International Journal for Quality in Health Care 2002; Volume 14, Number 5: 359–367

Impact of supervision and self-assessment

on doctor–patient communication in rural
Mexico
YOUNG-MI KIM1, MARIA ELENA FIGUEROA1, ANTONIETA MARTIN2, RICARDO SILVA3,
SIXTO F. ACOSTA3, MANUEL HURTADO4, PAUL RICHARDSON5 AND ADRIENNE KOLS1
1
Center for Communication Programs, Johns Hopkins University, School of Public Health, Baltimore, 5Quality Assurance Project,
Center for Human Services, Bethesda, MD, USA, 2Fronteras, The Population Council, Regional Office, Mexico City, 3Instituto
Mexicano del Seguro Social, Programa Solidaridad (IMSS/S), Mexico City, 4Universidad Veracruzana, Veracruz, Mexico

Abstract
Objective. To determine whether supervision and self-assessment activities can improve doctor–patient communication.
Setting and participants. Six supervisors, 60 doctors in their last year of training, and 232 primary health care patients at
rural health clinics in Michoacan, Mexico.
Design. The main evaluation compared post-intervention measures in control and intervention groups. A small panel study
also examined changes from baseline to post-intervention rounds in both groups.
Intervention. Over a 4-month period, specially trained supervisors added 1 hour of supervision on interpersonal communication
and counseling (IPC/C) to regular site visits. Doctors, who had received prior IPC/C training, periodically audiotaped and
assessed their own consultations.
Main outcome measures. These comprised frequency of doctors’ facilitative communication, doctors’ biomedical information-
giving, and patients’ active communication.
Results. The performance of all doctors improved markedly over the study period, but gains in facilitative communication
and information-giving were significantly greater in the intervention than the control group. No single component of the
intervention was responsible for the improvement; it resulted from the combination of activities. The doctors appreciated
the more supportive relationship with supervisors that resulted from the intervention and found listening to themselves on
audiotape a powerful, although initially stressful, experience.
Conclusion. Supportive supervision and self-assessment activities can reinforce IPC/C training, prompt reflection and
learning, and help novice doctors improve their interpersonal communication skills.
Keywords: communication, quality of care, physician–patient relations, self-assessment, supervision

Research shows that the quality of communication between cultural differences between indigenous communities and
doctors and their patients contributes to health outcomes as doctors. To provide health care services to rural populations,
well as patient satisfaction [1–5]. Doctors make more accurate the Mexican Institute of Social Security/Solidarity (IMSS/S)
diagnoses and more effective treatment plans when patients places resident doctors in rural clinics for a 9-month rotation
fully disclose their symptoms, concerns, and personal cir- as part of their training. Typically, one of these resident
cumstances. Patients feel more committed and better prepared doctors and a nurse staffs a two-room clinic. Most resident
to carry out a plan of action when doctors clearly explain doctors come from urban backgrounds, are middle to upper
the diagnosis, treatment options, and instructions. class, and speak Spanish. In contrast, the patients they serve
Good communication and counseling skills are especially come from a lower socioeconomic class and mostly speak
important in rural areas of Mexico, where there are wide indigenous languages. While most resident doctors establish

Address reprint requests to Y.-M. Kim, Senior Research and Evaluation Advisor, Center for Communication Programs,
Johns Hopkins University School of Public Health, 111 Market Place, Suite 310, Baltimore, MD 21202-4012, USA.
E-mail: ykim@jhuccp.org

