improving communication skills—a randomized …€¦ · improving communication skills—a...

9
Improving Communication Skills—A Randomized Controlled Behaviorally Oriented Intervention Study for Residents in Internal Medicine WOLF A. LANGEWITZ, MD, PHILIPP EICH, MD, ALEXANDER KISS, MD, AND BRIGITTA WOSSMER, PHD Objective: We investigated whether patient-centered communication skills can be taught to residents in Internal Medicine by using a time-limited behaviorally oriented intervention. Method: Residents working at the Department of Internal Medicine were randomly assigned to an intervention group (IG; N = 19) or a control group (CG; N = 23). In addition to 6 hours of standard medical education per week, the IG received specific communication training of 22.5 hours duration within a 6-month period. Initially and 10 months later, participants performed interviews with simulated patients. Interviews were rated by blinded raters who used the Maastricht History and Advice Checklist-Revised. Results: Compared with the CG, the IG improved substantially in many specific communication skills. Both groups improved in the "amount of medical information identified" and in the ability to "communicate about feasibility of treatment." Conclusion: Patient-centered communication skills such as those presented in this intervention study can be taught. The ability to gain medical information and the readiness to communicate about aspects of medical treatment seem to improve with more professional experience; however, they also profit from the intervention. Key words: communication skills, physician-patient relationship, patient-centered communication, intervention, simulated patients. PSQ = Patient Satisfaction Questionnaire; SP = simulated patients; GAS = Goal Attainment Scales; MAAS-R = Maastricht History and Advice Checklist- Revised; CME = Continuous Medical Education. INTRODUCTION The attainment of a common reality between two individ- uals is of paramount importance in the communication be- tween patient and physician; it is the basis for a psychoso- matic, integrative way of thinking in medicine (1). Many studies have shown that explanatory models for illness, therapeutic goals, or needs related to a single consultation differ between patient and physician; different standpoints are not the exception but the rule: On average, 50% of patients in ambulatory General Internal Medicine do not get what they want from a single consultation (2-4). It seems that physicians have difficulties listening to their patients; they underestimate the functional disability of their patients (5), they underesti- mate the information needs of patients (6), and they are more reluctant to initiate discussions about advance directives than the patients themselves (7-9). Patients are of little help in correcting physicians' misconceptions inasmuch as most of them seem very satisfied when asked to evaluate their physi- cians by using patient satisfaction questionnaires (10-12). Furthermore, physicians are unable to predict their patients' satisfaction with medical care (13). Recently, analyses into the motives of patients who sue their doctors have shown that complaints about bad communication are as important as complaints about bad treatment (14-18). In some studies, concepts of improving communication in the physician-patient interaction (19-21) have been proposed. It has been suggested that physicians should attempt to identify their patients' perspective (19), specific needs and feelings (22), their beliefs about diagnostic tests (23), thera- peutic options (24), and their health models (eg, Ref. 25). Essentially, these studies propagate a shift in the physician- patient relationship away from a mainly directive, physician- From the Division of Psychosomatic Medicine, Departments of Internal Medicine (W.A.L., A.K., B.W.) and Psychiatry (P.E.), University Hospital Basel, Basel, Switzer- land. Address reprint requests to: W. Langewitz, MD, Division of Psychosomatic Medicine, Department of Internal Medicine, Petersgraben 4, CH-4031 Basel, Switzerland. Received for publication January 6, 1997; revision received May 28, 1997. centered interaction style toward a nondirective, more patient- centered communication. This type of communication invites the patient to participate in decision-making and, in case of differences, includes the option to negotiate about the differ- ences. So far, intervention programs have used quite different outcome measures: these include patient satisfaction scores (26, 27), psychiatric knowledge (28), self-assessed compe- tence (29), or linguistic discourse characteristics (30). Other research groups have shown that patients benefit from a communication style that includes their active partic- ipation. Patients who are more actively involved in their health care perceive less discomfort and greater alleviation of their symptoms; they show greater improvement in their general medical condition (31-33). Also, their compliance is likely to improve (34-39). The more patients are involved in decision- making the less they consult other physicians for a second opinion (40). Not only patients, but also physicians will benefit from communication techniques that enable patient and physician to discuss their different positions explicitly: Sharpe et al. (41) concluded from their study on "difficult patients" that problems arise when patients and physicians have different expectations concerning the amount of infor- mation exchanged, the nature of the disease, and treatment options. When asked to comment on critical events in their daily practice, ie, moments when a job-related task was either particularly well done or was associated with specific diffi- culties, General Practitioners (42), General Practitioner train- ees (43), and house officers in their first clinical year (44) all mentioned communication problems between patients and physicians. To facilitate a negotiation process between pa- tients and physicians, we set out to determine whether a time-limited, behaviorally oriented intervention is likely to enhance a patient-oriented communication style. This study investigates, by way of a controlled randomized intervention, whether residents' abilities to use patient-centered communi- cation techniques can be improved by a time-limited interven- tion. This question is investigated by directly monitoring physician-patient interactions before and after the intervention in interviews with standardized patients. METHODS Study Population and Study Design All of the residents of the Department of Internal Medicine who had a contract lasting for at least 18 more months were asked to participate. Of 54 residents, 47 were eligible, 3 refused participation, and 1 left the clinic before the intervention started. There were 23 residents randomized into the control group and 20 into the interven- 268 0033-3174/98/6O03-0268$O3.O0/0 Copyright © 1998 by the American Psychosomatic Society Psychosomatic Medicine 60:268-276 (1998)

Upload: ledien

Post on 02-Jul-2018

230 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Improving Communication Skills—A Randomized …€¦ · Improving Communication Skills—A Randomized Controlled Behaviorally Oriented Intervention Stud for ... communication skill

Improving Communication Skills—A Randomized Controlled BehaviorallyOriented Intervention Study for Residents in Internal MedicineWOLF A. LANGEWITZ, MD, PHILIPP EICH, MD, ALEXANDER KISS, MD, AND BRIGITTA WOSSMER, P H D

Objective: We investigated whether patient-centered communication skills can be taught to residents in Internal Medicine by usinga time-limited behaviorally oriented intervention. Method: Residents working at the Department of Internal Medicine wererandomly assigned to an intervention group (IG; N = 19) or a control group (CG; N = 23). In addition to 6 hours of standardmedical education per week, the IG received specific communication training of 22.5 hours duration within a 6-month period.Initially and 10 months later, participants performed interviews with simulated patients. Interviews were rated by blinded raters whoused the Maastricht History and Advice Checklist-Revised. Results: Compared with the CG, the IG improved substantially in manyspecific communication skills. Both groups improved in the "amount of medical information identified" and in the ability to"communicate about feasibility of treatment." Conclusion: Patient-centered communication skills such as those presented in thisintervention study can be taught. The ability to gain medical information and the readiness to communicate about aspects of medicaltreatment seem to improve with more professional experience; however, they also profit from the intervention. Key words:communication skills, physician-patient relationship, patient-centered communication, intervention, simulated patients.

