Background. Efforts to make health care for adolescents with chronic illnesses more patient-centered must be grounded in an understanding and clear measures of adolescents' preferences and priorities.
Objective. To develop a measure of health care preferences of adolescents with chronic illnesses and to determine demographic, developmental, and health factors associated with adolescents' preferences.
Design. Mixed-method questionnaire development and survey.
Setting. Subspecialty clinics of a tertiary care children's hospital.
Participants. All adolescents (age: 11–19 years) with juvenile rheumatoid arthritis, sickle cell disease, inflammatory bowel disease, or cystic fibrosis of at least 2-year duration who were being treated at the participating center were eligible to participate, and 155 of 251 did so (62%). The participants had a mean age of 15.5 ± 2.4 years, 45% were male, and 75% were white.
Main Outcome Measures. Ratings of 65 items related to quality of care and 17 items related to physician-patient communication styles.
Results. An 82-item questionnaire, devised from qualitative analysis of focus group results, contained 65 Likert scale items that adolescents considered important for health care quality and 17 forced-choice items related to adolescents' preferences for communication. Among the first 65 items, the group of questions related to physician trust and respect had the highest rating of 5.24 ± 0.62 of 6, followed by patient power and control (mean rating: 4.72 ± 0.77) and then caring and closeness in the patient-doctor relationship (mean rating: 4.19 ± 0.91). For the communication items, the adolescents, on average, preferred communication directly to them rather than to their parents and were nearly neutral regarding physicians' inquiries about personal issues.
Conclusions. Participants rated aspects of interpersonal care (especially honesty, attention to pain, and items related to respect) as most important in their judgments of quality. As in most previous studies of adults, technical aspects of care were also rated highly, suggesting that adolescents understand and value both scientific and interpersonal aspects of care.
In 2001, the Institute of Medicine published a blueprint for improving the quality of health care in the United States. The report, titled Crossing the Quality Chasm,1 identified patient-centeredness as a key strategy for addressing the overall experience of care. The characteristics of patient-centered care include provider responsiveness to patient needs, mutual respect of patient and provider, and an alliance between patient and provider that encourages the sharing of power and responsibility.2 Studies of patient-centered care have demonstrated its association with positive health outcomes and processes, such as control of hypertension, hyperglycemia, and pain; health status; quality of life; treatment adherence; and satisfaction with care.3,4 However, Safran5 found that “whole-person care” typically ranks lowest among ratings of primary care quality.
The delivery of patient-centered care depends on provider’s understanding of patient desires and expectations. In a systematic review of 57 studies, Wensing et al6 identified humaneness, competency/accuracy, and shared decision-making as the top 3 priorities of adult outpatients. Much less has been written about the priorities of children, adolescents, or the family members involved in their care.7–9 Although studies of healthy adolescents suggest greater concern with the characteristics of providers than practice settings,7,8,10,11 there is no evidence linking adolescent priorities or experiences to current health status or future health outcomes.
Why might the health care experiences of adolescents with chronic conditions be of particular interest or importance? Analysis of the 1994 National Health Interview Study determined that 12.6 million youths (18% of the population) have 1 or more chronic physical, behavioral, emotional, or developmental needs necessitating health services beyond those required by children generally.12 Of those, 6.5% have conditions that limit daily function.12 The 2001 National Survey of Children with Special Health Care Needs, which used a different method to identify its population, reported that 12% of American children have a special health care need.13 Although they make 4 times as many physician visits and are hospitalized 5 times as often as their healthy peers,14 very little is known about their health care preferences or expectations. Adolescents with chronic conditions, with their more intensive and often critical health care experiences, may offer a unique and valuable source of data with which to modify and improve the overall effectiveness of the health care delivery system for adolescents. Furthermore, the high cost of their care creates a social and financial imperative to deliver services that are effective, efficient, and congruent with the expectations of patients and families.