 2002 International Society for Quality in Health Care and Oxford University Press 359
Y.-M. Kim et al.

a good rapport with patients and take time to ask questions assignment at the rural clinics), a second round of data was
and explain matters, formative research shows that they are collected.
less skilled in listening to clients, encouraging them to speak, The data are analyzed in two different ways: a cross-
and responding to individual client needs. sectional comparison and a panel study. The cross-sectional
IMSS/S has introduced training in interpersonal com- analysis compares post-intervention measures in the inter-
munication and counseling (IPC/C) to narrow the com- vention and control groups, and has the advantage of a larger
munication gap between young resident doctors and rural sample size. The panel study examines changes over time
patients. While experience elsewhere has demonstrated the from the baseline to post-intervention rounds in both the
effectiveness of IPC/C training [6,7], one-time training has intervention and control groups. It provides a more con-
not been sufficient to guarantee that health personnel apply servative measure of the intervention’s impact, since it takes
new communication skills on the job and maintain them into account changes in the control group during the inter-
over time [8]. Two opportunities exist for cost-effective vention period. However, the power of the panel study is
reinforcement of IPC/C skills among resident doctors at limited by its small sample size.
IMSS/S clinics. The first possibility is using the routine
supervision system already in place. Competent and ex- Study sample
perienced physician supervisors make regular 1-day site visits
to IMSS/S clinics to monitor technical standards of care. The study took place in the Zamora region of Michoacan,
With training and appropriate tools, they also could assess which is divided into seven supervision zones, each overseen
IPC/C performance and provide direct feedback to resident by a single supervisor. One zone was excluded from the
doctors. The second possibility is asking resident doctors study because the high proportion of indigenous peoples
to engage in self-assessment and self-directed learning, an made it atypical. The remaining six zones were randomly
approach that has maintained and improved health providers’ distributed into control (two zones) and experimental (four
communication skills in Indonesia, even in the absence of zones) conditions. This analysis uses data from a larger study
outside supervision and support [8]. conducted by IMSS/S, which included all 115 rural clinics in
In 1998–99, IMSS/S pilot tested both of these approaches the six zones, eliminating the need for random sampling. A
at rural clinics in the state of Michoacan. This study examines team of two research assistants visited each clinic for a day,
the impact of a combined intervention of supervision and self- and audiotaped and interviewed the first three patients to
assessment on the communication performance of resident come for services. These patients represented a small pro-
doctors. Specific objectives are: (1) to determine if supervision portion of the >15–30 patients who might be expected to
and self-assessment help doctors to apply newly learned visit a rural clinic in the course of a day. The larger study
communication skills on the job and to improve those skills involved 631 patients, 82 resident doctors, 33 general prac-
over time; and (2) to identify which activities (including titioners, and 115 nurses.
supervision visits, audiotaped consultations, self-assessment, The present study includes a subset of patients who were
homework logs, and job aids) are effective and acceptable to attended by resident doctors and for whom complete data
doctors. exists, including audiotapes, observations, and interviews.
Technical difficulties, including dead batteries, poor volume
control, and excessive background noise, rendered many
audiotapes unusable. In addition, some of the resident doctors
Methods had already left the rural clinics when the research assistants
arrived to collect the post-intervention data. Post-intervention
This study assessed a cohort of resident doctors who began data for the cross-sectional comparison are available for a
their assignment at an IMSS/S clinic in Michoacan, Mexico total of 157 patients and 60 doctors from 60 clinics scattered
in the summer of 1998. Soon after they arrived, all of the across all six supervision zones. Of these, 95 patients and 36
doctors attended a 2-day workshop on IPC/C, followed by doctors were in the intervention group, while 62 patients and
a half-day refresher course 5 months later. Baseline data were 24 doctors were in the control group.
collected immediately after the refresher course. The doctors The panel study includes every doctor for whom there is
were assigned to intervention and control groups depending matching baseline and post-intervention data. Matching data
on which supervision zone their clinics belonged to; the are available for a subgroup of 28 doctors, who were recorded
supervision zones included in the study were randomly divided with a total of 147 patients. Of these, 21 doctors were in the
into control and experimental conditions as described below. intervention group, and they saw 57 patients in the baseline
During the following 4 months, doctors in the intervention round and 54 patients in the post-intervention round. The
group received visits from supervisors who were specially remaining seven doctors were in the control group, and they
trained in IPC/C and who evaluated doctors’ interactions saw 18 patients in the baseline round and 17 patients in the
with clients; some of these doctors also conducted IPC/C post-intervention round.
self-assessment exercises. Doctors in the control group also
received regular supervision visits, but their supervisors were
Data collection
not trained in IPC/C and did not review how well they
communicated with clients. At the end of the 4-month Audiotaped consultations, which were coded for content, are
intervention period (which also marked the end of the doctors’ the primary source of data for this study. Based on an

