed, or dismissed with simple omnipotent formulations (Table 3). In contrast, when parents presented nonroutine topics as problems, 41% of doctors' discouraging responses were of the powerless type, suggesting that doctors had tried to give a substantive response but ultimately found themselves unable to do so successfully. For example,
Mother: He [her son] said, "Man, I hate you so much." My sister and I just laughed. I mean, we really didn't have any other choice. I could have whipped his [...]. If I put my hands on him, I am going to hurt him and they are going to put me in jail.... There've been times I have held him down, but I don't beat him.... I try to scare the [...] out of him.
Doctor: Sounds like you are struggling between wanting to hit him and wanting to be his parent. So we need to figure out how to make that better.
Mother: I made an appointment. He had an appointment for last week but I missed it.
Doctor: So his next appointment is next week And are you going to let me know what happens
Mother: I can get you copies if they get ...
Doctor: I would be glad to meet with [breaks off]. Do you have anyone to help you
Mother: He never listens to anyone. He is bad to everybody. Nobody can really handle him.
Doctor: Sounds like you have a lot to deal with. Make sure you let me know what happens.
Parent Factors Associated With Doctors' Discouraging Responses
Discouraging responses were not associated with the mother's age (mothers receiving a discouraging response were on average 0.6 years younger; 95% CI: –0.2.2 to 1.12), the mother's ethnicity (odds for response to be elicited by white mothers versus black mothers: 1.02; 95% CI: 0.36–2.9), the mother's distress as reflected on the General Health Questionnaire 31 (OR: 1.07; 95% CI: 0.58–1.96), or the length of the mother-doctor relationship (discouraging responses were associated with visit numbers that were on average 0.12 greater; 95% CI: –1.2 to 1.4). In visits with discouraging responses, mothers tended to give more medical information (average of 8.4 more information statements; 95% CI: 0.43–16.4) and more psychosocial information (average of 15.2 more information statements; 95% CI: –0.52 to 30.83), but they did not ask significantly more questions.
Doctor Factors Associated With Discouraging Responses
Female doctors were no more or less likely than male doctors to make discouraging responses (OR for discouraging response if doctor was female: 1.68; 95% CI: 0.58–4.8). Visits in which doctors made discouraging responses in psychosocial interactions were marked by doctors asking fewer psychosocial questions over the course of the entire visit. On average, in visits with discouraging responses, doctors asked 8 fewer psychosocial questions (95% CI: –13.1 to –2.8), a decrease of 24% in the number of psychosocial questions asked in an average visit. Discouraging responses were not related significantly to differences in doctors' patient-centeredness (mean difference in number of patient-centered utterances: –17; 95% CI: –53 to +19).
Of the 55 doctors included in the study, 10 had no psychosocial segments with discouraging responses and 22 had some segments with discouraging responses; for 23, all of the psychosocial segments had discouraging responses. There was no difference in gender among these 3 groups of doctors. The proportion of a doctor's visits containing discouraging responses was correlated with the proportion of visits in which parents raised nonroutine topics (r = 0.61; P < .001).
Responses That Encouraged Discussion of Psychosocial Topics
Among the 164 instances of psychosocial topics in the longitudinal study (encouraging responses were not coded in the first, cross-sectional study), doctors made responses that actively encouraged discussion 24 times (15%). The most common encouraging responses were building an alliance with empathetic statements (10 of 164 segments) and seeking additional information (13 of 164 segments). An example of an encouraging response, in which the doctor actively sought more information, is paraphrased here, because it included 2 minutes of interaction.
During the visit, a young child becomes insistent that he has to go to the bathroom. The parent remarks to the doctor that the child is "hard headed" as she tells the child to wait. The doctor asks the mother how she disciplines; she remarks that she "just makes him do things." The doctor asks specifically about spanking and the mother says that she does spank sometimes. Rather than going into a lecture about spanking, the doctor asks for more detail about how she spanks and how often. She replies but also asks several questions about specific behavior problems, which the doctor is able then to address.
Encouraging responses were no more common in visits with white versus black parents (OR: 1.2; 95% CI: 0.29–5.0) and were not related to the parent's age, the length of the parent-doctor relationship, or the doctor's gender. As with discouraging responses, encouraging responses were associated with the nature of the psychosocial topic and whether parents presented the topic as a problem (Table 3). In contrast to discouraging responses, however, encouraging responses were driven more by the nature of the topic (routine versus not) than by its being posed as a problem. Routine topics rarely received encouraging responses, whether or not they were presented as problems, whereas nonroutine topics, even if not presented as problems, had a 10-fold increased odds of receiving an encouraging response (OR: 10.9; 95% CI: 1.9–63.4). In contrast to discouraging responses, encouraging responses were related to the doctor's degree of patient-centered interview style. In visits with encouraging responses, doctors made on average 47 more patient-centered utterances (95% CI: 0–93), compared with visits with discouraging responses.
DISCUSSION
New Insights
Our findings are consistent with previous reports of doctors' responses to patients' disclosures of psychosocial concerns 2–9 and help us develop new hypotheses about why they occur. Discouraging responses did not seem to be related to doctors' characteristics, such as gender or patient-centered interview style, or to whether parents were black or white. Instead, what seemed to drive discouraging responses most powerfully, as a group, was a mother saying that she had a problem. This is consistent with findings from adult primary care visits, where doctors have been noted to be more negative, to disagree more, and to engage in less social conversation with patients who rate their physical or mental health as poor. 32
In addition, the type of problem presented influenced the type of discouraging response. Discouraging responses to routine problems tended to be dismissive or avoidant. That is, they seemed intended to provide quick reassurance and easy solutions or to avoid exploration of the topic. In contrast, when mothers posed nonroutine issues as problems, 41% of the discouraging responses took the form of failed attempts to offer more extensive advice. This pattern is consistent with findings observed in the treatment of patients with unexplained physical symptoms. In that setting, clinicians often manage patients' initial problem disclosures with quick simple advice, followed by a change of subject. If patients persist, giving additional perplexing details and revealing more distress, then doctors seem to feel compelled to respond (in the case of unexplained symptoms, by ordering tests they think the patient does not really need). 33 In the case of the residents in our study, failed attempts seemed to indicate an acknowledgment that the problem being presented by the parent was legitimate and should not be ignored but the resident lacked the skills or knowledge to be of help.
Another new insight comes from the contrast between predictors of discouraging and encouraging responses. Encouraging responses seem to occur when unusual problems are presented in the context of visits in which residents use a patient-centered interview style. This makes sense, given how patient-centeredness is defined, as a set of clinician behaviors that includes empathetically eliciting and understanding the patient's concerns. 34 Discouraging responses tend to occur when parents pose problems in the context of visits in which residents have been asking a smaller than average number of psychosocial questions. This also seems to make sense; in an atmosphere in which the conversational style seems aimed at limiting the disclosure of psychosocial information, statements about distress are likely to elicit additional avoidance.