PSQ = Patient Satisfaction Questionnaire; SP =simulated patients; GAS = Goal Attainment Scales;MAAS-R = Maastricht History and Advice Checklist-Revised; CME = Continuous Medical Education.

INTRODUCTIONThe attainment of a common reality between two individ-

uals is of paramount importance in the communication be-tween patient and physician; it is the basis for a psychoso-matic, integrative way of thinking in medicine (1). Manystudies have shown that explanatory models for illness,therapeutic goals, or needs related to a single consultationdiffer between patient and physician; different standpoints arenot the exception but the rule: On average, 50% of patients inambulatory General Internal Medicine do not get what theywant from a single consultation (2-4). It seems that physicianshave difficulties listening to their patients; they underestimatethe functional disability of their patients (5), they underesti-mate the information needs of patients (6), and they are morereluctant to initiate discussions about advance directives thanthe patients themselves (7-9). Patients are of little help incorrecting physicians' misconceptions inasmuch as most ofthem seem very satisfied when asked to evaluate their physi-cians by using patient satisfaction questionnaires (10-12).Furthermore, physicians are unable to predict their patients'satisfaction with medical care (13). Recently, analyses into themotives of patients who sue their doctors have shown thatcomplaints about bad communication are as important ascomplaints about bad treatment (14-18).

In some studies, concepts of improving communication inthe physician-patient interaction (19-21) have been proposed.It has been suggested that physicians should attempt toidentify their patients' perspective (19), specific needs andfeelings (22), their beliefs about diagnostic tests (23), thera-peutic options (24), and their health models (eg, Ref. 25).Essentially, these studies propagate a shift in the physician-patient relationship away from a mainly directive, physician-

From the Division of Psychosomatic Medicine, Departments of Internal Medicine(W.A.L., A.K., B.W.) and Psychiatry (P.E.), University Hospital Basel, Basel, Switzer-land.

Address reprint requests to: W. Langewitz, MD, Division of Psychosomatic Medicine,Department of Internal Medicine, Petersgraben 4, CH-4031 Basel, Switzerland.

Received for publication January 6, 1997; revision received May 28, 1997.

centered interaction style toward a nondirective, more patient-centered communication. This type of communication invitesthe patient to participate in decision-making and, in case ofdifferences, includes the option to negotiate about the differ-ences. So far, intervention programs have used quite differentoutcome measures: these include patient satisfaction scores(26, 27), psychiatric knowledge (28), self-assessed compe-tence (29), or linguistic discourse characteristics (30).

Other research groups have shown that patients benefitfrom a communication style that includes their active partic-ipation. Patients who are more actively involved in their healthcare perceive less discomfort and greater alleviation of theirsymptoms; they show greater improvement in their generalmedical condition (31-33). Also, their compliance is likely toimprove (34-39). The more patients are involved in decision-making the less they consult other physicians for a secondopinion (40). Not only patients, but also physicians willbenefit from communication techniques that enable patientand physician to discuss their different positions explicitly:Sharpe et al. (41) concluded from their study on "difficultpatients" that problems arise when patients and physicianshave different expectations concerning the amount of infor-mation exchanged, the nature of the disease, and treatmentoptions. When asked to comment on critical events in theirdaily practice, ie, moments when a job-related task was eitherparticularly well done or was associated with specific diffi-culties, General Practitioners (42), General Practitioner train-ees (43), and house officers in their first clinical year (44) allmentioned communication problems between patients andphysicians. To facilitate a negotiation process between pa-tients and physicians, we set out to determine whether atime-limited, behaviorally oriented intervention is likely toenhance a patient-oriented communication style. This studyinvestigates, by way of a controlled randomized intervention,whether residents' abilities to use patient-centered communi-cation techniques can be improved by a time-limited interven-tion. This question is investigated by directly monitoringphysician-patient interactions before and after the interventionin interviews with standardized patients.

M E T H O D SStudy Population and Study DesignAll of the residents of the Department of Internal Medicine who

had a contract lasting for at least 18 more months were asked toparticipate. Of 54 residents, 47 were eligible, 3 refused participation,and 1 left the clinic before the intervention started. There were 23residents randomized into the control group and 20 into the interven-

2680033-3174/98/6O03-0268$O3.O0/0Copyright © 1998 by the American Psychosomatic Society

Psychosomatic Medicine 60:268-276 (1998)

Page 2: Improving Communication Skills—A Randomized …€¦ · Improving Communication Skills—A Randomized Controlled Behaviorally Oriented Intervention Stud for ... communication skill

IMPROVING COMMUNICATION IN INTERNAL MEDICINE

tion group. After the initial seminar, one resident was lost tofollow-up because he moved to another hospital. The control groupreceived the standard medical training offered to all residents in ourclinic (on the average, 6.5 hours per week), the intervention groupreceived the standard training plus the specific intervention describedin this study. Subsequently, the subjects performed two videotapedinterviews with simulated patients, 3 weeks before the interventionstarted. Then the residents were randomized into the control or theintervention group. Ten months later all remaining subjects (N = 42)again performed two videotaped interviews with simulated patients.There was at least a 2-month interval between the last period ofpreceptor feedback and the second round of interviews. In bothgroups, professional characteristics and sociodemographic variableswere assessed.

Description of the InterventionTeaching Objectives. A list of teaching objectives and of specific

related skills is provided in Table 1. To offer a negotiation process orto invite patient participation in decision making, residents must helppatients clarify their concerns. Necessary communication skillsinclude techniques of active listening (eg, Ref. 45) and willingness tounderstand patients' expectations precisely. Negotiating is easierwhen the setting of the interview is clear. Therefore, residents areinstructed to clarify time limits and to announce explicitly a changein the topic and the structure of the communication, eg, by announc-ing a shift from a patient-centered phase to a physician-centered part.Furthermore, residents are advised to offer a shared assessment of theconsultation at the end of the interview. If patients are to shareresponsibility for their care, they need information. Thus, residentsshould deliberately provide as much information as is appropriate forthem and for their patients. To elicit relevant information in themedical and psychosocial domain, residents were encouraged also touse patient-centered communication techniques. The interventionwas performed by four trainers who received written instructionsexplaining intervention goals and counseling techniques, for exam-ple, on how to give feedback (46). Two of them were internists(W.A.L., A.K.), one was a clinical psychologist (B.W.), and one wasa psychiatrist (P.E.).

Intervention Elements. The intervention consists of three ele-ments: a 1.5-day seminar (14 hours), six progress assessmentmeetings (6 X 45 minutes), using GAS (47), and six one-to-onefeedback sessions during routine clinical work (6 X 20 minutes extratime).

