Haidet, Paul MD, MPH; Kelly, P Adam PhD, MBA; Chou, Calvin MD, PhD; The Communication, Curriculum, and Culture Study Group
Patient-centered care “acknowledges the patient as a person with unique needs and life history” and is essential for effective clinical practice.1 In the context of the patient–physician relationship, patient-centered care requires that both parties explicitly define their own unique views of illness and align them toward a common therapeutic goal.2 A growing body of empirical evidence demonstrates that patient-centered care is associated with a number of favorable biomedical, psychological, and social outcomes.3,4 In response, medical educators and accrediting bodies have recommended that patient-centered care form a central paradigm for teaching the tasks of clinical practice. Many medical schools have instituted broad arrays of educational approaches intended to foster patient-centered care by graduates.5–9
Despite attempts to incorporate principles of patient-centered care into formal medical curricula, there is evidence that such attempts are often undermined by latent social processes and messages that marginalize the learning and practice of patient-centered care.10–12 As defined by Hafferty, this “hidden curriculum” comprises “the commonly held understandings, customs, rituals, and taken-for-granted aspects of what goes on in the life-space we call medical education … training institutions are both cultural entities and moral communities intimately involved in constructing definitions about what is “good” and “bad” medicine.”13 For example, while students may be taught the importance of asking open-ended questions and listening to patients’ stories as part of their history and physical diagnosis courses, they often encounter a different pattern of behavior, characterized by closed-ended questions and frequent interruptions of patients, when they enter preceptorships or clinical clerkships.2,14–15 Contradictions such as this one create internal conflicts in students that diminish the credibility of medical teachers.16 Such contradictions also initiate students to an acculturation process that erodes altruism, undermines ethics, and creates cynicism. In fact, senior medical students typically display less patient-centered attitudes than do students in earlier years of school,17 likely due to the influence of the hidden curriculum.
As educators strive to design formal curricula that are maximally effective at fostering patient-centered care, they also need to address those elements of the hidden curriculum that have an impact on how students perceive and value patient-centered care.18 However, gaining an understanding of these cultural factors at an individual school has typically required data-gathering approaches that are time- and labor-intensive. A simple method of determining the degree to which a medical school's culture fosters patient-centered care would be valuable to educators as they try to understand and potentially modify the extracurricular issues that exist around their formal coursework. Our group, the C3 (Communication, Curriculum, and Culture) Study Group, formed in 2000 to develop and test a tool to measure hidden curricula with respect to patient-centered care. In this report, we describe the development and initial validation of the C3 Instrument, a survey tool that measures the patient-centeredness of a medical school's hidden curriculum.
We developed the C3 Instrument through a multistage process that included steps to produce items, refine items, identify study institutions and collect data, and psychometrically evaluate and select items for the final instrument. We describe these steps in detail below.
We began with a series of five iterations of literature review and discussion among a “background working group” consisting of an academic physician (PH), a sociologist (DAP), a general internal medicine fellow (DH), a medical education specialist (BFR), and a statistician (JS). This group performed a review of the medical, sociological, and educational literature with respect to the culture of medical schools and patient-centered care. Relevant literature sources were identified through a combination of database (Medline, ERIC, Current Contents, PsychInfo) searches, reviews of reference lists of pertinent articles and book chapters, and discussions with published authors in the field of patient-centered care. The reference list of this report is composed of representative references from our review. As a result of this literature review, the background working group formulated a plan that included writing items in five different areas of content (role modeling, students’ experiences, characteristics of courses, support for students’ own patient-centered behaviors, and students’ own perceptions of the patient-centeredness of their school). These areas of content were selected because of their potential to capture a wide range of student perceptions and experiences in the informal environment of medical school. To ensure that items consistently measured specific aspects of patient-centered care, we selected a definition of patient-centered care proposed by Street:
Physicians who practice patient-centered care:
* Communicate information that is understandable and appropriate to the patient's needs and desires
* Communicate care, concern, and interest in the patient as a person
* Encourage and legitimize the patient's participation in his/her care19
We chose this definition because of its concentration on specific observable behaviors. Since our goal was to develop an instrument that would be completed by students at a variety of medical schools, grounding items in specific observable behaviors would help to minimize the effects of respondents’ own attitudes toward patient-centered care.