We do not know why these different interactional contexts occur. There is some evidence that both parents and residents play a role. Of the 18 residents in the longitudinal study who had 1 visit with a discouraging response and for whom we had >1 visit in the data set, 8 had no visits in which they gave a psychosocial topic an encouraging response. These 8 might be residents who, for reasons we do not know, have a particularly difficult time with psychosocial distress. In contrast, the other 10 residents also had 1 visit in which they made encouraging responses, which suggests that some other factor or factors played a role in determining what they would do in any given visit. Parents' communication style might be one driving force here. Visits with discouraging responses were associated with parents giving more medical and psychosocial information, possibly increasing the chance that residents would feel overwhelmed and react defensively. Other factors that we did not measure could also play a role. Studies in a variety of clinical settings have found relationships between aspects of organizational climate and culture and patient care. Doctors who reported a greater sense of control over their work environment were seen as more participatory by their patients. 35 In community-based child and youth mental health services, a supportive and motivating work climate was associated with better clinician adherence to treatment guidelines 36 and with improved functioning among treated children. 37
These findings have implications for resident training in managing mental health problems in primary care. Focusing on skills that promote a participatory or patient-centered interchange 38 may be important but not sufficient in situations that can evoke doctors' emotional discomfort Additional training may need to focus on the discomfort itself and its causes, whether they are related to the substance of a parent's concerns or the volume of questions and information that the parent presents. This additional training might include communication skills that address explicitly residents' anxiety related to time management (such as skills for the empathetic redirection of rambling and the setting of an agenda for the visit), plus guidance on initial responses to common mental health and psychosocial problems. Practicing pediatricians say that they would tackle both parent and child problems more readily if they had better skills. 16,39 Tool kits such as Bright Futures in Practice: Mental Health 40 and the International Classification of Diseases, 10th Revision, Primary Care (recently revised with an expanded child mental health section) 41 offer efficient ways to improve knowledge of mental health and behavioral problems. Training programs have also been developed to address difficult emotional issues, such as helping families cope with domestic violence, 42 and to increase clinicians' willingness to elicit concerns from patients during cancer treatment. 43 These programs incorporate problem-specific knowledge, problem-specific communication skills, and opportunities for trainees to share their own feelings and ways of managing them. Finally, inevitably all clinicians find themselves confronted with patient problems for which they lack knowledge of what to do. Training in how to deal with such situations could be incorporated into the "difficult patient" training programs offered at many medical schools.
Another Coding System for Doctors' Responses to Patients' Concerns
Ours is one of several coding systems used to examine doctors' responses to psychosocial questions (Table 4). 2,3,7,8,43,44 The previous systems were empirically based. Two, those reported by Sharp et al 7 and Lynch et al, 44 examined doctors' actions, ie, whether they respond at all to parents' concerns and, if so, what concrete actions they take. Maguire et al, 43 Suchman et al, 2 and Levinson et al 3 developed systems that focus more explicitly on the affective and communication value of what doctors do, ie, whether they are responding in ways that facilitate or inhibit the discussion of concerns and emotions. Although our classification of responses contains categories that are similar to those developed by Levinson et al 3 and Maguire et al, 43 its foundation in both theory and clinicians' reflections allows us to generate hypotheses about the origins of pediatricians' behaviors. By differentiating among less-than-optimal responses, we arrive at a more optimistic conclusion than earlier studies; we see that an important proportion of inadequate responses represent not so much avoidance or dismissal of psychosocial issues but rather failed attempts to provide a meaningful response. In fact, the probability of both failed and potentially helpful responses increased as the problems mothers presented became more difficult.
We have used the term "discouraging responses" because, in the pioneering study of parent-pediatrician communication by Korsch et al, 6 parents often seemed to give up after doctors repeatedly seemed not to hear their statements of concern. In our own first examination of the data presented here, we also found that doctors' responses were associated with decreased introduction of psychosocial problems at subsequent visits. 11 Patients may not always give up voicing their concerns, however. In a study of pediatric visits to family medicine trainees, parents who ultimately elicited a doctor's response to psychosocial concerns had repeated the concern a minimum of 4 times. 44 Ring et al, 33 studying patients with difficult-to-explain medical symptoms, found that they, too, seemed to have developed conversational mechanisms to keep their doctors engaged with their problems. This represents an area for additional study. We need to understand more about how parents (and ultimately children and teens) choose to follow-up on doctors' lukewarm responses to concerns and how doctors can better gauge when their responses have been factually or emotionally inadequate. Not all routine concerns need elaborate replies, but others may represent bids for discussion about more troubling issues.
Limitations
Our study has several limitations. Perhaps most importantly, we do not know how results might have differed with more experienced doctors or in visits that included fathers or older children. Resident visits are very different from those of practicing community providers; they are longer, residents have less experience, and practices in the low-income urban communities where many residents train often involve patients with more medical and psychosocial problems. 45 Children's medical status was an unknown element in our analyses. All of the children enrolled in our studies came for health maintenance visits, but it is well recognized that medical and psychosocial issues compete for attention in primary care visits. 15,46 However, Levinson et al 3 found similarly high rates of inadequate responses to psychosocial issues in their study of community physicians seeing middle-class adult patients. We also would have liked to explore not only the parents' ethnicity but also the influence of ethnic concordance between mothers and residents. 24 Unfortunately, the data available to us included visits with only 1 black resident, and the large majority of parents in our study were black.
Another limitation is that our splitting of psychosocial problems into what we considered "routine" and what might be considered part of an expanded pediatric mission of responding to parent and family issues might have different meaning in other settings. 47 In our study, routine parenting issues fared relatively poorly; when presented as problems they tended to evoke dismissive responses, and when not presented as problems they received few encouraging ones. This might not be the case in training programs that place a greater emphasis on normal developmental and behavioral issues or in those that explicitly take a more family-centered approach to pediatric care.
CONCLUSIONS
Our findings are encouraging because they suggest that residents often do try to respond to the more difficult psychosocial issues that parents present. By differentiating between encouraging and discouraging responses and by classifying parent issues according to both topic and problem status, we may be able to develop new understandings of what drives doctors' responses to psychosocial issues and to develop better-focused and more-effective training for management of mental health problems in pediatric primary care.
ACKNOWLEDGMENTS
This work was supported by National Institute of Mental Health grants RO1MH 57782, RO1MH46134, and K24 MH01790
We are grateful to Dr Peter Salmon and 2 anonymous reviewers for thoughtful comments on earlier drafts of this manuscript.
FOOTNOTES
Accepted Sep 23, 2004.
Deceased.
This work was presented in part at the annual meeting of the Pediatric Academic Societies; May 1–4, 2004; San Francisco, CA.
No conflict of interest declared.