Seminar. To increase awareness of possible short-comings inphysician-patient communication, the initial 1.5-day workshop startswith plenary meetings presenting the current knowledge in doctor-patient relationships.

Individual communication skills deficits and strengths are illus-trated by the demonstration of instructive segments of participants'interviews with simulated patients in small group sessions (fiveparticipants/trainer). After outlining intervention goals for certain

segments of the interview (eg, the use of summarizing for the initialphase; see Table 1), the trainer hands out written transcripts of thevideo segments just shown. These are then used as the basis forrole-plays, allowing each participant first to reexperience his or herbehavior from the video clip as the patient or as the doctor, and thento replay the same sequence displaying alternative communicationtechniques.

A quarter of small group working time is devoted to theformulation of individual behavior goals. Each participant receives abooklet containing forms of Goal Attainment Scales that must befilled in with the current status of a distinct communication skill, thedefinite behavioral goal, and the first step toward the achievement ofthe definite behavioral goal.

Progress Assessment Meetings. In the next 6 to 8 months after theinitial seminar, these behavioral goals are checked six times in smallgroup sessions. By using the GAS protocols, participants assess theextent that they realized behavioral goals and the difficulties thatoccurred in the meantime. If possible, new behavioral goals aredefined, or existent behavioral goals are refined to increase thelikelihood of their realization.

One-to-One Preceptor Feedback. The third element of the inter-vention consists of six sessions of one-to-one feedback during routineclinical work of residents in the intervention group. This interventiontypically lasted for about 1.5 hours of being with the resident eitherduring their ward-rounds or during interviews with outpatients,followed by a 20-minute session during which the observed behaviorwas fed back to the resident. One-to-one preceptor feedback wasgiven by the two internists and the clinical psychologist, because thepsychiatric colleague was not available for this part of the interven-tion.

Evaluation. Central to the evaluation was the comparison of pre-to postintervention interviews with SP. Four actors were trained topresent a common medical problem plus a psychosocial problem thatwas important either for the patient's health-seeking behavior (eg,reluctance to be referred to the hospital), for low compliance, or thehopes and expectations of the SP from the encounter. Residents wereinformed that they would see standardized patients with the follow-ing instruction: "You may use up to 20 minutes consultation time toestablish a first differential diagnosis and to plan the next steps intreatment and diagnosis of a new patient who has just arrived at theoutpatient department. A physical examination will not be video-taped."

Interview Rating. Three third-year medical students received a1-day training course in the use of the MAAS-R (48), to rate theinterviews. Basically, MAAS-R contains two types of scores: globalscores ranging from 0 (does not occur), 1 (bad performance) to 5(very good performance), that rate either specific behaviors or thequality of, eg, data-gathering, and checklists where the occurrence ofa certain behavior or the mention of specific information is marked(see Table 2). Because medical and psychosocial information varies

TABLE 1. Teaching Objectives and Specific Related Skills

Teaching objective Specific skills

Help the patient clarify his/her concerns

Find relevant information

Offer a negotiation process

Invite patient participation in decisionmaking

Empathy in greeting behaviorAcknowledging initial complaintsTake up emotions instead of suppressing themClarify consultation reasonsSummarizing patient's statement in doctor's own wordsSummarizing, acknowledging initial complaints, clarify purpose of the consultationTake up emotions, structuring the consultationExplicit announcement of history-taking phaseGenerally structuring the consultationShared evaluation of the consultationConvey information as detailed as possible and as desired by the patient, ie, results and preliminary diagno-

sis, etiology, prognosis; communicate about treatment options, feasibility of treatment, future prospects

Psychosomatic Medicine 60:268-276 (1998) 269

Page 3: Improving Communication Skills—A Randomized …€¦ · Improving Communication Skills—A Randomized Controlled Behaviorally Oriented Intervention Stud for ... communication skill

W. A. LANGEWITZ et al.

TABLE 2. Maastricht History and Advice Checklist-Revised: Examples for Different Types of Items, Operationalization, andInstructions for Raters

Item description Operationalization Instructions for raters

Items using global scores concerning a specificbehavior or quality aspects. Example: Takeup emotions

Items using global scores concerning the entireconsultsition. Example: Patient-centeredcommunication style

Script-specific items (ie, ability to elicit historyin script No. 1: A young woman with irrita-ble bowel syndrome). Example: Characterizecomplaints

Resident names emotionsResident takes up emotional utterances, pursues

emotional utterancesResident asks for emotional correlates of symp-

tomsResident elicits patients' assumptions concern-

ing: His/her subjective explanation model forsymptoms, his/her assumptions about diagno-sis and treatment

Resident follows patients' ideasResident checks patients' understandingSevere, cramp-like left-sided pain in the abdo-

menDuring pain-free intervals heightened sensitivityVarying composition of bowelsNo blood smearsColor and odor normalChange in body weight ± 2 kg/weekNo feverNo joint pain

Provide a global rating from 0 = does not occur,1 = bad performance, to 5 = very good per-formance

Provide a global rating from 0 = does not occur,1 = bad performance, to 5 = very good per-formance

Provide a mark whenever the information ismentioned during the interview

between scripts, checklists were constructed for each interviewseparately, and contained script-specific items (see Table 2 for anexample). The last three questions on the evaluation form ask forglobal statements concerning the entire consultation. Raters wereblind as to the group assignment of residents (control vs. interventiongroup) and to the time of the interview (pre- vs. postinterventioninterview).

Patient Satisfaction. To obtain some information from the pa-tient's perspective, SP were asked to complete the German version ofthe 14-item PSQ issued by the American Board of Internal Medicineand used in various studies (eg, Refs. 9, 10, 49). The PSQ alsocontains two dichotomous (yes/no) variables, asking "Would yourecommend this doctor to a close friend or a relative?" and "Did youlike this doctor?". The internal consistency of the German version issatisfactory (Crohnbach's a = 0.71) (12). The Patient SatisfactionQuestionnaires were completed immediately after the interview bythe SPs. Standardized patients did not know whether a residentbelonged to the control or to the intervention group.

Data Reduction and Statistical AnalysesInterview Performance. To assess interview performance, first the

mean score of all three raters per interview was calculated, second themean of the scores from two pre- and two postintervention interviewswas calculated. Finally, a sum score of all medical informationobtained during an interview was calculated from the specific contentvariables. Furthermore, as suggested by the authors of MAAS-R (48),global scores on interview behavior were combined to yield a generalfactor "communication skills." To assess differences in scoresbetween groups from the MAAS-R protocol at one time of assess-ment, the Mann-Whitney ranked sum test was used. To assessdifferences between pre- and postintervention consultations withinthe control and within the intervention group, Wilcoxon's signedrank tests were computed.