After deciding on the general content and structure of items and on a definition of patient-centered care, we formed an “item-writing group” consisting of an academic physician (PH), a general internal medicine fellow (DH), and a second-year medical student (VB). This group wrote a total of 160 items through three iterations of individual item writing, group review, and group revision.
As a first refining step, we assembled a panel of thirteen experts in patient-centered care. This panel included practicing physicians, medical educators, researchers, administrators, and representatives of national certifying boards (e.g., the American Board of Internal Medicine), none of whom were previously involved in the project, but who had published in the medical literature on topics pertaining to patient-centered care. We used an Internet-based, modified Delphi technique to direct individual panel members’ review of the item pool. The Delphi technique is a method for forming group consensus and defining levels of agreement about issues of uncertainty among groups of individuals who are separated by time and space.20 During each of three rounds, we asked individual panel members to look at each item and score its contribution to the overall instrument on a scale of 1–10; a score of “1” indicated a weak contribution, and a score of “10” indicated a strong contribution to the overall instrument. For each item during the second and third rounds, we provided the panel members with their own scores from the previous round, the panel's median score from the previous round, and the panel's 25th to 75th percentiles for each item. Individual panel members were free to keep their original score or change their score on the second and third rounds; we asked panel members who scored items outside of the 25th to 75th percentile range to provide reasons for their score. We eliminated all items with a median score of “5” or below after the panel's third round of review. In addition to the modified Delphi study, we asked panel members to give written feedback on the wording of items that they found problematic.
In our second refinement step, we recruited 15 fourth-year students from the class of 2001 at Baylor College of Medicine to participate in a focused working group to identify and rewrite ambiguous or poorly worded items. At the beginning of the session, each student completed the entire item pool and identified all items that they found problematic in terms of meaning or wording. The students then participated in a facilitated discussion to make clear the intended meaning of problematic items and suggest wording changes to enhance understandability and relevance. Wording changes were made for approximately 10% of items during this discussion. The student working group discussion was co-facilitated by a psychometrician (ALS) and the clinician member of the item-writing team. The session lasted approximately two hours and was audiotaped for further review by the item-writing team. After the two phases of refinement (expert review and student working group), the final item pool contained 143 items. No new items were generated by either phase of refinement.
Selecting institutions; collecting data
To test the validity of our instrument, we aimed to collect data at both “patient-centered” and “non-patient-centered” institutions. Since no such information to guide our choice of institutions existed, we used reputation as a proxy. We conducted an informal poll of the membership of the American Academy on Physician and Patient (AAPP) and the faculty of the Bayer Institute for Health Care Communication (BIHCC). We selected these two organizations because of their focus on concepts related to patient-centered care, their experience in teaching principles of patient-centered care, and their diversity of membership (e.g., multiple medical specialties are represented in both organizations). In our poll, we provided Street's definition of patient-centered care and asked respondents to answer the following question: “In no particular order, please name three medical schools in North America that, in your opinion, are most likely to produce graduates that are patient-centered physicians.” Fifty medical schools in North America were named at least once in this poll.
We selected ten medical and osteopathic schools based on our AAPP/BIHCC poll results and several additional criteria that included: (1) a diversity of geographic regions represented in the overall sample, (2) a mixture of private and public institutions, (3) diversity in class size, and (4) willingness to participate in the project. We obtained institutional review board approval at each of the ten study institutions. Using class listserves, we e-mailed third- and fourth-year students in the classes of 2002 and 2003 and invited them to connect (via a hotlink in the body of the e-mail) and complete an Internet-based survey. The survey contained our item pool and additional demographic items about age, gender, undergraduate degree, ethnicity, other advanced degrees (e.g., PhD, masters’ degrees), parents’ education, extracurricular activities, and medical specialty of interest. The survey also included an eight-item version of the Patient Practitioner Orientation Scale (PPOS), a previously validated instrument that measures an individual's attitudes toward patient-centered care.21 We contacted the officers of the medical student classes to help build interest in the survey, and we offered a monetary gift to the class funds for participation in the survey. Survey data were collected between March 1, 2002 and May 30, 2003. We calculated a participation rate based on institutional rosters of the numbers of students in each medical school class.