REFERENCES
Horwitz SM, Leaf PJ, Leventhal JM. Identification of psychosocial problems in pediatric primary care: do family attitudes make a difference Arch Pediatr Adolesc Med. 1998;152 :367 –371
Suchman AL, Markakis K, Beckman HB, Frankel R. A model of empathic communication. JAMA. 1997;277 :678 –682
Levinson W, Gorawara-Bhat R, Lamb J. A study of patient clues and physician responses in primary care and surgical settings. JAMA. 2000;284 :1021 –1027
Goore Z, Mangione R, Elliot MN, McDonald L, Kravitz RL. How much explanation is enough A study of parent requests for information and physician responses. Ambul Pediatr. 2001;1 :326 –332
Baker PS, Yoels WC, Clair JM. Emotional expression during medical encounters: social disease and the medical gaze. In: James V, Gabe J, eds. Health and the Sociology of Emotions. Oxford, United Kingdom: Blackwell Publishers; 1996:173 –199
Korsch BM, Gozzi EK, Francis V. Gaps in doctor-patient communication, I: doctor-patient interaction and patient satisfaction. Pediatrics. 1968;42 :855 –871
Sharp L, Pantell RH, Murphy LO, Lewis CC. Psychosocial problems during child health supervision visits: eliciting, then what Pediatrics. 1992;89 :619 –623
Wasserman RC, Inui TS, Barriatua RD, Carter WB, Lippincott P. Responsiveness to maternal concerns in preventive child health visits: an analysis of clinician-parent interactions. Dev Behav Pediatr. 1983;4 :171 –176
Wasserman RC, Inui TS, Barriatua RD, Carter WB, Lippincott P. Pediatric clinicians' support for parents makes a difference: an outcome-based analysis of clinician-parent interaction. Pediatrics. 1984;74 :1047 –1053
Hickson GB, Altemeier WA, O'Conner S. Concerns of mothers seeking care in private pediatric offices: opportunities for expanding services. Pediatrics. 1983;72 :619 –624
Wissow LS, Roter D, Larson S, Wang M-C, Hwang W-T, Johnson R. Mechanisms behind the failure of longitudinal primary care to promote the disclosure and discussion of psychosocial issues. Arch Pediatr Adolesc Med. 2002;156 :685 –692
Groves JE. Taking care of the hateful patient. N Engl J Med. 1978;298 :883 –887
Stern TA, Prager LM, Cremens MC. Autognosis rounds for medical house staff. Psychosomatics. 1993;34 :1 –7
Pollak J, Levy S. Countertransference and failure to report child abuse and neglect. Child Abuse Negl. 1989;13 :515 –522
Jaén CR, Stange KC, Nutting PA. Competing demands of primary care: a model for the delivery of clinical preventive services. J Fam Pract. 1994;38 :166 –171
Olson AL, Kelleher KJ, Kemper KJ, Zuckerman BS, Hammond CS, Dietrich AJ. Primary care pediatricians' roles and perceived responsibilities in the identification and management of depression in children and adolescents. Ambul Pediatr. 2001;1 :91 –98
Byrne A, Morton J, Salmon P. Defending against patients' pain: a qualitative analysis of nurses' responses to children's postoperative pain. J Psychosom Res. 2001;50 :69 –76
Hadjiisky E, Agostini D, Dardel F, Thouvenin C. Du Cri au Silence. 2nd ed. Vanves, France: Centre Technique National d'études et de Recherches sur les Handicaps et les Inadaptations; 1993
Gabel S, Bemporad J. An expanded concept of countertransference. J Am Acad Child Adolesc Psychiatry. 1994;33 :140 –141
Roter DL, Hall JA. Gender differences in patient-physician communication. In: Gallant SJ, Keita GP, Royak-Schaler R, eds. Health Care for Women: Psychological, Social, and Behavioral Influences. Washington, DC: American Psychological Association; 1997:57 –71
Thom DH, Stanford Trust Study Physicians. Physician behaviors that predict trust. J Fam Pract. 2001;50 :323 –328
Hall J, Roter D, Katz N. Correlates of provider behavior: a meta-analysis. Med Care. 1988;26 :657 –675
Doescher MP, Saver BG, Franks P, Fiscella K. Racial and ethnic disparities in perceptions of physician style and trust. Arch Fam Med. 2000;9 :1156 –1163
Cooper LA, Roter DL, Johnson RL, Ford DE, Steinwachs DM, Powe NR. Patient-centered communication, ratings of care, and concordance of patient and physician race. Ann Intern Med. 2003;139 :907 –915
Wissow LS, Roter D, Wilson MEH. Pediatrician interview style and mothers' disclosure of psychosocial issues. Pediatrics. 1994;93 :289 –295
Roter DL, Stewart M, Putnam SM, Lipkin M, Stiles W, Inui TS. Communication patterns of primary care physicians. JAMA. 1997;277 :350 –356
Wissow LS, Roter D, Larson S. Longitudinal care improves disclosure of psychosocial information. Arch Pediatr Adolesc Med. 2003;157 :419 –425
McCormick KF. Attitudes of primary care physicians toward corporal punishment. JAMA. 1992;267 :3161 –3165
Roter DL, Hall JA, Kern DE, Barker LR, Cole KA, Roca RP. Improving physicians' interviewing skills and reducing patients' emotional distress: a randomized clinical trial. Arch Intern Med. 1995;155 :1877 –1884
Diggle PJ, Liang KY, Zeger SL. Analysis of Longitudinal Data. Oxford, United Kingdom: Clarendon Press; 1994
Goldberg D, Hillier V. A scaled version of the General Health Questionnaire. Psychol Med. 1979;9 :139 –145
Hall JA, Roter DL, Milburn MA, Daltroy LH. Patients' health as a predictor of physician and patient behavior in medical visits: a synthesis of four studies. Med Care. 1996;34 :1205 –1218
Ring A, Dowrick C, Humphris G, Salmon P. Do patients with unexplained physical symptoms pressurize general practitioners for somatic treatment A qualitative study. BMJ. 2004;328 :1057
Mead N, Bower P. Measuring patient-centeredness: a comparison of three observation-based instruments. Patient Educ Couns. 2000;39 :71 –80
Kaplan SH, Greenfield S, Gandek B, Rogers WH, Ware JE Jr. Characteristics of physicians with participatory decision-making styles. Ann Intern Med. 1996;124 :497 –504
Henggeler SW, Schoenwald SK. The role of quality assurance in achieving outcomes in MST programs. J Juv Justice Detent Serv. 1999;14 :1 –17
Glisson C, Hemmelgarn A. The effects of organizational climate and interorganizational coordination on the quality and outcomes of children's service systems. Child Abuse Negl. 1998;22 :401 –421
Smith RC, Marshall-Dorsey AA, Osborn GG, et al. Evidence-based guidelines for teaching patient-centered interviewing. Patient Educ Couns. 2000;39 :27 –36
Olson AL, Kemper KJ, Kelleher KJ, Hammond CS, Zuckerman BS, Dietrich AJ. Primary care pediatricians' roles and perceived responsibilities in the identification and management of maternal depression. Pediatrics. 2002;110 :1169 –1176
National Center for Education in Maternal and Child Health. Bright Futures in Practice: Mental Health. Washington, DC: National Center for Education in Maternal and Child Health; 2002
World Health Organization. WHO Guide to Mental and Neurological Health in Primary Care: A Guide to Mental and Neurological Ill Health in Adults, Adolescents and Children. 2nd ed. London, United Kingdom: Royal Society of Medicine Press; 2004
Berger RP, Bogen D, Dulani T, Broussard E. Implementation of a program to teach pediatric residents and faculty about domestic violence. Arch Pediatr Adolesc Med. 2002;156 :804 –810
Maguire P, Faulkner A, Booth K, Elliott C, Hillier V. Helping cancer patients disclose their concerns. Eur J Cancer. 1996;32A :78 –81
Lynch TR, Wildman BG, Smucker WD. Parental disclosure of child psychosocial concerns: relationship to physician identification and management. J Fam Pract. 1997;44 :273 –280
Ambulatory Pediatric Association. Educating Pediatric Residents to Provide Health Care to Underserved Children. Alexandria, VA: Ambulatory Pediatric Association; 1990
Epner JEG, Levenberg PB, Schoeny ME. Primary care providers' responsiveness to health-risk behaviors reported by adolescent patients. Arch Pediatr Adolesc Med. 1998;152 :774 –780
Heneghan AM, Mercer MB, DeLeone NL. Will mothers discuss parenting stress and depressive symptoms with their child's pediatrician Pediatrics. 2004;113 :460 –467(Lawrence S. Wissow, MD, M)
Department of Medicine, Sinai Hospital, Baltimore, Maryland
Société Psychanalytique de Paris, Paris, France
ABSTRACT
Objective. Studies spanning nearly 4 decades demonstrate that doctors ignore or dismiss many patient bids for discussion of psychosocial topics. We sought to understand characteristics of doctors, patients, and visits in which this occurs.