The purpose of this study was to use items directly derived from adolescents with chronic illnesses to create a measure of health care preferences that could guide and help monitor new approaches to adolescent health care. Because the adult literature and the small body of existing literature pertaining to adolescents suggest that health care preferences are shaped by multiple sociocultural and experiential factors, we also investigated how preferences vary with maturation, health status, and demographic factors. This information may help tailor future interventions for groups of adolescent patients with different needs and desires. Throughout the study, combined qualitative/quantitative methods were used to generate adolescent-centered data and to retain the adolescent perspective during the stages of analysis and interpretation.
Overview and Human Subject Protection
The study was conducted in 3 phases, with data from the first 2 being reported here. In phase I, a new health care preference measure for adolescents was developed. In phase II, the measure was administered to a cohort of 155 adolescents with 1 of 4 chronic diseases, and its preliminary factor structure and associations with demographic and developmental factors were determined. In phase III, longitudinal analysis will determine how preferences change with maturation and disease progression.
All phases of the study were approved by the participating institution's committee for the protection of human subjects. Subjects were recruited from hospital-based clinics, with their attending physicians' consent. Adolescents gave written assent or consent, as did guardians of those <18 years of age. Participants were assured of anonymity, and their clinicians were not involved in data collection or analysis. Adolescents received $10.00 for participating in any study visit (focus group or questionnaire administration).
Phase I: Questionnaire Development
In the first phase, 29 adolescents (age: 11–19 years) with sickle cell disease, cystic fibrosis, inflammatory bowel disease, or juvenile rheumatoid arthritis participated in 7 focus groups. For comparison, 25 healthy adolescents recruited from community centers participated in 5 focus groups. The group sessions lasted 60 to 90 minutes and were led by a professional facilitator. The facilitator was trained in marketing and worked full-time as a focus group facilitator. She had extensive experience facilitating adolescent group discussions of health and wellness topics, but she did not have any clinical training. Topics included health concerns, desired provider characteristics, and desired organization of care. For example, the opening question regarding desired provider characteristics was, “What makes a good doctor?” All group sessions were audio-taped and transcribed. A team of 4 researchers analyzed the transcripts with qualitative methods. They read each transcript, summarized it, and then met as a group to discuss the content and major themes. After each transcript had been analyzed, a series of individual and group analysis sessions were conducted to generate a comprehensive organized list, or typology, of the items described by the adolescents. The details of the focus groups are reported elsewhere.15
Initial Item Generation
A psychometrician (R.F.D.) and a physician (M.T.B.) wrote the items using the typology generated by the focus groups. All major elements related to desired elements of health care were included as items in the questionnaire. The item stems were constructed to represent an extreme view (eg, doctor always...), to enhance response variability. The items naturally split into 2 types. The first type represented items that all of the adolescents thought were desirable (eg, physician honesty and comfortable office furniture) but with differing opinions regarding the item's importance. The second type represented items for which the adolescents differed regarding desirability. These items were mostly related to communication, such as whether the adolescent's physician talked primarily to the teen or to her or his parent. A few additional items, such as hand-washing, were added on the basis of their importance in previous research.11
For the 65 items that were universally desirable, Likert scales were constructed with 6 response options ranging from “not at all important to my judgment of quality” to “extremely important to my judgment of quality.” This wording was chosen from multiple versions after trials with professionals and adolescents. For the 17 items of variable desirability, adolescents chose among 5 levels of agreement between 2 extremes (Fig 1). The wording and visual layout of these questions were also pretested with professionals and adolescents.
Versions of the questionnaire were pilot-tested sequentially with adolescents and professionals.
Phase II: Determination of Factor Structure and Cross-Sectional Analysis
Adolescents (age: 11–19 years) with sickle cell disease, cystic fibrosis, inflammatory bowel disease, or juvenile rheumatoid arthritis who had been diagnosed at least 2 years previously were eligible to participate. We chose these conditions because they are well defined, are relatively common, and share similar health care delivery models (usually cared for by multidisciplinary teams at tertiary care centers as well as by primary care providers). Subjects who were expected to leave the geographic area permanently during the 18-month study period or who were too cognitively impaired or too ill to participate were excluded. Two hundred fifty-one eligible patients could be located, and 155 (62%) participated. Adolescents with sickle cell disease were statistically more likely to participate, and those with cystic fibrosis were less likely to participate. Participants were younger than nonparticipants (15.4 vs 16.4 years, P = .01). Most subjects who declined cited lack of interest in this study or lack of time.