360
 Doctor–patient communication

interaction analysis of 15 consultations recorded earlier at at IMSS/S clinics. Participating supervisors attended a 3-day
the study site, researchers adapted the Roter Interaction training course that covered the importance of interpersonal
Analysis System (RIAS) to code the consultations [9]. RIAS communication, a five-step supervision model for evaluating
was designed to analyze doctor–patient interactions and has its quality, and key supervision skills. They were trained on
been extensively tested in medical settings in both developed how to conduct IPC/C supervision using a specially designed
and developing countries; studies have reported adequate assessment tool, and they focused on six skill areas deemed
inter-coder reliability [7,8,10,11]. The system assigns each essential to the quality of care: listening, being responsive to
utterance made by a doctor or patient to one of 48 mutually clients, expressing positive emotions, eliciting information,
exclusive coding categories (utterances consist of a phrase or giving information, and encouraging patient participation.
sentence that conveys a complete thought). Some examples The 4-month intervention has been called ‘partnership
of coding categories are: gives medical information, asks supervision’ because responsibility for enhancing com-
open-ended lifestyle question, shows concern or worry, or munication skills was shared by supervisors and doctors.
checks for understanding. Supervisors visited the doctors at 2-month intervals and
Two Mexican physicians, both of whom were familiar with engaged in a series of special IPC/C activities: they observed
the services of IMSS/S, performed the RIAS coding. One a consultation, used a checklist to assess the doctors’ com-
physician coded all of the baseline data and then trained and munication skills, gave feedback, discussed issues raised by
supervised a second physician to code the post-intervention the doctor, and helped doctors identify specific com-
data. As they listened to the audiotapes, the physicians used munication skills that needed work. The doctors recorded
a computerized data entry screen to assign codes to each these assignments in a homework log and reviewed their
utterance. The coders were blind to the intervention status progress with the supervisor during the next visit.
of the doctors. To test for inter-coder reliability, the first Between supervision visits, the doctors continued to work
physician also coded 22 consultations from the post-inter- on improving their communication skills, especially those
vention round. Agreement between the two coders exceeded listed in the homework log. Doctors were encouraged to
90%. The coders also calculated the length of each con- consider every encounter with a patient as an opportunity to
sultation, based on the counter numbers on the tape recorder. practice desired behaviors and to improve their com-
To ensure the consistency of these measurements, the same munication skills. To prompt self-assessment and self-learn-
brand and model of tape recorder was used to audiotape all ing, they were also given a more formal assignment in the
consultations. form of the following:
Data on the sociodemographic characteristics and work
experience of the supervisors, doctors, and patients were (1) Each doctor was supposed to audiotape two con-
collected in individual interviews. sultations a month, with the permission of the patients.
Qualitative data were collected at the end of the study to (2) The doctors listened to the tapes and assessed their
help explain the findings. Providers participated in focus communication performance with the help of a job
group discussions while supervisors were interviewed in- aid.
dividually. Facilitators and interviewers explored their re- (2) Some doctors also completed written self-assessment
actions to the intervention and their perceptions of its impact. forms focusing on specific communication skills. (Their
Researchers also used unstructured observations made during supervisors received additional training to support this
the implementation process to help explain the findings. activity.)

Supervision, self-assessment, and self-learning The job aid consisted of six color-coded sections, each
intervention covering one of the essential IPC/C skill areas listed above.
As described above, each doctor attended a 2-day workshop Each section explained the meaning and the importance of
and a half-day refresher course on IPC/C. The curriculum the skill, gave detailed examples of how to perform it with
was designed to help the doctors develop skills in counseling, warmth, and listed behaviors to be avoided.
verbal and non-verbal communication, interviewing, listening, In the control group, doctors also received IPC/C training,
and helping the client to make a decision. This curriculum but there was no follow up or reinforcement. Although
was institutionalized by IMSS/S in a previous project and supervisors made their usual 1-day visits to control clinics,
had become a standard part of training by the time this they were not trained in IPC/C supervision nor were they
study took place. Thus, all of the doctors—whether in the given the special assessment tool. Researchers asked the two
intervention or control groups—received the same IPC/C supervisors in the control condition to be on a waiting list
training. However, doctors in the intervention group were so as not to contaminate the experiment. Therefore, doctors
given instructions on the intervention itself during the re- in the control group did not receive IPC/C supervision, nor
fresher course. did they receive the job aid, a tape recorder, or any other
The supervision, self-assessment, and self-learning inter- intervention materials. They continued with their usual routine
vention was designed to reinforce this training, to help young of reviewing issues in the technical quality of care and in the
doctors apply communication skills on the job, and to improve adequacy of medical supplies during monthly supervision
those communication skills over the course of their residency visits.