Methods. Reanalysis of 167 audiotapes from 2 studies of parent-doctor communication in a pediatric residents' continuity clinic was performed. Tapes included visits by 100 mothers or female guardians to 55 residents who were the children's primary care providers. Coders identified mentions of psychosocial topics and noted the content and the doctor's response. Responses were classified with an adaptation of a previously described, psychoanalytically derived typology of avoidant or discouraging responses.
Results. Discouraging responses occurred in 34 (77%) of 44 discussions that involved corporal punishment and 51 (34%) of 64 discussions that involved other psychosocial topics. The particular topic (parent/family versus routine parenting issue) and how the topic was framed (as a problem versus simply mentioned) were associated with doctors' discouraging responses (OR: 3.07; 95% confidence interval: 1.56–6.05; and OR: 7.57; 95% confidence interval: 3.50–16.44; respectively). Discouraging responses were not related to the doctor's gender, parent's ethnicity, length of the parent-doctor relationship, or doctor's overall interview style (patient-centeredness). Discouraging responses to routine problems tended to be dismissive, but 41% of discouraging responses to parent/family problems were failed attempts to provide advice.
Conclusions. Discouraging responses seem to be related less to doctor or patient characteristics than to the type and acuity of the psychosocial topic. These responses may originate with doctors' discomfort with particular subject areas and thus might be approached with training that combines communication and emotion-handling skills with clinical tools such as Bright Futures in Practice: Mental Health or the International Classification of Diseases, 10th Revision, Primary Care.
Key Words: mental health primary care resident training
One persistent difficulty in pediatric primary care is the contrast between the high prevalence of parents' concerns about child behavior or development and the limited extent to which such concerns are discussed with doctors. The analysis by Horwitz et al 1 of pediatric visits in the New Haven, Connecticut, area found that 80% of parents of 4- to 8-year-old children thought that they had a psychosocial concern appropriate to discuss with their pediatricians but only approximately one half actually did so. In adult primary care visits, patients often approach psychosocial issues tentatively, starting with hints or feelers that they hope doctors will recognize. 2
Doctors may contribute to these hesitations consciously or unconsciously. Studies in both adult and pediatric primary care, spanning several decades, found that doctors ignore or dismiss the majority of patients' hints and disclosures of emotional distress. 2–9 Some patients may take these responses as a message that, in the future, the doctor would rather not hear about similar issues. 10,11
Many things may lead doctors to avoid discussion of patients' distress. Doctors may be troubled by the emotions involved and by feeling that they lack the skills, time, or referral resources required to manage emotional problems. 12–16 Clinicians also may chose to distance themselves from patients whose needs they see as going beyond what is expected or reasonable for their specialty or type of interaction. 17
Although previous studies documented the extent to which primary care providers avoid discussion of patient psychosocial concerns or engage in discussions that fail to provide advice, we lack an understanding of the kinds of concerns that evoke avoidance and of which doctors are most likely to be avoidant. Understanding when and why clinicians become avoidant could facilitate development of training, structural alterations to practice settings, and incentives that might promote better care of emotional issues. To that end, Hadjiisky et al 18 interviewed clinicians from a variety of backgrounds (social workers, day care center workers and directors, nurses, and physicians) who had encountered a situation widely acknowledged to promote discomfort among professionals, ie, interaction with a family for which there was suspicion of child abuse. Those investigators asked clinicians to describe their feelings and responses and then used psychoanalytic theory to classify and to interpret the reactions the clinicians reported. Psychoanalytic theory posits that clinicians develop emotional responses to patients and their life situations, just as patients develop emotional responses to their clinicians and to what is discussed during treatment. When these responses are uncomfortable, clinicians, like their patients, can manage them in ways that are more or less adaptive. 19 Hadjiisky et al 18 identified 3 main mechanisms clinicians used to distance themselves from the unpleasant emotions associated with cases of suspected abuse. First, they might reframe or reexplain what they had heard, to make it seem as if it should not be labeled as abuse or to make it seem less severe ("doubt" or denial of reality). Second, they might acknowledge the severity of the problem but propose an overly simplistic solution, as if their recommendation was sufficient to make everything better ("all-powerfulness"). Finally, they might rationalize inaction or silence by saying to themselves that the problem was insoluble ("helplessness").
We developed a system to identify apparent manifestations of the distress responses described by Hadjiisky et al 18 in audiotape recordings of primary care visits between pediatric residents and their patients' mothers. We hypothesized that these responses would be more common when mothers presented residents with psychosocial issues that were outside the expected content of a pediatric visit or that were presented as problems, that is, associated with potential or overt emotional distress. We also wanted to explore the influence of the length of the mother-doctor relationship, the doctor's gender and "patient-centered" style of communicating, and the mother's ethnicity (Fig 1). Female physicians are noted to be more concerned with psychosocial issues, 20 and some studies find that longer patient-doctor relationships increase trust and satisfaction, 21 as does doctors' use of a patient-centered style of communication. 22 In the United States, black patients report difficulty communicating with doctors of different ethnicities. 23,24 Finally, we hoped that the classification of responses described by Hadjiisky et al 18 would provide us with clues about the mechanisms behind residents' responses and thus lead us to hypotheses about possible interventions.
METHODS
Clinical Setting
Data came from 2 studies performed 8 years apart in the same pediatric residents' continuity clinic. 11,25 Both studies were approved by the Joint Committee on Clinical Investigation of the Johns Hopkins Medical Institutions, and the subsequent reanalysis was approved by the Committee of Human Research of the Johns Hopkins Bloomberg School of Public Health. Both studies involved asking parents and residents for permission to audiotape scheduled well-child and follow-up visits.
Initial Coding
Tapes were initially coded with the Roter Interaction Analysis System, 26 a scheme for classifying talk in clinical conversations that identifies the purpose of the talk (eg, question or information giving), the speaker, and the intended recipient. Doctors' patient-centeredness was calculated as the sum of all doctor talk classified as giving information (medical and psychosocial), asking psychosocial questions, showing empathy, giving reassurance, and building a partnership. In our previous work, this formulation of patient centeredness was associated with higher levels of parents' psychosocial talk. 27
Visits Included in the Analysis
In the first study, 25 coders identified 44 of 234 visits as including discussion of physical punishment. In 26 of the 44 visits, the parents stated that they used physical punishment; they mentioned its use as a possibility in 13 visits, and there was some other discussion of discipline that raised concerns about its use in the remaining 5 visits. Twenty-nine doctors provided care in these visits; 20 (69%) were female, 27 were white, 1 was black, and 1 was Asian American. Eleven doctors were in their third year of training, 10 were in their second year, and 8 were at the end of their first year. In these visits, the majority of children (33 of 44 children [75%]) were 3 years of age, and the oldest was 10 years of age. Twenty (45%) of the 44 mothers had a high school education or some college, and all except 4 were black. The mothers' average age was 26.6 years (range: 16–68 years). Mothers were asked to estimate how many times they had seen the same doctor previously. The average estimate was 11, with a range of 0 to 16.
In the second study, 11 coders identified 164 visit segments in 123 visits (of a total of 692 visits) containing discussion of a psychosocial topic (Table 1). Coders grouped the topics into categories (routine parenting, parents' relationships with partners or parents, stresses related to a child's medical condition, parents' concerns about their own schooling, job, or child care arrangements, and parents' substance abuse or legal problems) and noted whether the topic was simply mentioned or was framed as a problem for the parent. Simple mentions included parents' passive responses to doctors' questions and explicit statements that there was no problem. An example of an interchange with a passive parent answer is as follows.