Between-group differences in sociodemographic data and profes-sional characteristics were assessed by independent t tests or x2 testsfor interval and nominal data, respectively. The dichotomous vari-ables of the PSQ were analyzed by computing the odds ratio for theproportion of "yes/no" answers before and after the intervention (50).To analyze the relationship between professional characteristics,communication skills, the amount of medical or psychosocial infor-mation, and patient satisfaction, Spearman's rank correlations werecomputed.

According to the recommendations of Grant (51), the reliability ofthe MAAS-R rating form was assessed by calculating mean interrateragreement as mean percent agreement between three rater pairs.Interrater agreement was assumed if the difference between tworaters did not exceed ± 1 score.

Mean interrater agreement was 88.5% (raters 1 and 2, 86.1%;raters 1 and 3, 89.6%; raters 2 and 3, 90%), with 18 of 27 items above90%, three items 80 to 90%, four items 70 and 80% and two items 60and 70%.

To increase the comparability among script-specific scores, theseitems were normalized to a scale ranging from 0 to 10. All statisticalanalyses were run on SPSS-Windows Programs (52). For data afterthe communication skills training, two-tailed significance will bereported.

RESULTSSociodemographic CharacteristicsTable 3 lists the sociodemographic characteristics of study

participants, divided into control and intervention groups.They represent a typical sample of Swiss residents in InternalMedicine at a major hospital. Statistical analysis revealed nosignificant differences between groups.

TABLE 3. Sociodemographic and Professional Characteristics ofControl and Intervention Groups

Age (yr)Female/male residentsMedical finalsClinical experience (yr)Experience in internal medicine (yr)Years at Department of Internal MedicineParticipation in interview skills groups as

studentPsychiatric training

Controlgroup

33.5 ± 3.24/19

1986 ± 2.44.7 ± 2.33.1 ± 2.21.2 ± 1.2

4/19

1/22

Interventiongroup

32.1 ± 1.84/15

1986 ± 2.34.6 i 1.93.0 ± 1.51.2 ± 1.1

4/15

1/18

Psychosomatic Medicine 60:268-276 (1998)

Page 4: Improving Communication Skills—A Randomized …€¦ · Improving Communication Skills—A Randomized Controlled Behaviorally Oriented Intervention Stud for ... communication skill

IMPROVING COMMUNICATION IN INTERNAL MEDICINE

Communication SkillsThe presentation of the results from the videotaped consul-

tations follows the list of teaching objectives (Tables 4 to 7).In addition, results from global ratings concerning the entireinterview are given in a separate table (Table 8).

Concerning the teaching objective "help the patient clarifyhis or her concerns," Table 4 shows that pre- and postinter-vention results in the control group (columns 2 and 3) do notdiffer substantially, whereas the intervention group improvesin most items (columns 5 and 6). This is also reflected in thenonsignificant p values for the Wilcoxon's signed rank test inthe control group (column 4), and significant p values for thecomparison of pre- and postintervention values in the inter-vention group (column 7). Bold numbers in columns 3 and 6indicate significant differences between control and interven-tion groups after the intervention (Mann-Whitney t/-test).Similar results are given in Tables 6 and 7: there is asubstantial improvement in most items relating to the teachingobjectives "invite patient participation in decision-making"(see Table 6) and "offer a negotiation process" (see Table 7)in the intervention group, whereas the control group does notimprove substantially.

Most script-specific items relating to the teaching objective"find relevant information" (see Table 5) demonstrate parallelimprovements in the control and intervention groups; the sameholds for items concerning communication about treatmentoptions and future prospects (see Table 6). Concerning thesum score of medical facts, both the control group and theintervention group obtain more medical information at thesecond series of interviews. However, the improvement in theintervention group is more pronounced, yielding a signifi-cantly higher score than the control group after the interven-tion. Concerning the sum score of psychosocial facts obtained,there is no intervention effect in either group: scores remainbasically unchanged.

Global items relating to the entire consultation mainlycorroborate the more specific data from Tables 4 to 7 (seeTable 8): raters thought that residents structured the consul-tation better, used a more patient-centered communicationstyle, and were more willing to involve patients in decision-making after the intervention.

Patient SatisfactionA major short-coming of PSQ results in general is their

skewed distribution (10-12). However, compared with realpatients (mean PSQ score: 60.1 ± 7.1 points) (12), scores ofsimulated patients are distributed slightly more evenly, adding

to their methodological reliability. Overall, simulated patientsare satisfied with the interviews: before the intervention meanPSQ scores are 51.7 ± 7.5 for the control group and 56.4 ±6.7 for the intervention group (maximum possible score: 70points); after the intervention, the respective scores are 54.1 ±6.3 and 60.5 ± 8.6. There is no significant Time by Groupinteraction in single items of the 14-item questionnaire or inthe total score. However, there is a significant increase in theproportion of positive answers to the dichotomous question:"Would you recommend this doctor to a close friend or arelative?" The proportion of SP answering "yes" increasedfrom 23 of 38 interviews to 31 of 38 interviews in theintervention group, and remained stable with 21 of 46 inter-views before and 24 of 46 interviews after the intervention inthe control group. Whereas the 95% confidence intervals forthe odds ratio of "yes" and "no" answers includes 1 before theintervention, confidence intervals are 1.36 to 13.0 for theobserved odds ratio of 4.06 after the intervention.

Communication Skills, Patient Satisfaction,Amount of Information Identified, andProfessional CharacteristicsThe general indicator "communication skills" is correlated

with the amount of medical and psychosocial informationelicited during the interview: r = .47 (p = .002) and r — .45(p = .003), respectively. Of the global items covering theentire consultation "consultation structure" is related to theidentification of medical information (r = .52; p < .001). Thepatient satisfaction score is related to the general factor"communication skills" (r = .68; p < .001) as well as to allthree global items rating the entire consultation: "consultationstructure" (r = .65; p < .001); "patient-centered communica-tion style" (r = .68; p < .001); and "involvement of patientsin decision-making" (r = .69; p < .001). Of specific commu-nication skills, "clarify consultation reasons" was the skilldemonstrating the highest correlation with patient satisfaction(r = .71; p < .001). None of the professional characteristics,such as years of clinical experience, was correlated with thesum score of medical facts, the sum score of psychosocialfacts, with patient satisfaction, or with the general indicator"communication skills."

There are, however, some positive correlations: years ofclinical experience is related to "information concerningprevious treatment" (r = .38; p = .012), "conveying prognos-tic information" (r = .33; p = .035), and "communicatingabout the feasibility of treatment" (r = .27; p = .09).