361
Y.-M. Kim et al.

Outcome measures with the purpose of the visit. About half (48%) of the patients
came for general medical services, such as colds, stomach
The main outcome measure is doctor facilitative com-
pain, and diabetes; their average age was 51 years. One-
munication, i.e. communication that promotes an interactive
third (34%) came for reproductive health services, including
relationship between patient and doctor by fostering dialogue,
prenatal care, family planning, sexually transmitted infections
rapport, and patient participation. This concept has been
(STIs), and adolescent counseling; their average age was 22
developed by some of the authors over the course of previous
years. About one-fifth (18%), usually mothers, brought a
studies analyzing client–provider interaction in family plan-
child who was sick or needed immunization.
ning consultations in Kenya and Indonesia [8,12,13]. Fa-
The average age of the resident doctors was 25 years, and
cilitative communication is operationally defined as a set of
36% of them were male. All of the supervisors were male
RIAS coding categories that past research suggests is related
physicians, and their average age was 37 years. All worked
to clients playing an active role in the consultation. These
full-time as supervisors for IMSS/S, and they had an average
include partnership building, showing agreement or under-
of 7 years experience in the job.
standing, discussion of personal and social issues, expression
of positive emotions, and asking or giving information on Process evaluation
lifestyle and psychosocial issues. Four of the intervention’s six
IPC/C content areas were designed to encourage facilitative Supervision. Doctors in both the control and intervention
communication: active listening, being responsive to patients, groups received an average of 1.7 visits from supervisors
encouraging patient participation, and expressing positive during the 4-month study period, i.e. about one every 2
emotions. months. In the control group, none of these visits included
Information-giving by doctors is a second outcome meas- supervision on IPC/C. In the intervention group, all of the
ure. Earlier qualitative studies conducted in Michoacan found visits included >1 hour of supervision on IPC/C. During
that giving insufficient information was a common weakness most visits in the intervention group, supervisors and doctors
among resident doctors and that patients wanted better reviewed the homework log together (1.4 times).
explanations. One of the intervention’s IPC/C content areas In focus group discussions, doctors in the intervention
encouraged doctors to provide more and better medical and group reported that supervisors offered them more and better
technical information to patients. feedback on communication and counseling issues after the
In theory, facilitative communication by doctors should intervention began. Doctors also noted changes in super-
encourage patients to take a more active part in the con- visors’ interpersonal communication: supervisors began work-
sultation. Hence a third outcome measure is patient active ing with the doctors as partners, listening to their ideas, and
communication, which includes: asking questions, asking for engaging them in discussion, and were more appreciative of
clarification, expressing an opinion, expressing concerns, and their efforts. While doctors praised supervisors for being
discussing personal and social issues. kind, accessible, and not scolding, some wanted more time
with supervisors and more specific feedback from them.
Data analysis Self-assessment and self-learning. Doctors audiotaped an average
of 7.2 consultations, a little less than the eight tapes they
The analysis consistently examines the frequency of each were asked to make, and performed an average of 23.1 self-
outcome variable (i.e. the number of utterances per con- assessments, about four in each of the six IPC/C skill areas.
sultation) rather than its proportion. In the cross-sectional Thus, doctors listened to each tape several times, assessing
study, ANOVA was performed to test the significance of a different skill each time. Each self-assessment and self-
differences between the control and intervention groups. In learning session included listening to an audiotaped con-
the panel study, ANOVA was used to test the significance of sultation, and took 30–60 minutes. Nearly all doctors (97%)
changes over time (from the baseline to the post-intervention reported using the job aid regularly and found it useful.
rounds) within the intervention and control groups. The Wald Doctors reported using the homework log 8.6 times, on
test was used to test the significance of differences in the average, as part of their self-improvement efforts.
rate of change between the intervention and control groups. According to focus group discussions, doctors initially
Multiple regression analyses were conducted as part of the found the self-assessment process stressful, especially those
cross-sectional and panel studies to control for three potential who did not receive written self-assessment forms and in-
confounding factors: the purpose of the visit, the sex of the structions. The doctors worried about asking patients for
doctor, and the length of the session. permission to record the session, they were afraid of hearing
their own mistakes on tape, they were anxious about following
the steps laid out in the job aid, they felt nervous and self-
conscious while the taping was going on, and they were
Results
anxious about sharing the tapes with supervisors or nurses.
With repetition, however, doctors became proficient at self-
Characteristics of study participants
evaluation and found that listening to themselves on tape
Most patients were married (84%), women (80%), and had was a powerful and eye-opening experience. The tapes helped
a primary education or less (81%). The age of the patients, them recognize their strengths and weaknesses and provided
but not their marital status, sex or educational level, varied strong motivation to improve.