Health Care Preference Questionnaire
The health care preference questionnaire (82 items) developed in phase I was used.
Adolescent preferences and expectations regarding information and counseling from health care providers are correlated with pubertal maturation.16,17 Because data collection did not otherwise require physical examinations, the subjects self-assessed pubertal maturation with the pictorial format devised by Morris and Udry.18 This self-assessment technique has been shown to correlate well with physician ratings.18,19
The transition from concrete operational thinking to formal operational thinking generally occurs in early or middle adolescence. Adolescents who are concrete thinkers have difficulty with abstraction and are more focused on the present. Adolescents who have progressed to formal operational thinking are able to understand abstract concepts and to consider possibilities not previously experienced. These adolescents may have different preferences for interactions with their physicians and their health concerns may reflect greater future orientation, compared with adolescents who have not yet reached cognitive maturity. For example, mature adolescents may desire a more consultative, rather than paternalistic, relationship with their physicians. The How Is Your Logic? scale20 was chosen to measure qualitative cognitive development. This 13-item test of operational thought consists of 5 items measuring concrete thought and 8 items measuring levels of formal thought.
Health-Related Quality of Life
Adolescents with different degrees of physical and psychosocial functioning are likely to express different preferences and needs for health care. The Child Health Questionnaire, Youth Version, is an 87-item, health-related quality of life measure. It has been tested in school-based and clinic samples and has demonstrated adequate reliability and discriminant validity.21 Scale scores range from 0 to 100, with higher scores reflecting better function or greater well-being. We chose 5 scales, ie, global health, global behavior, mental health, general health perceptions, and changes in health.
Adolescents who have initiated behaviors such as sexual activity are more likely to report a desire to discuss these behaviors with health care providers.16,22 Selected questions from the Youth Risk Behavior Survey23 were used to assess lifetime and current sexual activity, alcohol, tobacco, and marijuana use, school performance, and seatbelt use. The Youth Risk Behavior Survey has been used widely and found to be reliable.24
Gender, age, race, and ethnicity were determined through youth self-report. Information on health insurance status, parental education, and household income was obtained from a parent or guardian at the time of enrollment (except for participants ≥18 years of age, who self-reported this information).
The 65 items assessing importance to quality and the 17 items related to communication were factor analyzed separately in principal-components analyses with oblique rotation. Items were grouped together in a factor if they loaded >0.4 on 1 factor and <0.3 on the other factors.
Descriptive statistics were generated for all variables. We first computed the mean Likert score for each of the 65 items, in rank order from largest to smallest. Multivariable linear regression analysis was used to identify variables independently associated with scores for the factors and for individual items in the preference measure that were considered a priori to be important but that did not load on any of the factors. Because the 3 factor scores based on the first 65 items (importance to quality) were correlated, as were the 2 factor scores based on the next 17 items (communication), we also performed separate multivariate linear regression analyses of the 2 sets of intercorrelated variables. Therefore, we used the factor scores derived from the first 65 items as 3 dependent variables in 1 analysis and the 2 factor scores derived from the last 17 items as dependent variables in a separate model. This analysis examined the effects of individual patient characteristics on the vector of factor scores taken as a group response.
The participant characteristics are shown in Table 1. Although we inquired about Hispanic ethnicity, 10 participants responded “don't know,” whereas only 5 responded “yes”; therefore, we did not include this variable in any analyses. Pubertal development was skewed toward maturity, whereas cognitive development was more varied, with only 24% of respondents reaching formal operations, or full cognitive maturity.