Doctor: And things with you are going OK
Parent: Mm.
Doctor: Good, glad to hear it.
An explicit statement of "no problem" is as follows.
Doctor: And how are you guys getting along [now that the infant is here]
Mother: Good.
Doctor: You getting help from dad
Mother: [pause] Yeah.
Framing as a problem included hints that there might be a problem, statements of parents' increased vigilance, explicit disclosures of parents' distress or concern, and explicit requests for help by parents. A hint from a parent might be as follows.
Doctor: Are you having fun with him
Parent: He's a handful.
Doctor: What's that
Parent: He's a handful.
Doctor: Yeah, he is.
A parent's disclosure of distress or a problem would be as follows.
Parent: I'm just having a lot of problems 'cause I graduated from truck driver school and got a job offer.
Doctor: Oh, really
Parent: But I don't have anybody to leave the kids with.
An example of an explicit request for help is as follows.
Parent: Is there a family counselor here Because I have this problem. Right With her.
We approached reliability of coding for topic and problem type by having initial discussions about groupings and definitions, openly discussing several examples to develop a coding manual, and then having 2 investigators (S.L. and L.S.W.) independently code 25 parent-doctor psychosocial interchanges. Agreement between the 2 raters ranged from 88% to 100% for the 7 topic categories and from 84% to 100% for the 6 problem/distress categories.
All of the children in visits in the second study were infants (<1 year of age). Because this was a longitudinal study, the 123 visits involved only 56 mothers, of whom 50 (89%) were black and 6 were white. Mothers ranged in age from 17 to 64 years (mean: 26.1 years; range: 17–64 years). Most (39 of 56 [70%]) had finished high school or had taken some college-level courses. Twenty-six doctors were represented; 23 (88%) were white, 3 identified themselves as Asian, and 20 (76%) were female. Fifteen doctors were in their first year of training, and 11 were in their second year. In this study, the original goal was to recruit parent-doctor dyads as close to the first newborn visit as possible and then to record subsequent visits over the course of a year. Therefore, the visits in this sample ranged from the first to the eighth meeting between the doctor and patient (median: second visit).
Coding of Doctors' Discouraging Responses to Psychosocial Topics
We developed an initial set of codes and definitions by transcribing excerpts from visits with discussions of physical punishment. We chose these excerpts because of their similarity to the situations in the original work by Hadjiisky et al 18 and because physical punishment is a subject toward which attitudes are known to vary among doctors. 28 Two of the authors (E.H. and L.S.W.) reviewed the transcripts independently and then together, matching clinicians' responses to the broad categories of reactions noted in the interviews with clinicians by Hadjiisky et al. 18 On the basis of these discussions, we developed 9 categories of responses, 4 discouraging psychosocial discussion and 5 encouraging discussion (Table 2). The 4 discouraging responses corresponded to the 3 broad categories described by Hadjiisky et al 18 plus a category of seemingly ignored parent remarks. The 5 categories of responses that seemed to encourage disclosure particularly well (openly seeking help from other sources, building alliance with empathy, reducing the crisis without minimizing, bolstering credibility without omnipotence, and seeking additional information from the parent before giving advice) corresponded to elements of doctors' problem-identification and emotion-handling skills that were shown to help reduce psychosocial distress among adult primary care patients. 29
Another pair of authors (S.L. and L.S.W.) then applied these codes to excerpts from audiotapes. They assigned codes independently and then met to discuss disagreements and to modify category definitions. When definitions were finalized, one of the authors (S.L.) coded psychosocial discussion segments from study 2, with the other author duplicate-coding a 15% sample; agreement was >90% for each of the doctor response categories. A third author (J.A.) learned the system later. When she and another author reached 100% agreement coding independently, the third author coded visits from study 1.
The coding process involved going to the place in a recording where the target discussion took place and listening to it 1 times, paying attention to tone of voice as well as the words spoken. Tone and context were particularly useful in labeling doctor responses that in isolation might have seemed ambiguous. For example, "I wouldn't worry about it," might be labeled as dismissive and discouraging of discussion if it was said curtly and without any accompanying explanation. In contrast, it might be labeled as empathetic and positive if it was said warmly and after an explanation and an opportunity for the parent to make clear her concerns.
Not all responses were classified as discouraging or particularly encouraging. Responses classified as neutral were those that gave appropriate information but did not make additional attempts to explore the parent's feelings or to provide empathy. A neutral response to a disclosure of fatigue related to child sleep problems might be as follows.
Parent: I'm feeling really wiped out because I'm going in to check every time I hear him cry at night.
Doctor: At this age it's OK to let them cry for a few minutes before going in; they need to learn how to wake up and settle themselves down. How's he feeding
Coders were allowed to make their own determination of when a discussion ended and were asked to assign the single response code that seemed to fit best. The final coding process thus involved (1) initial coding of tapes to mark locations of psychosocial discussion, (2) study of the coding manual (descriptions of the codes, with examples), (3) training to agreement with someone already familiar with the system, and (4) independent coding, with quality checks.
Statistical Methods
Issues of statistical independence complicated analysis of the 2 studies. In both studies, doctors had visits by >1 parent; in study 2, there could be >1 visit between unique doctor-parent pairs and >1 psychosocial segment within each visit. We computed final results with generalized estimating equation-based procedures in the statistical software program Stata 6.0 (Stata Corp, College Station, TX). These procedures compute population-averaged statistics that take into account the nonindependence of observations and also require few assumptions about the distributions of the variables explored. 30
RESULTS
Topics Eliciting Discouraging Responses
Among the 208 psychosocial discussions from the 2 studies, 89 (43%) contained discouraging responses to parent psychosocial topics. These responses were more common among visits discussing corporal punishment (34 of 44 visits [77%]) than they were among the sample of segments containing a range of topics (55 of 164 segments [34%]). Among these more varied visits, the odds of a discouraging doctor response was greater (odds ratio [OR]: 3.07; 95% confidence interval [CI]: 1.56–6.05) if the topic discussed deviated from routine parenting (Table 3). The odds of a discouraging response were also increased if the topic was presented by the parent as a problem, rather than in a neutral manner (OR: 7.57; 95% CI: 3.50–16.44). The odds of a discouraging response did not vary according to who initiated the discussion of the topic (parent versus doctor; OR: 1.41; 95% CI: 0.56–3.53).
Parents were much more likely to present nonroutine topics as problems (57 of 73 topics [78%]), compared with routine topics (23 of 91 topics [25%]) (P < .001). Because of this relationship, we generated 4 classes of psychosocial issues, representing the 4 possible combinations of routine versus nonroutine topic and problem versus neutral presentation by the parent (Table 3). Presentation of a topic as a problem had a stronger association with discouraging responses than did the type of topic (nonroutine versus routine). The greatest odds of a discouraging response were for nonroutine topics presented as problems (OR: 16.6), followed by routine topics presented as problems (OR: 10.8) and nonroutine topics presented in a neutral way (OR: 5.5). Stated another way, doctors were relatively unlikely to give discouraging responses as long as the parent did not present a topic as being a problem. Presentation of even routine topics as problems, however, resulted in discouraging responses more than one half of the time.