362
 Doctor–patient communication

a complete thought) per session was significantly greater in

the intervention than the control group (196 versus 128,
P<0.001) at the end of the study, and both providers and
clients contributed to the disparity. In other words, both
providers and clients in the intervention group uttered more
thoughts per minute than their peers in the control group.
According to the panel study, providers’ utterance rate in-
creased significantly over the study period in the intervention
group (from 6.9 to 9.3 utterances per minute, P<0.001) but
not in the control group (from 7.5 to 8.7, not significant).
The client utterance rate increased more among the inter-
vention (3.6 to 5.7, P<0.001) than the control group (4.0 to
5.0, P<0.05).
A qualitative review of the audiotapes identified three
Figure 1 Frequency of the doctors’ use of facilitative and
behavioral changes that led to increased utterance rates in
information-giving communication after the intervention,
the intervention group. Firstly, providers spent less time in
control versus intervention groups. Facilitative, com-
silence while writing notes on the patient’s chart. Secondly,
munication that promotes an interactive relationship between
providers lectured less. Thirdly, providers paused more fre-
patient and doctor.
quently to allow clients to speak.

Impact on length of sessions and utterance rate Impact on doctors’ communication

There was no significant difference in the length of the Facilitative communication. Doctors in the intervention group
consultation in the intervention and control groups (13.4 and outperformed the others during the post-intervention round,
11.8 minutes, respectively). The panel study found the average with an overall frequency of facilitative communication of
length of the consultation increased significantly over the 4- 48 compared with 30 for the control group (P<0.001) (Figure
month study period in both the intervention (from 7.0 to 1). Even after controlling for the purpose of the visit, the sex
13.3 minutes, P<0.01) and control groups (6.3 to 9.8 minutes, of the doctor, and the length of the session, the intervention
P<0.001). showed a significant impact on facilitative communication
These numbers mask a significant change in the amount (
=0.28, P<0.001). As Figure 2 shows, doctors in the
of conversation exchanged between providers and clients. intervention group performed significantly better than those
The number of utterances (phrases or sentences expressing in the control group on three of the six types of facilitative

Figure 2 Doctors’ frequency of use of six types of facilitative communication after the intervention, control versus
intervention groups. Partnership, builds a sense of partnership between doctor and patient; Acknowledge, communicates
understanding of what patient is saying; Pers/social, includes remarks on personal or social aspects; Positive emotion,
gives praise, reassurance; Info-psychosocial, provides counselling on psychosocial aspects; Ques-psychosocial, asks about
psychosocial aspects.

363
Y.-M. Kim et al.

Figure 3 Doctors’ facilitative communication: panel study. Figure 4 Doctors’ bio-medical information and counseling:
panel study.

communication: partnership building (12.7 versus 7.3, information and counseling than those in the control group
P<0.001), acknowledgement (12.3 versus 6.2, P<0.001), and (27.5 versus 16.6, P<0.001) (Figure 1), and this difference
expressing positive emotions (5.9 versus 2.9, P<0.001). remained significant even after controlling for other factors
The panel study confirms the intervention’s impact on (
=0.26, P<0.001). The panel study confirms this finding:
facilitative communication. While doctors’ communication information-giving increased from 7.8 to 25.1 (P<0.001) in
improved markedly over time in both groups, the gains were the intervention group, compared with a rise from 7.7 to
significantly greater in the intervention than the control group 16.6 (P<0.001) in the control group (Figure 4). After con-
(P=0.004). Levels of facilitative communication rose 238% trolling for other factors, these increases remained significant
in the intervention group (from 13.6 to 45.9, P<0.001) and both in the intervention (
=0.44, P<0.001) and control
124% in the control group (from 14.6 to 32.7, P<0.001) groups (
=0.42, P<0.05). However, the rate of change was
(Figure 3). After controlling for other factors in a multiple significantly greater in the intervention than control group
regression analysis, this rise was significant in the intervention (P=0.0001). RIAS coding does not permit us to measure
group (
=0.23, P<0.01) but not in the control group (
= the quality of information provided, such as its accuracy and
0.20, not significant). In anecdotal reports, doctors and relevance.
supervisors said the initial IPC/C training, daily practice with Multiple regression analyses found a somewhat different
patients, weekly outreach services in the community, and pattern of associations between individual intervention com-
supervision had helped doctors become better com- ponents and information-giving than was revealed for fa-
municators. Since the control group also attended IPC/C cilitative communication. After controlling for other factors,
training, received routine supervision, and learned from their just two components had a significant impact: the number
growing experience with patients, it is no wonder that their of times the homework log was used (
=0.18, P<0.01) and
levels of facilitative communication increased as well. the number of audiotapes made (
=0.17, P<0.01), while
A series of multiple regression analyses were conducted the number of supervision visits was of borderline significance
to determine which components of the intervention were (
=0.14, P=0.052). Once all of the intervention com-
most effective. These analyses controlled for: (1) the purpose ponents were entered in the regression, none of the individual
of the visit, which varied between the two data collection components remained significant.
rounds, and between control and intervention groups; (2) the Qualitative findings. In focus group discussions, doctors
sex of the doctor, which was associated with levels of reported that their new communication skills not only im-
facilitative communication; and (3) the length of the session, proved their interactions with patients but also carried over
which varied widely. When the impact of each component to their relationships with nurses, supervisors, community
on facilitative communication was assessed separately, a sig- members, friends, and family. Doctors also said they found
nificant positive association was found with the number of it more satisfying to view their patient in a larger context, as
supervision visits received (
=0.25, P<0.001), the number a person rather than as a diagnosis. Thus they felt the
of sessions audiotaped (
=0.20, P<0.01), the number of intervention had contributed to their personal and pro-
self-assessments performed (
=0.19, P<0.01), and the num- fessional lives, both for the present and in the future.
ber of times the homework log was used (
=0.13, P<0.05).
(It was impossible to assess the impact of the job aid, since
Impact on patients’ communication
all doctors reported using it frequently.) Only the number of
supervision visits remained significant, however, when all of The frequency of patient active communication did not differ
the intervention components were entered in the regression significantly between the intervention and control groups
(
=0.20, P<0.05). (13.3 compared with 11.4, respectively, not significant). The
Information-giving. Following the intervention, doctors in panel study showed that the frequency of patient active
the intervention group provided 63% more biomedical communication increased dramatically over the study period