TABLE 4. Results on Items Relating to the Teaching Objective: Help the Patient Clarify His or Her Concern

Interview evaluation item

Empathy in greeting behaviorAcknowledging initial complaintsTake up emotionsUse of checkingClarify consultation reasons

Control group" (N = 23)

Preintervention

2.96 ± 0.492.69 ± 0.692.53 ± 1.071.60 ± .802.81 ±.67

Postintervention

3.07 ± 0.532.89 ± 0.682.30 ± 0.791.08 ± 0.873.08 ± 0.72

Wilcoxon'sp Value"

NSNSNSNSNS

Intervention

Preintervention

3.07 ;2.93 ;2.67;1.75 ;3.15;

i 0.53tO.64t 1.27t 0.86;0.83

groupb (/V = 19)

Postintervention

3.23 ± 0.453.60 ± 0.98; p = .0053.34 ± 0.91 ;p < .0014.00 ± 1.12; p < .0014.00 ± 0.74; p<. 001

Wilcoxon'sp Value"

NS/? = .0llp = .07p < . 0 0 l

"To compare within-groups differences between pre- and postintervention values, the Wilcoxon's signed rank test was computed (see columns 4 and 7).b To compare between-groups differences at one time of assessment, the Mann-Whitney (/-test was performed. Bold numbers indicate significant differences: p valuesappear after the bold numbers in column 6. Range of scores, 0 to 5.

Psychosomatic Medicine 60:268-276 (1998)

Page 5: Improving Communication Skills—A Randomized …€¦ · Improving Communication Skills—A Randomized Controlled Behaviorally Oriented Intervention Stud for ... communication skill

W. A. LANGEWITZ et al.

TABLE 5. Results on Script-Specific Items Relating to the Teaching Objective: Find Relevant Information

Interview evaluation item

Characterize complaintsTime course of complaintsEtiological factorsImpact on daily lifePrevious treatmentFamily historyPersonal historyAddiction, drugsSystem historySum score of medical factsSum score of psychosocial facts

Control group" (N = 23)

Preintervention

6.70 ± 1.327.34 ± 1.496.86 ± 1.525.35 ± 2.234.63 ±2.213.80 ±.1535.86 ± 2.876.49 ± 3.236.33 ± 2.865.51 ± 1.364.27 ± 1.73

Postintervention

8.50 ± 0.799.31 ±0.865.65 ± 1.826.00 ±2.516.53 ± 3.264.28 ± 2.747.89 ± 2.926.31 ±2.715.24 ± 3.316.10 ± 1.073.92 + 1.64

Wilcoxon'sp Value"

p < .001p < .001p = .02/? = .02p = .004

NSp = .002

NSp = .05p = .013

NS

Intervention ;

Preintervention

6.66 ± 1.178.39 ± 1.51;/? = .016.87 ± 1.664.20 ±2.38;/? = .104.31 ±2.204.56 ± 2.744.61 ±2.87;/? = .106.89 ± 2.656.60 ± 2.645.46 ± 1.214.79 ± 1.52

group" (N = 19)

Postintervention

8.87 ± 1.089.23 ± 1.156.42 ± 1.57; p =6.42 + 2.266.73 ± 3.265.31 + 2.838.96 ± 1.466.51 ± 3.606.68 ± 2.81; p =6.89 ± 1.14;/? =4.93 ± 1.78;/? =

.12

.15

.02

.14

V

PP

PP

P

P

Vilcoxon'sp Value"

< .001= .07

NS= .01= .03

NS< .001

NSNS

= .002NS

" To compare within-groups differences between pre- and postintervention values, the Wilcoxon's signed rank test was computed (see columns 4 and 7).h To compare between-groups differences at one time of assessment, the Mann-Whitney (/-test was performed. Bold numbers indicate significant differences; p valuesappear after the bold numbers in column 5 and 6. Range of normalized scores, 0 to 10.

TABLE 6. Results on Items Relating to the Teaching Objective: Invite Patient Participation in Decision-Making

Interview evaluation item

Convey information aboutResults, preliminary diagnosisEtiology of symptomsPrognosis

Communicate aboutTreatment optionsFeasibility of treatmentFuture prospects

Control group" (N = 23) Intervention group" (/V = 19)Wilcoxon's

_ . . „ . . p Value" „ . . „ . p Value"Preintervention Postintervention Preintervention Postintervention

4.82 ± 2.934.60 ± 2.932.36 ± 1.99

1.97 ± 1.071.69 ± 1.362.85 ± 1.36

5.96 ± 3.545.83 ± 3.542.43 ± 2.40

2.86 ± 1.512.80 ± 2.293.82 ± 1.65

P =P =P =

P =P =P =

.12

.13

.08

.02

.06

.04

3.73 ± 2.964.19 ± 2.741.60 ± 1.96

1.61 ± 1.141.11 ± 1.082.85 ± 1.36

7.32 ± 3.51;/? = .137.41 ± 3.17;p = .123.29+ 1.95;/? = .21

4.23±1.59,p = .0094.84 ±2.04,/? = .0045.39 ± 2.7,5; p = .02

p = .004p = .002p = .01

p < .001p < .001

00

"To compare within-groups differences between pre- and postintervention values, the Wilcoxon's signed rank test was computed (see columns 4 and 7)." To compare between-groups differences at one time of assessment, the Mann-Whitney (/-test was performed. Bold numbers indicate significant differences; p valuesappear after the bold numbers in column 6. Range of normalized scores, 0 to 10.

TABLE 7. Results on Items Relating to the Teaching Objective: Offer a Negotiation Process

Interview evaluation item

Explicit announcement of history-taking phaseGenerally structuring the conversationShared assessment of the consultation

Control group" (N = 23)

Preintervention

1.74 ± 2.052.28 + 0.601.98 ± 1.51

Postintervention

1.30 ± 1.582.55 ± 0.611.79 ± 1.73

Wilcoxon'sp Value"

NSNSNS

Intervention group" (N = 19)

Preintervention

1.49+ 1.722.53 ± 0.671.67 ± 1.68

Postintervention

3.79 ± 1.61; p < .0013.81 ± 0.73; p < .0013.61 ± 0.77; p < .001

Wilcoxon'sp Value"

p .001/?<.00lp = .002

"To compare within-groups differences between pre- and postintervention values, the Wilcoxon's signed rank test was computed (see columns 4 and 7)." To compare between-groups differences at one time of assessment, the Mann-Whitney (/-test was performed. Bold numbers indicate significant differences; p valuesappear after the bold numbers in column 6. Range of scores, 0 to 5.