Types of Discouraging Responses
Among the 89 visits or segments with discouraging responses, ignoring or dropping the subject was the most common (32 segments [36%]), followed by seeming powerless (25 segments [28%]), acting omnipotent (19 segments [21%]), and changing reality (13 segments [15%]). The type of discouraging response was associated with the type of problem mothers presented. "Routine" psychosocial issues, when they elicited discouraging responses, tended to be ignored, reframed, or dismissed with simple omnipotent formulations (Table 3). In contrast, when parents presented nonroutine topics as problems, 41% of doctors' discouraging responses were of the powerless type, suggesting that doctors had tried to give a substantive response but ultimately found themselves unable to do so successfully. For example,
Mother: He [her son] said, "Man, I hate you so much." My sister and I just laughed. I mean, we really didn't have any other choice. I could have whipped his [...]. If I put my hands on him, I am going to hurt him and they are going to put me in jail.... There've been times I have held him down, but I don't beat him.... I try to scare the [...] out of him.
Doctor: Sounds like you are struggling between wanting to hit him and wanting to be his parent. So we need to figure out how to make that better.
Mother: I made an appointment. He had an appointment for last week but I missed it.
Doctor: So his next appointment is next week And are you going to let me know what happens
Mother: I can get you copies if they get ...
Doctor: I would be glad to meet with [breaks off]. Do you have anyone to help you
Mother: He never listens to anyone. He is bad to everybody. Nobody can really handle him.
Doctor: Sounds like you have a lot to deal with. Make sure you let me know what happens.
Parent Factors Associated With Doctors' Discouraging Responses
Discouraging responses were not associated with the mother's age (mothers receiving a discouraging response were on average 0.6 years younger; 95% CI: –0.2.2 to 1.12), the mother's ethnicity (odds for response to be elicited by white mothers versus black mothers: 1.02; 95% CI: 0.36–2.9), the mother's distress as reflected on the General Health Questionnaire 31 (OR: 1.07; 95% CI: 0.58–1.96), or the length of the mother-doctor relationship (discouraging responses were associated with visit numbers that were on average 0.12 greater; 95% CI: –1.2 to 1.4). In visits with discouraging responses, mothers tended to give more medical information (average of 8.4 more information statements; 95% CI: 0.43–16.4) and more psychosocial information (average of 15.2 more information statements; 95% CI: –0.52 to 30.83), but they did not ask significantly more questions.
Doctor Factors Associated With Discouraging Responses
Female doctors were no more or less likely than male doctors to make discouraging responses (OR for discouraging response if doctor was female: 1.68; 95% CI: 0.58–4.8). Visits in which doctors made discouraging responses in psychosocial interactions were marked by doctors asking fewer psychosocial questions over the course of the entire visit. On average, in visits with discouraging responses, doctors asked 8 fewer psychosocial questions (95% CI: –13.1 to –2.8), a decrease of 24% in the number of psychosocial questions asked in an average visit. Discouraging responses were not related significantly to differences in doctors' patient-centeredness (mean difference in number of patient-centered utterances: –17; 95% CI: –53 to +19).
Of the 55 doctors included in the study, 10 had no psychosocial segments with discouraging responses and 22 had some segments with discouraging responses; for 23, all of the psychosocial segments had discouraging responses. There was no difference in gender among these 3 groups of doctors. The proportion of a doctor's visits containing discouraging responses was correlated with the proportion of visits in which parents raised nonroutine topics (r = 0.61; P < .001).
Responses That Encouraged Discussion of Psychosocial Topics
Among the 164 instances of psychosocial topics in the longitudinal study (encouraging responses were not coded in the first, cross-sectional study), doctors made responses that actively encouraged discussion 24 times (15%). The most common encouraging responses were building an alliance with empathetic statements (10 of 164 segments) and seeking additional information (13 of 164 segments). An example of an encouraging response, in which the doctor actively sought more information, is paraphrased here, because it included 2 minutes of interaction.
During the visit, a young child becomes insistent that he has to go to the bathroom. The parent remarks to the doctor that the child is "hard headed" as she tells the child to wait. The doctor asks the mother how she disciplines; she remarks that she "just makes him do things." The doctor asks specifically about spanking and the mother says that she does spank sometimes. Rather than going into a lecture about spanking, the doctor asks for more detail about how she spanks and how often. She replies but also asks several questions about specific behavior problems, which the doctor is able then to address.
Encouraging responses were no more common in visits with white versus black parents (OR: 1.2; 95% CI: 0.29–5.0) and were not related to the parent's age, the length of the parent-doctor relationship, or the doctor's gender. As with discouraging responses, encouraging responses were associated with the nature of the psychosocial topic and whether parents presented the topic as a problem (Table 3). In contrast to discouraging responses, however, encouraging responses were driven more by the nature of the topic (routine versus not) than by its being posed as a problem. Routine topics rarely received encouraging responses, whether or not they were presented as problems, whereas nonroutine topics, even if not presented as problems, had a 10-fold increased odds of receiving an encouraging response (OR: 10.9; 95% CI: 1.9–63.4). In contrast to discouraging responses, encouraging responses were related to the doctor's degree of patient-centered interview style. In visits with encouraging responses, doctors made on average 47 more patient-centered utterances (95% CI: 0–93), compared with visits with discouraging responses.
DISCUSSION
New Insights
Our findings are consistent with previous reports of doctors' responses to patients' disclosures of psychosocial concerns 2–9 and help us develop new hypotheses about why they occur. Discouraging responses did not seem to be related to doctors' characteristics, such as gender or patient-centered interview style, or to whether parents were black or white. Instead, what seemed to drive discouraging responses most powerfully, as a group, was a mother saying that she had a problem. This is consistent with findings from adult primary care visits, where doctors have been noted to be more negative, to disagree more, and to engage in less social conversation with patients who rate their physical or mental health as poor. 32
In addition, the type of problem presented influenced the type of discouraging response. Discouraging responses to routine problems tended to be dismissive or avoidant. That is, they seemed intended to provide quick reassurance and easy solutions or to avoid exploration of the topic. In contrast, when mothers posed nonroutine issues as problems, 41% of the discouraging responses took the form of failed attempts to offer more extensive advice. This pattern is consistent with findings observed in the treatment of patients with unexplained physical symptoms. In that setting, clinicians often manage patients' initial problem disclosures with quick simple advice, followed by a change of subject. If patients persist, giving additional perplexing details and revealing more distress, then doctors seem to feel compelled to respond (in the case of unexplained symptoms, by ordering tests they think the patient does not really need). 33 In the case of the residents in our study, failed attempts seemed to indicate an acknowledgment that the problem being presented by the parent was legitimate and should not be ignored but the resident lacked the skills or knowledge to be of help.
Another new insight comes from the contrast between predictors of discouraging and encouraging responses. Encouraging responses seem to occur when unusual problems are presented in the context of visits in which residents use a patient-centered interview style. This makes sense, given how patient-centeredness is defined, as a set of clinician behaviors that includes empathetically eliciting and understanding the patient's concerns. 34 Discouraging responses tend to occur when parents pose problems in the context of visits in which residents have been asking a smaller than average number of psychosocial questions. This also seems to make sense; in an atmosphere in which the conversational style seems aimed at limiting the disclosure of psychosocial information, statements about distress are likely to elicit additional avoidance.