364
 Doctor–patient communication

in both the intervention (from 2.4 to 12.7,
=0.07, P<0.001) scope of the analysis also was limited by technical difficulties
and control groups (from 2.6 to 13.0,
=0.13, P<0.01), with the audio recording and the departure of some doctors
with no significant difference in the rate of change between prior to the post-intervention round of data collection. About
the two groups. This general increase in active communication one-quarter (27%) of the resident doctors who participated
may be due to providers’ growing experience and the increased in the study were dropped entirely from the analysis, and
length of the sessions, rather than the indirect impact of the less than half (47%) of those remaining were included in the
intervention. These also may explain qualitative reports by panel study. Due to the lack of random sampling, the findings
doctors in the intervention group: in focus group discussions, must be interpreted with caution. Since the data lost, however,
they said patients noticed and responded to the changes in was due to recording problems and scheduling difficulties,
their interpersonal communication, appreciated the additional there is no reason to believe it systematically biased the
time spent on talking about their problems, opened up more, results.
and were more likely to make return visits. This intervention is rooted in new, supportive approaches
to supervision that have broadened the supervisor’s re-
sponsibilities in an effort to improve the quality of care [17,
Discussion 18]. According to a widely accepted model, clinical supervisors
have three primary functions: (1) normative, ensuring that staff
Supportive supervision and self-assessment changed pro- adhere to standards; (2) formative, facilitating learning and
viders’ communication patterns, increasing the amount of professional development by staff members; and (3) restorative,
facilitative communication, shortening their utterances, and providing emotional support to, and ensuring the personal
accelerating the exchange of conversations. These alterations well-being of, staff members [15,19].
suggest that doctors adopted a more client-centered, less The supervision intervention implemented in Mexico ac-
authoritarian approach to care along with a more participatory knowledged the continuing importance of supervisors’ norm-
style of communication—changes that researchers have found ative function in the creation of an observation checklist to
produce better health outcomes [2–5,14]. assess doctors’ IPC/C performance. However, the emphasis
In contrast, changes in patient behavior due to the inter- on feedback, two-way discussion, and the homework log
vention were neither observed nor expected, since the inter- added a formative, educational dimension that helped doctors
vention could have only an indirect impact upon them. improve their skills. Training in interpersonal communication
However, patient active communication in both the inter- also helped supervisors perform the restorative function,
vention and control groups increased over time, probably which takes on even more importance when young, in-
due to the growing familiarity between patients and doctors. experienced doctors are assigned to live and work in isolated
The resident doctors were strangers when they first arrived rural clinics where they have no peers or support network.
at the IMSS/S clinics. Over the course of their 9-month stint Research also points to the importance of reflection for
at the clinic, which included making home visits 1 day a professional decision making and adult learning [20]. Re-
week, the doctors gradually met the local people, gained an flective practice requires active observation of events and,
appreciation of the local culture, and came to know their later, reflection on them to understand better and learn from
patients. By the end of their stay, they had forged a personal experience. While supervisors can and do prompt reflection
relationship with many patients, making it easier for patients [19], this study demonstrates that listening to yourself on
to speak out. audiotape also stimulates reflection, self-assessment, and self-
Studying these young doctors offered both benefits and learning. For doctors, listening to the audiotapes was a
challenges. Because they had just finished training and had powerful experience, and self-criticism was a more compelling
not yet established patterns of communication with patients, motivator than outside criticism. While health care providers
these resident doctors may have been more open to the in Indonesia successfully performed IPC/C self-assessments
influence of the intervention than veteran health care pro- without using audiotapes, relying on memory alone was
viders. Indeed, two studies of nurses in the UK found that difficult, and providers were not as deeply moved by the
clinical supervision, including its educational component, had process [8].
a far greater impact on the least experienced and most junior Partnership supervision may not be suitable for all settings,
nurses [15,16]. However, it can be difficult to assess the however. Above all, it requires that a functioning supervision
impact of an intervention on doctors just entering practice system be in place. Because IMMS/S already had competent
because their skills rapidly improve with experience. The and experienced supervisors making regular visits to rural
panel study enabled us to distinguish between the impact of clinics, it was relatively easy to add IPC/C supervision to
the intervention and doctors’ naturally steep learning curve, their responsibilities. In many developing countries, however,
since doctors in the control group shared the same IPC/C supervisors are few in number, poorly trained, and lack
training, routine supervision, and patient experiences as the transportation to visit facilities [20–22]. Even in developed
intervention group. countries, the costs of time and training pose a barrier to
The study suffers from certain other limitations. Audio supervision of clinical personnel [19,23]. When the super-
taping, while less intrusive than having an observer present, vision system is not fully functioning, alternative approaches
inevitably affects the behavior both of the doctors, who may become more attractive; for example, self-assessment, re-
try harder, and the patients, who may feel inhibited. The flective diaries, and peer review [8,23]. Yet the Mexican