DISCUSSIONOutcome of the InterventionOur study shows that the training program yields substan-

tial improvements in communication skills. Some items suchas those listed in Table 7 ("offer a negotiation process") revealno improvement in the control group and a significant im-provement in the intervention group. Changes in these itemsmay be attributed exclusively to the intervention. Other itemssuch as those relating to communication about future treat-ment (Table 6) display two significant changes: first, bothgroups improve over time; and second, the intervention groupscores significantly better than the control group after the

intervention. These results point to an interaction of twofactors: some skills improve with time and with the acquisi-tion of specific skills. A third category of results shows noadvantage for the intervention group compared with thecontrol group after the intervention, but significant improve-ments over time in both groups; this relates namely to theteaching objective, "find relevant information" (see Table 5).With regard to the presumed time-effect, one may assume thatclinical experience per se does improve some communicationskills, specifically those related to the identification of medicalinformation and those related to the communication of treat-ment options. This interpretation is supported by the observed

Psychosomatic Medicine 60:268-276 (1998)

Page 6: Improving Communication Skills—A Randomized …€¦ · Improving Communication Skills—A Randomized Controlled Behaviorally Oriented Intervention Stud for ... communication skill

IMPROVING COMMUNICATION IN INTERNAL MEDICINE

TABLE 8. Results on Global Items Rating the Entire Consultation

Global evaluation item

Consultation structurePatient-centered communication styleInvolvement of patients in decision-making

Control groupb (N = 23)

Preintervention

2.28 ± 0.602.67 ± 0.812.66 ± 0.83

Postintervention

2.5S ± 0.613.14 ± 0.852.89 ± 0.98

Wilcoxon'sp Value"

p = .09p = .09

NS

Intervention

Preintervention

2.44 ± 0.833.03 ± 0.892.95 ± 0.94

groupb(W = 19)

Postintervention

3.87 ± 0.77; p< .0013.92 ± 0.74; p = .0024.01 ± 0.74;p<.001

Wilcoxon'sp Value"

p < .001p = .002p < .001

" To compare within-groups differences between pre- and postintervention values, the Wilcoxon's signed rank test was computed (see columns 4 and 7).bTo compare betwcen-groups differences at one time of assessment, the Mann-Whitney (/-test was performed. Bold numbers indicate significant differences;p values appear after the bold numbers in column 6. Range of scores, 0 to 5.

positive correlations between years of clinical experience andsome of the items listed in Tables 5 and 6. However, thecorrelational analyses also reveal that specific communicationskills additionally improve the ability of residents to identifyrelevant medical and psychosocial information.

The comparison of our results with those of similar studiesshows that behavior change either was not documented oroccurred to a lesser extent than in the present investigation:Roter et al. (53) showed that after a 1-month rotation perioddevoted to interviewing skills and the psychosocial domain inmedical practice, interns in the intervention group used lessleading questions and more open questions in an interviewwith a simulated patient than interns in the control group. Amore recent publication by Smith et al. (29) describes a studythat randomly assigned 26 residents either to a control or anintervention group. Besides an increase in knowledge, resi-dents in the intervention group showed more self-reportedconfidence in using psychosocial skills; an attempt to monitora change in observed behavior, however, was not documented.Another recent study with an intensity of training similar toours has yielded mixed results. Levinson and Roter (30)describe the effects of a 2.5-day communication skills pro-gram on interview content and behavior: the number of openquestions asked by physicians increased significantly; how-ever, physicians spent relatively more time talking after theintervention, which was contrary to the authors' expectations.

Effectiveness of the InterventionOn the basis of spontaneous comments of residents during

the seminar and during one-to-one preceptor feedback, webelieve that the following points must be considered inexplaining the effectiveness of our intervention:

1. By seeing segments of their own interviews, participantsare immediately personally involved during the initialseminar. They do not discuss "communication in gen-eral" but are reminded of their observations concerning,for example, 45 seconds of the interview with Dr. B.,who repeatedly does not "take up" the patient's concern.

2. Once the deficit had been defined, participants tryalternative behaviors, replaying the consultation se-quence just seen and trying to "take up" emotions in theexample given above.

3. Trying alternative behavior often elicits uncomfortablefeelings in the physician and in participants who watchtheir colleagues during role-play. We believe that beingin the patients' position is a crucial experience forparticipants and the only means to comprehend certaincommunication techniques.

4. Participants varied according to their communicationskills. Therefore, it seemed important to have eachresident define one or two personally relevant behaviorgoals for a 4- to 6-week period after the initial workshopand then again between GAS meetings. One physician,for example, chose a basic behavior such as "summa-rizing" and decided to practice it at least twice withinthe initial 3 minutes of the consultation; another resi-dent, who already allowed the patient to clarify herconcerns, chose the behavioral goal "announce a shiftfrom the patient-centered to the physician-centeredphase of the interview."

5. The transfer from seminar behavior into clinical practiceis a difficult task. Accompanying residents in theirattempts to use alternative behavior during clinicalroutine seemed to encourage them; it helped to enlargeand redefine behavioral goals when appropriate. Indealing with "difficult" patient-physician interactions,the intervention goal of initiating a negotiation processbetween patient and physician had an immediate facevalue for most residents: it was evident that most"difficult" patients had expectations and hopes differentfrom those of residents or the ward team. Applying thenegotiation approach by explicitly clarifying treatmentgoals and by discussing treatment options was usuallyhelpful in settling difficult physician-patient interac-tions.

Identification of Medical and Psychosocial FactsCompared with some specific communication skills, the

intervention was not especially successful in increasing theamount of medical information identified during the inter-view; regarding psychosocial facts, no improvement could bedemonstrated. As mentioned above, the increase in the amountof medical facts identified in both groups may be due to anincrease in professional knowledge achieved during almost 1year of regular training in Internal Medicine. Another inter-pretation relates to the simulated patients: We are not surewhether the diseases presented during the second round ofinterviews (coronary artery disease, peptic ulcer, diabetes, andasthma) were more familiar to residents than those presentedduring the preintervention interviews (irritable colon, rheuma-toid arthritis, possible recent HIV infection, and upper gastro-intestinal bleeding).

Also, one must keep in mind that the percentage of 55 to69% of medical information identified by residents in thepresent study compares favorably with figures from a study byRoter and Hall (54): Experienced primary care practitioners

Psychosomatic Medicine 60:268-276 (1998)

Page 7: Improving Communication Skills—A Randomized …€¦ · Improving Communication Skills—A Randomized Controlled Behaviorally Oriented Intervention Stud for ... communication skill

W. A. LANCEWITZ et al.

(mean age, 47 years) identified on average 55% of medicalfacts during interviews with simulated patients performed aspart of an CME evaluation. Therefore, the lack of a morepronounced improvement in this regard could possibly also beattributed to a ceiling effect.

The same may hold for the lack of improvement in theidentification of psychosocial facts: Given the detailed psy-chosocial history of the simulated patients' scripts containing,for example, data on parents' psychological characteristics, onthe relationship with siblings, and facts on the patients'educational and professional development, we were surprisedto find that between 39 and 48% of psychosocial informationwas identified before the intervention; this may be due toresidents' assumptions that patients presented by the Divisionof Psychosomatic Medicine would be prone to have importantpsychosocial findings, thus increasing the attention given tothe psychosocial part of their interview.