Factor structure and selection of items
Since items in the five different content areas were written with different formats and different response options, we conducted five separate factor analyses for items in each content area. For these factor analyses, we performed principal components analyses with promax (oblique) rotation. We first examined uncorrelated factors, but found that a correlated factor structure better modeled the data. This finding corroborated our a priori assumption that the factors in each defined content area would be interrelated.
After 682 students at seven schools had completed data collection, we performed exploratory factor analyses to identify dimensions in each content area. We defined a dimension as any factor: (1) with an eigenvalue greater than 1.0, and (2) that had at least two items with loadings of greater than .50 on the factor. Two content areas (“students’ own perceptions of the patient-centeredness of their school,” and “characteristics of courses”) did not reveal the emergence of any dimensions, and were therefore dropped from inclusion in the final instrument.
Next, we selected items with loadings greater than .50 on their respective dimensions in the three remaining content areas. These items comprise the final instrument that appears in the Appendix. For each item on the final instrument, we computed item response frequencies. We examined the response frequencies to look for unusual patterns indicative of poorly worded or misunderstood items, and found none.
After data collection was completed by 890 students at all ten schools, we confirmed the dimensions of the final instrument by performing follow-up factor analyses on the final items chosen. These follow-up analyses revealed the presence of dimensions similar to those seen in the exploratory factor analysis; we report psychometric statistics for these dimensions using data from the follow-up factor analyses on the entire dataset.
Reliability and validity
We examined the internal consistency of each dimension in the three content areas using Cronbach's alpha statistic. To examine validity, we compared scores on the content areas of our instrument to school reputation reflected in our AAPP/BIHCC poll. We selected the three schools with the highest number of votes (HI schools) and three schools with the lowest number of votes (LO schools) on the poll for this analysis. We used multivariable analysis of variance to examine associations between HI/LO status and mean scores for each of the C3 Instrument content areas. We controlled for student gender, age, attitudes toward patient-centered care as measured by the PPOS, and number of days since the beginning of the third year of medical school. We hypothesized that HI schools would have higher mean scores in each of the C3 Instrument content areas than LO schools.
Description of the C3 Instrument
The final version of the C3 Instrument appears in the Appendix. Since much of the content of the instrument is based on patient-care experiences, we suggest that it be used among students who have finished preclerkship classwork and who have completed at least some core clerkships. Three content areas constitute the C3 Instrument: “students’ experiences’ contains three dimensions, and “role modeling” and “support for students’ actions” each contain one dimension. For ease of interpretation, we advocate summing the item scores in each content area to calculate three overall “content scores” for the instrument. Since no empirical data exist about the relative weights of the three content areas to the overall “patient-centeredness” of the hidden curriculum, content area scores should be viewed as three independent scores, rather than being combined into a single score.
Each of the content areas on the instrument has its own set of instructions for respondents and uses its own unique response scale. Based on the amount of time that the focused working group subjects needed to complete the entire item pool, we estimate that completion of the 29 items on the final instrument should take an average respondent 10–15 minutes.
A total of 890 students completed the Internet-based survey, representing an overall participation rate of 45%. Characteristics of these students are shown in Table 1. The PPOS scores reflect individual students’ attitudes toward patient-centered care; the mean score of 4.8 for the cohort is similar to published mean scores for students in the clinical years of medical school.17,22 Most students reported their race/ethnicity to be white; other students identified themselves as African American (4%), Asian (20%), and Hispanic (4%). Only 28% of the sample indicated an interest in pursuing a primary care specialty, which we defined as “general internal medicine,” “general pediatrics,” or “family and/or community medicine.” The “days into clinical experience” variable reflects the number of days students had experienced in their clerkship years at the time that they completed the survey.
The ten schools participating in this study represent the Northeastern, Mid-Atlantic, Southern, Midwestern, and Western geographic regions of the United States. Six of the schools are publicly funded institutions. Class size for the institutions ranges from less than 100 students to over 170 students. Schools participated in the study with our promise that their names would remain undisclosed.