We do not know why these different interactional contexts occur. There is some evidence that both parents and residents play a role. Of the 18 residents in the longitudinal study who had 1 visit with a discouraging response and for whom we had >1 visit in the data set, 8 had no visits in which they gave a psychosocial topic an encouraging response. These 8 might be residents who, for reasons we do not know, have a particularly difficult time with psychosocial distress. In contrast, the other 10 residents also had 1 visit in which they made encouraging responses, which suggests that some other factor or factors played a role in determining what they would do in any given visit. Parents' communication style might be one driving force here. Visits with discouraging responses were associated with parents giving more medical and psychosocial information, possibly increasing the chance that residents would feel overwhelmed and react defensively. Other factors that we did not measure could also play a role. Studies in a variety of clinical settings have found relationships between aspects of organizational climate and culture and patient care. Doctors who reported a greater sense of control over their work environment were seen as more participatory by their patients. 35 In community-based child and youth mental health services, a supportive and motivating work climate was associated with better clinician adherence to treatment guidelines 36 and with improved functioning among treated children. 37
These findings have implications for resident training in managing mental health problems in primary care. Focusing on skills that promote a participatory or patient-centered interchange 38 may be important but not sufficient in situations that can evoke doctors' emotional discomfort Additional training may need to focus on the discomfort itself and its causes, whether they are related to the substance of a parent's concerns or the volume of questions and information that the parent presents. This additional training might include communication skills that address explicitly residents' anxiety related to time management (such as skills for the empathetic redirection of rambling and the setting of an agenda for the visit), plus guidance on initial responses to common mental health and psychosocial problems. Practicing pediatricians say that they would tackle both parent and child problems more readily if they had better skills. 16,39 Tool kits such as Bright Futures in Practice: Mental Health 40 and the International Classification of Diseases, 10th Revision, Primary Care (recently revised with an expanded child mental health section) 41 offer efficient ways to improve knowledge of mental health and behavioral problems. Training programs have also been developed to address difficult emotional issues, such as helping families cope with domestic violence, 42 and to increase clinicians' willingness to elicit concerns from patients during cancer treatment. 43 These programs incorporate problem-specific knowledge, problem-specific communication skills, and opportunities for trainees to share their own feelings and ways of managing them. Finally, inevitably all clinicians find themselves confronted with patient problems for which they lack knowledge of what to do. Training in how to deal with such situations could be incorporated into the "difficult patient" training programs offered at many medical schools.
Another Coding System for Doctors' Responses to Patients' Concerns
Ours is one of several coding systems used to examine doctors' responses to psychosocial questions (Table 4). 2,3,7,8,43,44 The previous systems were empirically based. Two, those reported by Sharp et al 7 and Lynch et al, 44 examined doctors' actions, ie, whether they respond at all to parents' concerns and, if so, what concrete actions they take. Maguire et al, 43 Suchman et al, 2 and Levinson et al 3 developed systems that focus more explicitly on the affective and communication value of what doctors do, ie, whether they are responding in ways that facilitate or inhibit the discussion of concerns and emotions. Although our classification of responses contains categories that are similar to those developed by Levinson et al 3 and Maguire et al, 43 its foundation in both theory and clinicians' reflections allows us to generate hypotheses about the origins of pediatricians' behaviors. By differentiating among less-than-optimal responses, we arrive at a more optimistic conclusion than earlier studies; we see that an important proportion of inadequate responses represent not so much avoidance or dismissal of psychosocial issues but rather failed attempts to provide a meaningful response. In fact, the probability of both failed and potentially helpful responses increased as the problems mothers presented became more difficult.
We have used the term "discouraging responses" because, in the pioneering study of parent-pediatrician communication by Korsch et al, 6 parents often seemed to give up after doctors repeatedly seemed not to hear their statements of concern. In our own first examination of the data presented here, we also found that doctors' responses were associated with decreased introduction of psychosocial problems at subsequent visits. 11 Patients may not always give up voicing their concerns, however. In a study of pediatric visits to family medicine trainees, parents who ultimately elicited a doctor's response to psychosocial concerns had repeated the concern a minimum of 4 times. 44 Ring et al, 33 studying patients with difficult-to-explain medical symptoms, found that they, too, seemed to have developed conversational mechanisms to keep their doctors engaged with their problems. This represents an area for additional study. We need to understand more about how parents (and ultimately children and teens) choose to follow-up on doctors' lukewarm responses to concerns and how doctors can better gauge when their responses have been factually or emotionally inadequate. Not all routine concerns need elaborate replies, but others may represent bids for discussion about more troubling issues.
Limitations
Our study has several limitations. Perhaps most importantly, we do not know how results might have differed with more experienced doctors or in visits that included fathers or older children. Resident visits are very different from those of practicing community providers; they are longer, residents have less experience, and practices in the low-income urban communities where many residents train often involve patients with more medical and psychosocial problems. 45 Children's medical status was an unknown element in our analyses. All of the children enrolled in our studies came for health maintenance visits, but it is well recognized that medical and psychosocial issues compete for attention in primary care visits. 15,46 However, Levinson et al 3 found similarly high rates of inadequate responses to psychosocial issues in their study of community physicians seeing middle-class adult patients. We also would have liked to explore not only the parents' ethnicity but also the influence of ethnic concordance between mothers and residents. 24 Unfortunately, the data available to us included visits with only 1 black resident, and the large majority of parents in our study were black.
Another limitation is that our splitting of psychosocial problems into what we considered "routine" and what might be considered part of an expanded pediatric mission of responding to parent and family issues might have different meaning in other settings. 47 In our study, routine parenting issues fared relatively poorly; when presented as problems they tended to evoke dismissive responses, and when not presented as problems they received few encouraging ones. This might not be the case in training programs that place a greater emphasis on normal developmental and behavioral issues or in those that explicitly take a more family-centered approach to pediatric care.
CONCLUSIONS
Our findings are encouraging because they suggest that residents often do try to respond to the more difficult psychosocial issues that parents present. By differentiating between encouraging and discouraging responses and by classifying parent issues according to both topic and problem status, we may be able to develop new understandings of what drives doctors' responses to psychosocial issues and to develop better-focused and more-effective training for management of mental health problems in pediatric primary care.
ACKNOWLEDGMENTS
This work was supported by National Institute of Mental Health grants RO1MH 57782, RO1MH46134, and K24 MH01790
We are grateful to Dr Peter Salmon and 2 anonymous reviewers for thoughtful comments on earlier drafts of this manuscript.
FOOTNOTES
Accepted Sep 23, 2004.
Deceased.
This work was presented in part at the annual meeting of the Pediatric Academic Societies; May 1–4, 2004; San Francisco, CA.
No conflict of interest declared.