Comparing the small but significant increase in the identi-fication of medical facts with the lack of an improvement inpsychosocial facts, one should also consider that residentswithin the setting of a traditional training in Internal Medicineare under powerful developmental forces favoring their skillsto elicit medical facts and to devote less time and attention tothe psychosocial domain. If our approach is to be replicated inother settings, we would suggest the following changes in theintervention: Instead of using simulated patients with acuteproblems only, who report for the first time to an outpatientclinic, half of the patients should present with a chronicproblem for which they have been hospitalized several timesin the last few months or which led them to see the physicianin the outpatient clinic several times before. If residents wereprovided with a complete list of previous findings before theinterview with these patients, they might be more inclined topay attention to psychosocial factors leading to repetitiveadmissions of chronic patients.

Study DesignAnother question is whether the effectiveness of our

intervention is due to some bias. Two error sources seemrelevant in the discussion of our data: selection bias and localfactors specific to the setting of our clinic.

First, when some kind of self-referral is possible or whenparticipants are not properly randomized, interventions suchas the one described in this study run the risk of "preaching tothe converted." Both faults were avoided in the presentinvestigation: We were able to include almost all residents atour clinic and randomization occurred after the initial consul-tation with simulated patients. Many other studies are charac-terized by a lack of randomization: In addition to the study byLevinson and Roter (30) and a recent publication by Smith etal. (29), Smith et al. (55) in 1994 could find only one studythat assessed the effect of an intervention in a randomizedfashion (53). Second, the fact that 44 of 47 eligible residentsparticipated in the study without exactly knowing what wouldhappen during the intervention or into which group theywould be randomized, points either to a high acceptance of ourservice within the Department of Internal Medicine in generalor perhaps even to the good reputation of the teachers. Thus,as long as our approach has not been replicated in otherinstitutions, we cannot definitely answer the question whetherthe efficacy of the program is due to the content, the teaching

methods of the program itself, or to the close relationship ofour group with the Department of Internal Medicine.

Cost of the InterventionA likely advantage of our teaching program was that

compared with most other studies and to the ones listed bySmith et al. (55), our intervention demands much less timefrom the residents with 22.5 hours extra time during a 6-monthperiod compared with 120 and 900 hours in other studies.However, for trainers the intervention was more time-consum-ing, reaching a total of 140 hours for the training of fiveresidents.

Deficits of the EvaluationA major limitation of the present study and of all other

intervention studies in this field is the lack of patient-derivedoutcome variables other than patient satisfaction data anddirectly observed physician behavior interacting with real orsimulated patients. Within the setting of our hospital, we wereunable to investigate whether better communication skillsresult in better health outcomes. The residents in our hospitalhave brief rotation periods and are assigned to a single ward,so that the reactions of patients treated by physicians from theintervention group could not be compared with those in thecontrol group. As mentioned in the introduction, however,there is some data showing that a more patient-centeredcommunication is related to better health outcome (31-39).One could, furthermore, argue that the quality of diagnosisand treatment depends on the quality of information gatheredduring an interview. Then the significant correlation betweenthe amount of medical and psychosocial facts identified andcommunication skills would suggest that a successful com-munication training program is related to a better health careoutcome.

We did not investigate whether the intervention will havelong-lasting effects on communication skills. This long-termeffect is possible, however, because a comparably limitedintervention in students was shown to have an impact oncommunication skills in young doctors, 5 years later (56, 57).

CONCLUSIONWe investigated whether a time-limited, highly structured,

and individualized intervention program can improve resi-dents' communication skills. We showed that residents canlearn to listen actively to patients, to invite the patient to sharedecisions, and to negotiate about the process of health care.Both patients and physicians profit from this approach.

The study was supported by Grant 32-36379.92 from theSwiss National Science Foundation.

We appreciate the support received from. Prof. Dr. WernerStauffacher, the former Head of the Department of InternalMedicine. We are very thankful for the editorial comments ofJ. Wyler-Harper, MD.

REFERENCES1. Von Uexkiill T: Anthropology and the theory of medicine. Theor

Med 16:93-114, 19952. Kravitz RL, Cope DW, Bhrany V, et al: Internal medicine

patients' expectations for care during office visits. J Gen InternMed 9:75-81, 1994

Psychosomatic Medicine 60:268-276 (1998)

Page 8: Improving Communication Skills—A Randomized …€¦ · Improving Communication Skills—A Randomized Controlled Behaviorally Oriented Intervention Stud for ... communication skill

IMPROVING COMMUNICATION IN INTERNAL MEDICINE

3. Sanchez-Menegay C, Stalder H: Do physicians take into accountpatient's expectations? J Gen Intern Med 9:404-406, 1994

4. Joos SK, Hickam DH, Borders LM: Patients' desires andsatisfaction in general medicine clinics. Public Health Rep108:751-759, 1993

5. Calkins DR, Rubenstein LV, Cleary PD, et al: Failure ofphysicians to recognize functional disability in ambulatorypatients. Ann Intern Med 114:451-454, 1991

6. Sardell AN, Trierweiler SJ: Disclosing the cancer diagnosis:Procedures that influence patient hopefulness. Cancer 72:3355-3365, 1993

7. Reilly BM, Magnussen CR, Ross J, et al: Can we talk? Inpatientdiscussions about advance directives in a community hospital.Arch Intern Med 154:2299-2308, 1994

8. Hill ME, MacQuillan G, Forsyth M, et al: Cardiopulmonaryresuscitation: Who makes the decision? Br Med J 308:1677,1994

9. Pfeifer MP, Sidorov JE, Smith AC, et al: The discussion ofend-of-life medical care by primary care patients and physicians.J Gen Intern Med 9: 882-888, 1994

10. Matthews DA, Feinstein AR: A new instrument for patients'ratings of physician performance in the hospital setting. J GenIntern Med 4:14-22, 1989

11. Roberts JG, Tugwell P: Comparison of questionnaires determin-ing patient satisfaction with medical care. Health Serv Res22:637-654, 1987

12. Langewitz W, Keller A, Denz M, et al: Patientenzufriedenheits-Fragebogen (PZF): Ein taugliches Mittel zur Qualitatskontrolleder Arzt-Patient-Beziehung? Psychother Psychosom Med Psy-chol 45:351-357, 1995

13. Merkel WT: Physician perception of patient satisfaction: Dodoctors know which patients are satisfied? Med Care 22:453-459, 1984

14. Beckman HB, Markakis KM, Suchman AL, et al: The doctor-patient relationship and malpractice: Lessons from plaintiffdepositions. Arch Intern Med 154:1365-1370, 1994

15. Levinson W: Physician-patient communication—A key to mal-practice prevention. J Am Med Assoc 272:1619-1620, 1994

16. Entman SS, Glass CA, Hickson GB, et al: The relationshipbetween malpractice claims history and subsequent obstetriccare. J Am Med Assoc 272:1588-1591, 1994

17. Hickson GB, Clayton EW, Entman SS, et al: Obstetrician's priormalpractice experience and patients' satisfaction with care. J AmMed Assoc 272:1583-1587, 1994