Reliability data for the C3 instrument are shown in Table 2. All of the dimensions that emerged from the factor analysis had very good to excellent internal consistency, as measured by Cronbach's alpha statistics ranging from .67 to .93. The central theme of each dimension as reflected in its title was based on a posthoc review of the meaning of the items that loaded on the dimension.
Validity data are shown in Table 3. For this analysis, we compared content-area scores for the three schools that were mentioned most often as having a reputation for graduating patient-centered students (HI schools) with content-area scores for the 3 schools that were mentioned the least (LO schools). As is shown in Table 3, schools selected as having a reputation for patient-centeredness had significantly greater patient-centered scores than schools not mentioned; this finding was consistent for each of the three content areas of the C3 instrument. None of the additional independent variables (student gender, age, attitudes toward patient-centered care, and days into clinical experience) correlated with C3 instrument scores on individual univariate analyses.
The C3 Instrument is an easily administered tool that measures three important aspects of the experiences of medical students. The role modeling that occurs on clerkships, the degree of support that students perceive for their own patient-centered behaviors, and the experiences that students have in the medical care arena all may send strong messages to students about the need (or lack thereof) for patient-centered behaviors with patients.
Since the hidden curriculum represents “common-sense” notions about medical practice, students may slowly change their attitudes and behaviors without even realizing that they are doing so. Therefore, the hidden curriculum often remains impervious to educational interventions that involve simply more time, more courses, and more subject areas.23 Educators need a way to identify and directly address specific aspects of the hidden curriculum in order to modify its overall impact on the attitudes and behaviors of students. The C3 Instrument is a potentially useful tool in this regard. It is designed to be easily administered and to give educators information that can guide their choices on what areas to focus potential interventions. For example, a school that notes low (i.e., not patient-centered) scores in the “role modeling” content area might be served best by one educational approach (e.g., teacher development), whereas a school that scores low in the “support for my actions” content area might be best served by a different approach (e.g., revision of grading and feedback criteria). In this way, the C3 Instrument can provide important and specific information that can help educators to modify curricular objectives or approaches as they address those particular aspects of the hidden curriculum that exist at their own institutions.
Previous approaches toward understanding the hidden curriculum have generally involved in-depth qualitative inquiry at a single school.10,12,13 These important studies have led to identification of the existence of the hidden curriculum, as well as understanding of its general domains. In contrast, the C3 Instrument focuses on three particular areas of education that impact one domain (i.e., patient-centered care). In addition, the C3 Instrument's brevity and ease of administration allow for widespread study of hidden curricula across multiple schools. Given core competencies related to patient-centered care that are currently espoused by multiple accrediting bodies, we assert that further systematic study of the dynamics between formal teaching and informal school culture is needed in order to widely improve educational effectiveness in this area. We also suggest that our design that used an Internet-based survey is feasible for studying impacts of the hidden curriculum on other content areas as well (e.g., evidence-based medicine, personal–professional balance, professionalism issues).
Our study has several limitations. First, since no “gold standard” for the patient-centeredness of a school's hidden curriculum exists, our validation of how closely the C3 Instrument's findings mirror our findings regarding U.S. schools’ reputations for patient-centeredness should be viewed with caution. Even though members of the AAPP and BIHCC are generally regarded as highly interested in patient-centered care, many individual poll respondents noted that they had in-depth exposure to a limited number of schools and were therefore able to comment on only those few schools. More work will be needed to further establish the validity of the C3 Instrument. Second, and perhaps more important, the relative impact of the three C3 Instrument content areas is unknown. One content area may have greater influence on student attitudes and behaviors than the others. Also, since no outcomes data currently exist, it is difficult to interpret the clinical significance of various effect sizes as measured by the C3 Instrument. Further work is needed to examine how each of the C3 Instrument content areas affects patient-centered educational outcomes.
In conclusion, we have designed a reliable and valid tool to measure salient aspects of a medical school's hidden curriculum with respect to patient-centered care. We anticipate that medical educators could use the C3 Instrument to guide educational interventions by addressing the context that exists around formal teaching activities. The C3 Instrument also makes possible the widespread study of hidden curricula across multiple medical schools. Further research on the hidden curriculum should be aimed at developing a greater understanding of the dynamics between teaching activities and school culture.