The items generated by the adolescents in the focus groups fell into 2 categories, ie, those almost universally desirable but of variable importance and those of variable desirability. The 65 items of universal desirability are listed in decreasing order of mean importance to quality in Table 2. A majority of the items described aspects of the adolescents' interpersonal relationships with their physicians. Because of variable SEs and multiple comparisons, no formal statistical comparison of these means was undertaken. With inspection, however, items related to physician honesty, attention to pain, and knowledge of the adolescent's conditions were judged as being most important to quality, with means of >5 and medians of 6. In contrast, items related to the physical environment of the office all had means of <3. Privacy and hand-washing ranked between these extremes. Table 3 lists the items of variable desirability in order of the strength of adolescents' preferences for 1 choice over the other. Responses could range from “prefer A a lot better” (coded −2) to “prefer B a lot better” (coded +2), with 0 representing “like A and B equally” (Fig 1). The most extreme responses were for adolescents preferring that things be explained to them directly, rather than to their parents (−1.49), and for adolescents always having a chance to speak privately with their physicians (1.24). Respondents were nearly neutral about being asked about personal issues (0.03) and about whether their physicians joked around rather than being serious during visits (0.08).
Principal-components factor analysis with oblique rotation identified 3 underlying domains among the 65 items related to quality and 2 domains among the remaining 17 items related to communication. Tables 4 and 5 present the items, means, and Cronbach's α values for these factors. All of the factors had adequate reliability except for directness of communication about sensitive topics (α = .54). Among the 3 universally desirable factors of quality of care, feeling respected by and trusting the physician had the highest mean value of 5.2 (of a possible score of 6). The items assessing adolescents' feeling of power and control in the relationship had a mean score of 4.7, whereas those assessing closeness and caring had a mean score of 4.2. All of these means were significantly different (P < .001). Of the 2 factors related to communication, the first factor concerned communication through parents versus directly with the adolescent. Higher scores reflected a preference for more direct communication with the adolescent. A score of 0 reflected no preference. The mean factor score was 0.76, demonstrating a modest group preference for communication directly with the adolescent. The final factor was related to directness of discussion of sensitive topics, although 1 item was related to control of decision-making. Those items had a mean score of 0.11, reflecting a nearly neutral overall response for this factor.
We then analyzed demographic, developmental, and functional health status characteristics that we had hypothesized, on the basis of previous literature findings, might be associated with care preferences. Separate models were built for each factor. The multivariable analyses are summarized in Table 6. For the respect and trust factor, parental education was the only significant predictor. Higher education was associated with lower means for this factor. White race was associated with a trend in the same direction, although it was not significant (P = .09). Female subjects viewed power and control as more important than did male subjects. For caring and closeness, increasing parental education and increasing cognitive development were associated with lower scores for the importance of this factor in the adolescent's judgment of quality. None of the other potential independent variables was significant in any model. Multivariate modeling (which considered all 3 factor scores together) yielded similar results, with parental education being negatively associated with scores for all 3 factors and white race being marginally associated with lower scores for caring/closeness and respect/trust.
A larger number of variables were independently associated with differences in preference for communication directly to the adolescent. Older age, better behavior scores, higher risk scores, and having commercial health insurance were all associated with higher mean scores on this scale (ie, preference for communication to teens). Higher mental health scores were associated with a preference in the opposite direction. For the final factor, directness of communication about personal issues, older age and positive changes in health were associated with a preference for more direct communication.
The adolescents in this study rated aspects of interpersonal care as most important to their judgment of quality. As in most previous studies of adults, technical aspects of care (ie, knows everything about taking care of someone with my condition and has a lot of experience taking care of people with my condition) were also ranked highly, which suggests that adolescents understand and value both the scientific and interpersonal components of care.
Physician honesty had the highest mean score, followed closely by attention to pain. Pain was a frequent and emotionally charged topic in the focus groups that we conducted to guide development of our questionnaire. Regard for pain has not been a high-ranking item in most surveys of adults' general preferences25 and has not even been included as an item in many surveys. Similarly, in the study by Ginsburg et al11 of factors that affect healthy adolescents' decisions to seek care, pain management was not important enough to be included in the final survey. However, comfort and pain were important aspects of quality of care for hospitalized patients in the Picker Institute studies of patient-centered quality.26 Our subjects likely have more in common with hospitalized patients than with the general population.