REFERENCES
Horwitz SM, Leaf PJ, Leventhal JM. Identification of psychosocial problems in pediatric primary care: do family attitudes make a difference Arch Pediatr Adolesc Med. 1998;152 :367 –371
Suchman AL, Markakis K, Beckman HB, Frankel R. A model of empathic communication. JAMA. 1997;277 :678 –682
Levinson W, Gorawara-Bhat R, Lamb J. A study of patient clues and physician responses in primary care and surgical settings. JAMA. 2000;284 :1021 –1027
Goore Z, Mangione R, Elliot MN, McDonald L, Kravitz RL. How much explanation is enough A study of parent requests for information and physician responses. Ambul Pediatr. 2001;1 :326 –332
Baker PS, Yoels WC, Clair JM. Emotional expression during medical encounters: social disease and the medical gaze. In: James V, Gabe J, eds. Health and the Sociology of Emotions. Oxford, United Kingdom: Blackwell Publishers; 1996:173 –199
Korsch BM, Gozzi EK, Francis V. Gaps in doctor-patient communication, I: doctor-patient interaction and patient satisfaction. Pediatrics. 1968;42 :855 –871
Sharp L, Pantell RH, Murphy LO, Lewis CC. Psychosocial problems during child health supervision visits: eliciting, then what Pediatrics. 1992;89 :619 –623
Wasserman RC, Inui TS, Barriatua RD, Carter WB, Lippincott P. Responsiveness to maternal concerns in preventive child health visits: an analysis of clinician-parent interactions. Dev Behav Pediatr. 1983;4 :171 –176
Wasserman RC, Inui TS, Barriatua RD, Carter WB, Lippincott P. Pediatric clinicians' support for parents makes a difference: an outcome-based analysis of clinician-parent interaction. Pediatrics. 1984;74 :1047 –1053
Hickson GB, Altemeier WA, O'Conner S. Concerns of mothers seeking care in private pediatric offices: opportunities for expanding services. Pediatrics. 1983;72 :619 –624
Wissow LS, Roter D, Larson S, Wang M-C, Hwang W-T, Johnson R. Mechanisms behind the failure of longitudinal primary care to promote the disclosure and discussion of psychosocial issues. Arch Pediatr Adolesc Med. 2002;156 :685 –692
Groves JE. Taking care of the hateful patient. N Engl J Med. 1978;298 :883 –887
Stern TA, Prager LM, Cremens MC. Autognosis rounds for medical house staff. Psychosomatics. 1993;34 :1 –7
Pollak J, Levy S. Countertransference and failure to report child abuse and neglect. Child Abuse Negl. 1989;13 :515 –522
Jaén CR, Stange KC, Nutting PA. Competing demands of primary care: a model for the delivery of clinical preventive services. J Fam Pract. 1994;38 :166 –171
Olson AL, Kelleher KJ, Kemper KJ, Zuckerman BS, Hammond CS, Dietrich AJ. Primary care pediatricians' roles and perceived responsibilities in the identification and management of depression in children and adolescents. Ambul Pediatr. 2001;1 :91 –98
Byrne A, Morton J, Salmon P. Defending against patients' pain: a qualitative analysis of nurses' responses to children's postoperative pain. J Psychosom Res. 2001;50 :69 –76
Hadjiisky E, Agostini D, Dardel F, Thouvenin C. Du Cri au Silence. 2nd ed. Vanves, France: Centre Technique National d'études et de Recherches sur les Handicaps et les Inadaptations; 1993
Gabel S, Bemporad J. An expanded concept of countertransference. J Am Acad Child Adolesc Psychiatry. 1994;33 :140 –141
Roter DL, Hall JA. Gender differences in patient-physician communication. In: Gallant SJ, Keita GP, Royak-Schaler R, eds. Health Care for Women: Psychological, Social, and Behavioral Influences. Washington, DC: American Psychological Association; 1997:57 –71
Thom DH, Stanford Trust Study Physicians. Physician behaviors that predict trust. J Fam Pract. 2001;50 :323 –328
Hall J, Roter D, Katz N. Correlates of provider behavior: a meta-analysis. Med Care. 1988;26 :657 –675
Doescher MP, Saver BG, Franks P, Fiscella K. Racial and ethnic disparities in perceptions of physician style and trust. Arch Fam Med. 2000;9 :1156 –1163
Cooper LA, Roter DL, Johnson RL, Ford DE, Steinwachs DM, Powe NR. Patient-centered communication, ratings of care, and concordance of patient and physician race. Ann Intern Med. 2003;139 :907 –915
Wissow LS, Roter D, Wilson MEH. Pediatrician interview style and mothers' disclosure of psychosocial issues. Pediatrics. 1994;93 :289 –295
Roter DL, Stewart M, Putnam SM, Lipkin M, Stiles W, Inui TS. Communication patterns of primary care physicians. JAMA. 1997;277 :350 –356
Wissow LS, Roter D, Larson S. Longitudinal care improves disclosure of psychosocial information. Arch Pediatr Adolesc Med. 2003;157 :419 –425
McCormick KF. Attitudes of primary care physicians toward corporal punishment. JAMA. 1992;267 :3161 –3165
Roter DL, Hall JA, Kern DE, Barker LR, Cole KA, Roca RP. Improving physicians' interviewing skills and reducing patients' emotional distress: a randomized clinical trial. Arch Intern Med. 1995;155 :1877 –1884
Diggle PJ, Liang KY, Zeger SL. Analysis of Longitudinal Data. Oxford, United Kingdom: Clarendon Press; 1994
Goldberg D, Hillier V. A scaled version of the General Health Questionnaire. Psychol Med. 1979;9 :139 –145
Hall JA, Roter DL, Milburn MA, Daltroy LH. Patients' health as a predictor of physician and patient behavior in medical visits: a synthesis of four studies. Med Care. 1996;34 :1205 –1218
Ring A, Dowrick C, Humphris G, Salmon P. Do patients with unexplained physical symptoms pressurize general practitioners for somatic treatment A qualitative study. BMJ. 2004;328 :1057
Mead N, Bower P. Measuring patient-centeredness: a comparison of three observation-based instruments. Patient Educ Couns. 2000;39 :71 –80
Kaplan SH, Greenfield S, Gandek B, Rogers WH, Ware JE Jr. Characteristics of physicians with participatory decision-making styles. Ann Intern Med. 1996;124 :497 –504
Henggeler SW, Schoenwald SK. The role of quality assurance in achieving outcomes in MST programs. J Juv Justice Detent Serv. 1999;14 :1 –17
Glisson C, Hemmelgarn A. The effects of organizational climate and interorganizational coordination on the quality and outcomes of children's service systems. Child Abuse Negl. 1998;22 :401 –421
Smith RC, Marshall-Dorsey AA, Osborn GG, et al. Evidence-based guidelines for teaching patient-centered interviewing. Patient Educ Couns. 2000;39 :27 –36
Olson AL, Kemper KJ, Kelleher KJ, Hammond CS, Zuckerman BS, Dietrich AJ. Primary care pediatricians' roles and perceived responsibilities in the identification and management of maternal depression. Pediatrics. 2002;110 :1169 –1176
National Center for Education in Maternal and Child Health. Bright Futures in Practice: Mental Health. Washington, DC: National Center for Education in Maternal and Child Health; 2002
World Health Organization. WHO Guide to Mental and Neurological Health in Primary Care: A Guide to Mental and Neurological Ill Health in Adults, Adolescents and Children. 2nd ed. London, United Kingdom: Royal Society of Medicine Press; 2004
Berger RP, Bogen D, Dulani T, Broussard E. Implementation of a program to teach pediatric residents and faculty about domestic violence. Arch Pediatr Adolesc Med. 2002;156 :804 –810
Maguire P, Faulkner A, Booth K, Elliott C, Hillier V. Helping cancer patients disclose their concerns. Eur J Cancer. 1996;32A :78 –81
Lynch TR, Wildman BG, Smucker WD. Parental disclosure of child psychosocial concerns: relationship to physician identification and management. J Fam Pract. 1997;44 :273 –280
Ambulatory Pediatric Association. Educating Pediatric Residents to Provide Health Care to Underserved Children. Alexandria, VA: Ambulatory Pediatric Association; 1990
Epner JEG, Levenberg PB, Schoeny ME. Primary care providers' responsiveness to health-risk behaviors reported by adolescent patients. Arch Pediatr Adolesc Med. 1998;152 :774 –780
Heneghan AM, Mercer MB, DeLeone NL. Will mothers discuss parenting stress and depressive symptoms with their child's pediatrician Pediatrics. 2004;113 :460 –467(Lawrence S. Wissow, MD, M)