The three lead authors were investigators in the Communication, Curriculum, and Culture Study Group, all of whom are listed below in alpha order. The information about these investigators was current when this article was submitted for publication. Vinay Babu, MD, resident physician, The Albert Einstein Medical Center, Philadelphia, Pennsylvania; Benjamin Blatt, MD, director, CLASS Clinical Education Center, and associate professor of medicine, George Washington University School of Medicine and Health Sciences, Washington, DC; Susan Bentley, DO, fellow and acting instructor, Department of Psychiatry and Behavioral Sciences, The University of Washington School of Medicine, Seattle, Washington; Calvin Chou, MD, PhD, assistant clinical professor of medicine, San Francisco Veterans Affairs Medical Center and the University of California, San Francisco, School of Medicine, San Francisco, California; Auguste H. Fortin VI, MD, MPH, assistant clinical professor of medicine, Yale University School of Medicine, New Haven, Connecticut, Geoffrey Gordon, MD, associate professor of medicine and psychiatry and director, John Benson Program on Professionalism, Oregon Health and Science University, Portland, Oregon; Catherine Gracey, MD, assistant professor of medicine, University of Rochester School of Medicine and Dentistry, Rochester, New York; Paul Haidet, MD MPH, assistant professor of medicine, The Michael E DeBakey Veterans Affairs Medical Center and Baylor College of Medicine, Houston, Texas; Heather Harrell, MD, assistant professor of medicine, University of Florida College of Medicine, Gainesville, Florida; David Hatem, MD, associate professor of medicine, University of Massachusetts Medical School, Worcester, Massachusetts; Drew A. Helmer, MD, MS, assistant professor of medicine, The Veterans Affairs–New Jersey Health Care System, and UMDNJ–New Jersey Medical School, East Orange, New Jersey; Thomas Inui, ScM, MD, president and CEO, Regenstrief Institute, the Sam Regenstrief Professor and Associate Dean for Health Care Research, Indiana University School of Medicine and the Roudebush Veterans Affairs Medical Center, Indianapolis, Indiana; P. Adam Kelly, PhD MBA, assistant professor of medicine, The Michael E DeBakey Veterans Affairs Medical Center and Baylor College of Medicine, Houston, Texas; Mark Kuebeler, MS, health services research analyst, The Houston Center for Quality of Care and Utilization Studies and Baylor College of Medicine, Houston, Texas; Gregory Makoul, PhD, associate professor of medicine and director, Program in Communication and Medicine, Northwestern University Feinberg School of Medicine, Chicago, Illinois; Debora A. Paterniti, PhD, Associate Director, Center for Health Services Research in Primary Care, University of California, Davis Medical Center, Sacramento, CA, Linda Perkowski, PhD, associate dean for educational and curricular development, University of Minnesota Medical School, Minneapolis, Minnesota; Boyd F. Richards, PhD, director, Office of Curriculum, Baylor College of Medicine, Houston, Texas; A. Lynn Snow, PhD, assistant professor of medicine and psychiatry, The Michael E DeBakey Veterans Affairs Medical Center and Baylor College of Medicine, Houston, Texas; Julianne Souchek, PhD, senior research scientist and associate professor of medicine, The Michael E DeBakey Veterans Affairs Medical Center and Baylor College of Medicine, Houston, Texas; and Dianne Wagner, MD, associate professor of medicine, Michigan State University College of Human Medicine, East Lansing, Michigan.
The authors would like to acknowledge the contributions of the members of the expert review panel, the medical students who participated in the focused working group, and the numerous individuals who promoted the study at participating institutions. The authors would also like to acknowledge Nelda P. Wray, MD, MPH, for guidance and input during the planning stages of this project.
This work was supported by grants from the Bayer Institute for Health Care Communication and the Baylor College of Medicine Academy of Distinguished Educators Fullbright and Jaworski Fund for Medical Education Research. Drs. Haidet, Kelly, and Helmer are supported by career development awards from the Office of Research and Development, Health Services R&D Service, U.S. Department of Veterans Affairs. A mentoring relationship between Drs. Inui (mentor) and Haidet (mentee) is supported by a research mentorship award from the Society of General Internal Medicine.