365
Y.-M. Kim et al.

experience points to practical limitations here as well. While References

audiotaping consultations proved to be an effective learning
tool, IMMS/S found it difficult to supply tape recorders to 1. Barry CA, Bradley CP, Britten N, Stevenson FA, Barber N.
scattered rural clinics and maintain them in working order Patients’ unvoiced agendas in general practice consultations:
once the intervention was scaled up. qualitative study. Br Med J 2000; 320: 1246–1250.
Because supervision and self-assessment activities and 2. Greenfield S, Kaplan S, Ware JE, Yano EM, Frank HJ. Patients’
materials were designed to complement and build on each participation in medical care: effects on blood sugar control
other, it is difficult to single out the effectiveness of any one and quality of life in diabetes. J Intern Med 1988; 3: 448–457.
component of the intervention. Results suggest instead the
importance of multiple, reinforcing interventions for pro- 3. Kaplan SH, Greenfield S, Ware JW. Assessing the effect of
the physician–patient interaction on the outcomes of chronic
moting self-learning and behavioral change. Doctors valued
disease. Med Care 1989; 27: S110–S127.
every element of the intervention, including the supervision
visits, homework log, job aid, audiotapes, and self-assessment. 4. Stewart M. Effective physician–patient communication and
Perhaps because the self-assessment process (including the health outcomes: a review. Can Med Assoc J 1995; 152: 1423–
audiotapes) occurred four times more often than supervision 1433.
visits and occupied so much more of their time, doctors 5. Roter D, Steward M, Putman SM, Lipkin M, Stiles W, Inui TS.
emphasized self-assessment during focus group discussions. The patient–physician relationship: communication patterns of
However, they also asked for more time with supervisors primary care physicians. J Am Med Assoc 1997; 277: 350–356.
and more feedback from them.
6. De Negri B, Brown L, Hernandez O et al. Improving interpersonal
communication between 7 health care providers and clients. Bethesda,
MD: Quality Assurance Project, 1997. Available online at
Conclusion www.qaproject.org under ‘Products’. Last accessed 29 April
2002.
This study demonstrates that a combination of supportive 7. DiPrete Brown LD, de Negri B, Hernandez O, Dominguez L,
supervision and self-assessment can reinforce IPC/C training, Sanchak JH, Roter D. An evaluation of the impact of training
help doctors apply newly learned skills on the job, and Honduran health care providers in interpersonal communica-
contribute to continuing improvement in doctor–patient com- tion. Int J Qual Health Care 2000; 12: 495–501.
munication. Because supervision is a standard part of most 8. Kim YM, Putjuk F, Basuki E, Kols A. Self-assessment and peer
health care systems, it offers a highly effective way of reaching review: improving Indonesian service providers’ communication
doctors, with training and reinforcement on interpersonal with clients. Int Fam Plann Perspect 2000; 26: 4–20.
communication. However, supervisors typically do not give
9. Roter D. The Roter Interaction Analysis System (RIAS) Coding Manual.
feedback on doctor–patient interaction. Specially designed
Baltimore, MD: Johns Hopkins University School of Hygiene
training and assessment tools can direct supervisors’ time and Public Health, 1997.
and energy to these important issues. Self-assessment extends
and magnifies the impact of supervision by sharing re- 10. van den Brink-Muinen A, Verhaak PFM, Bensing JM et al. The