18. Vincent C, Young M, Phillips A: Why do people sue doctors? Astudy of patients and relatives taking legal action. Lancet343:1609-1613, 1994

19. Delbanco TL: Enriching the doctor-patient relationship by invitingthe patient's perspective. Ann Intern Med 116:414-418, 1992

20. Smith RC, Hoppe RB: The patient's story: Integrating thepatient- and physician-centered approaches to interviewing. AnnIntern Med 115:470-477, 1991

21. Lipkin M, Quill TE, Napodano RJ: The medical interview: Acore curriculum for residents in internal medicine. Ann InternMed 100:277-284, 1984

22. Zinn WM: Transference phenomena in medical practice: Beingwhom the patient needs. Ann Intern Med 113:293-298, 1990

23. Sox HC, Nease RF: When patients and doctors disagree. J GenIntern Med 8:580-581, 1993

24. Kassirer JP: Incorporating patients' preferences into medicaldecisions. N Engl J Med 330:1895-1896, 1994

25. Gillick MR: Common-sense models of health and disease.N Engl J Med 313:700-703, 1985

26. Putnam SM, Stiles WB, Casey Jacob M, et al: Teaching themedical interview. An intervention study. J Gen Intern Med3:38-47, 1988

27. Evans BJ, Kiellerup FD, Stanley RO, et al: A communicationskills programme for increasing patients' satisfaction with gen-eral practice consultations. Br J Med Psychol 60:373-378, 1987

28. Cohen-Cole SA, Boker J, Bird J, et al: Psychiatric educationimproves internists' knowledge: A three-year randomized, con-trolled evaluation. Psychosom Med 55:212-218, 1993

29. Smith RC, Mettler JA, Stoffelmayr BE, et al: Improving resi-dents' confidence in using psychosocial skills. J Gen Intern Med10:315-320, 1995

30. Levinson W, Roter D: The effects of two continuing medicaleducation programs on communication skills of practicing pri-mary care physicians. J Gen Intern Med 8:318-324, 1993

31. Brody DS, Miller SM, Lerman CE, et al: Patient perception ofinvolvement in medical care. J Gen Intern Med 4:506 -511, 1989

32. Kaplan SH, Greenfield S, Ware JE: Assessing the effects ofphysician-patient interactions on the outcomes of chronic dis-ease. Med Care 27:110-126, 1989

33. Greenfield S, Kaplan SH, Ware JE, et al: Patients' participationin medical care: Effects on blood sugar control and quality of lifein diabetes. J Gen Intern Med 3:448-457, 1988

34. Roter DL: Patient participation in patient provider interaction:The effects of patient question-asking on the quality of interac-tion, satisfaction, and compliance. Health Educ Monogr 5:281-315, 1977

35. Martin FJ, Bass MJ: The impact of discussion of non-medi-cal problems in the physician's office. Fam Pract 6:254-258,1989

36. Medication regimens: Causes of noncompliance. Washington,DC: Office of the Inspector General, Department of Health andHuman Services, June 1990, 12-14

37. Rost K, Carter W, Inui T: Introduction of information during theinitial medical visit: Consequences for patient follow-throughwith physician recommendations for medication. Soc Sci Med28:325-331, 1989

38. Bartlett EE, Grayson M, Barker R, et al: The effects of physiciancommunications skills on patient satisfaction; recall and adher-ence. J Chron Dis 37:755-764, 1984

39. Stewart MA: What is a successful doctor-patient interview? Astudy of interactions and outcomes. Soc Sci Med 19:167-175,1984

40. Kaplan SH, Greenfield S, Gandek B, et al: Characteristics ofphysicians with participatory decision-making styles. Ann InternMed 124:497-504, 1996

41. Sharpe M, Mayou R, Seagrott V, et al: Why do doctors findsome patients difficult to help? Q J Med 87:187-193, 1994

42. Windefield H, Murrell T, Clifford J: Sources of occupation stressfor Australian GP's, and their implications for postgraduatetraining. Fam Pract 11:413-417, 1994

43. Diamond MR, Kamien M, Sim MGB, et al: A critical incidentstudy of general practice trainees in their basic general practiceterm. Med J Aust 162:321-324, 1995

44. Caiman KC, Donaldson M: The pre-registration house officeryear: A critical incident study. Med Educ 25:51-59, 1991

45. Bird J, Cohen-Cole SA: The three-function model of the medicalinterview. Adv Psychosom Med 20:65-88, 1990

46. Ende J: Feedback in clinical medical education. J Am MedAssoc 250:777-781, 1983

Psychosomatic Medicine 60:268-276 (1998)

Page 9: Improving Communication Skills—A Randomized …€¦ · Improving Communication Skills—A Randomized Controlled Behaviorally Oriented Intervention Stud for ... communication skill

W. A. LANGEWITZ et al.

47. Kiresuk R, Sherman RE: Goal attainment scaling: A generalmethod for evaluating comprehensive community mental healthprograms. Community Mental Health J 44:443-453, 1986

48. van Thiel J, Kraan HF, van der Vleuten CP: Reliability andfeasibility in measuring medical interviewing skills: The revisedMaastricht History-Taking and Advice Checklist (MAAS-R).Med Educ 25:224-229, 1991

49. Cherkin DC, Hart LG, Rosenblatt RA: Patient satisfaction withfamily physicians and general internists: Is there a difference? JFam Pract 26:543-551, 1988

50. Dean A, Dean JA, Burton AH, et al: EPI-INFO, Version 5: A wordprocessing, database, and statistics program for epidemiology onmicro-computers. USD, Inc., Stone Mountain, GA, 1990

51. Grant JM: The fetal heart rate is normal, isn't it? Observeragreement of categorical assessments. Lancet 337:215-218, 1991

52. SPSS for Windows, SPSS Inc., Chicago, IL, 1993

53. Roter DL, Cole KA, Kern DE, et al: An evaluation of residencytraining in interviewing skills and the psychosocial domain ofmedical practice. J Gen Intern Med 5:347-354, 1990

54. Roter DL, Hall JA: Physicians' interviewing styles and medicalinformation obtained from patients. J Gen Intern Med 2:325-329, 1987

55. Smith RC, Marshall AA, Cohen-Cole SA: The efficacy ofintensive biopsychosocial teaching programs for residents: Areview of the literature and guidelines for teaching. J Gen InternMed 9:390-396, 1994

56. McGuire P, Fairbairn S, Fletcher CH: Consultation skills ofyoung doctors: I. Benefits of feedback training in interviewing asstudents persist. Br Med J 292:1573-1576, 1986

57. McGuire P, Roe P, Goldberg D, et al: The value of feedback inteaching interviewing skills to medical students. Psychol Med8:695-704, 1978

276 Psychosomatic Medicine 57:268-276 (1995)