sponsibility for performance improvement and enhancing the Euro-Communication Study: an International Comparative Study in Six
partnership between doctor and supervisor. Further research European Countries on Doctor–Patient Communication in General
Practice. Utrecht, The Netherlands: NIVEL, 1999. Available
is needed to test different forms of IPC/C supervision and
at http://www.nivel.nl/publicaties/W8.shtml. Last accessed 29
self-assessment, and to refine the balance between them. April 2002.
11. Roter D, Rosenbaum J, de Negri B, Renaud D, DiPrete Brown
L, Hernandez O. The effects of a continuing medical education
Acknowledgements programme in interpersonal communication skills on doctor
practice and patient satisfaction in Trinidad and Tobago. Med
The study was carried out by the Quality Assurance Project Educ 1998; 32: 181–189.
(QAP), Instituto Mexicano del Seguro Social/Solidaridad 12. Kim YM, Odallo D, Thuo M, Kols A. Client participation
(IMSS/S), and Johns Hopkins University Center for Com- and provider communication in family planning counseling:
munication Programs (JHU/CCP), a sub-contractor of QAP. transcript analysis in Kenya. Health Commun 1999; 11: 1–19.
QAP is managed by the Center for Human Services (CHS;
Bethesda, MD), and funded by the U.S. Agency for Inter- 13. Kim YM, Kols A, Bonnin C, Richardson P, Roter D. Client
communication behaviors with health care providers in In-
national Development (USAID) contract number HRN-C-
donesia. Patient Educ Couns 2001; 45: 59–68.
00-96-900013. The authors thank Javier Cabral, Celia Es-
candon, Jesus Castellanos, Maribel Rodriguez (IMSS/S); Phyl- 14. Ong LM, De Haes JCJM, Hoos AM, Lammes FB. Doctor–
lis Piotrow, Elizabeth Costenbader, Gary Lewis, (JHU/CCP); patient communication: a review of the literature. Soc Sci Med
Debra Roter, Susan Larson (Johns Hopkins School of Public 1995; 40: 903–918.
Health); Jim Heiby (USAID); and David Nicholas, Bart 15. Bowles N, Young C. An evaluative study of clinical supervision
Burkhalter, and Paula Tavrow (QAP/CHS) for their as- based on Proctor’s three function interactive model. J Adv Nurs
sistance. 1999; 30: 958–964.

366
 Doctor–patient communication

16. Teasdale K, Brocklehurst N, Thom N. Clinical supervision and 21. Combary P, Newman C, Glover K et al. Study of the Effects of
support for nurses: an evaluation study. J Adv Nurs 2001; 33: Technical Supervision Training on CBD Supervisors’ Performance in Seven
216–224. Regions of Ghana. Chapel Hill, NC: University of North Carolina
at Chapel Hill, School of Medicine, Program for International
17. Ben Salem B, Beattie KJ. Facilitative supervision: a vital link in Training in Health (INTRAH), 1999. Available at http://
quality reproductive health service delivery. AVSC Working Paper www.prime2.org/pdf/TR07.pdf. Last accessed 29 April 2002.
#10, 1996. Available at http://www.engenderhealth.org/pubs/
workpap/wp10/wp 10.html. Last accessed 29 April 2002. 22. Valadez J, Vargas W, Diprete L. Supervision of primary health
care in Costa Rica: time well spent? Health Policy Plann 1990; 5:
18. Lammerink M. Ways of working. Health Action 1994; 8: 10. 118–125.
19. Kilminster SM, Jolly BC. Effective supervision in clinical practice 23. Teasdale K. Practical approaches to clinical supervision. Prof
settings: a literature review. Med Educ 2000; 34: 827–840. Nurse 2000; 15: 579–582.
20. Ahmed AM, Gavyole A, Omar HM, Munisi W. The national
guidelines for supervision checklist: a tool for monitoring super-
vision activities at district level in Tanzania. Ann Ig 1994; 6:
161–166. Accepted for publication 12 June 2002

367