There are multiple reasons why pain management might be so critical for these adolescents with chronic illnesses. First, their conditions are all associated with significant pain. In the focus groups, adolescents with each of the conditions expressed concern about pain related to the condition. They were equally concerned, however, about pain related to procedures, ranging from surgery to venipuncture. Given the near certainty of procedure-related pain for these adolescents, it is perhaps not surprising that they rank attention to pain as a key quality aspect, although it was not as important for adolescents or adults in the general population.
Honesty appears to be equally important to adolescents with chronic conditions and those in the general population, receiving a high ranking similar to that found in the study by Ginsburg et al.11 In their qualitative exploration of the rankings, Ginsburg et al11 found that honesty involved “straight talk” and not giving false reassurance, which was interpreted as lying. Similarly, the adolescents in our study considered lack of full disclosure and “saying one thing and doing another” (eg, the physician changing her or her mind about drawing blood) to be examples of dishonesty (B. K. Klostermann, PhD, M.T.B., D. Nebrig, MSW, T. Tivorsak, BS, and G.B.S., unpublished data, 2003). Honesty has typically not been explicitly addressed in studies of adult preferences, although it has clearly been identified as a key aspect of the patient-physician relationship.
Cleanliness (ie, hand-washing and clean instruments) unexpectedly ranked highest among factors affecting the decision to seek care in the study by Ginsburg et al.11 Hand-washing was less important for our participants, however, with a mean importance score of 4.66. Although this factor was not considered unimportant, it ranked below the median for all items. This likely reflects our participants' greater experience with the health care system and perhaps less concern about nosicomial human immunodeficiency virus infection than was expressed by the adolescents in the study by Ginsburg et al.11
Items related to access to care were infrequently mentioned by adolescents in our focus groups, and the items on the questionnaire related to location and convenience were among the lowest rated. This finding contrasts with those from adult studies,6 in which access issues are usually ranked quite high. Two factors may account for this difference. First, our participants were already well established within the specialist system and thus had relatively good access to care. Second, their parents still had most of the responsibility for scheduling and other logistic issues. The lack of importance of amenities such as comfortable furniture and snacks mirrors findings from other studies of adults and adolescents.
When we investigated the influence of individual patient characteristics on ratings of quality and on communication preferences, only a few factors were significant. Parental education was inversely associated with mean scores for the quality of care items. Ginsburg et al11 also found lower scores for 4 of their factors among adolescents whose parents were college educated, compared with all others.
The lack of difference between girls and boys in our study was surprising. In most studies of adults and adolescents, girls rated aspects of quality more highly than did boys. For example, in the study by Ginsburg et al11 of factors affecting adolescents' access, girls rated 29 items higher than did boys, whereas boys rated only 2 items higher than did girls. Similarly, the Commonwealth Fund survey of >6000 adolescents in the general population found girls more likely to have preferences with respect to provider gender and parent-attended examinations.27 It is possible that chronic illness experience was a more powerful factor than general, sociocultural, gender-related predisposition. Many adolescents in our study had male physicians and were happy with their care. Their lack of gender preference may thus reflect their own positive experiences, rather than a hypothetical preference for providers of the same gender. Clearly, additional work is necessary to elucidate this potential interaction between health status and gender and to determine how it might affect desires for, and perceptions of, medical care. There are numerous studies documenting the fact that sicker patients tend to be less satisfied with their health care. We are not aware, however, of studies explicitly investigating the impact of a chronic condition on the prioritization of health care services. Functional health status was not independently associated with patient priorities or preferences in our study, except that adolescents whose health had recently improved were more likely to desire communication with the physician regarding personal issues.
Our study has some limitations. First, our participants received care at a single hospital, and their experiences and preferences may not be generalizable to other regions or other sociocultural groups. Because the subjects were well established in the health care system, their views may not reflect those of subjects who have recently received diagnoses or who lack established relationships with providers. In particular, adolescents of Hispanic ethnicity were not well represented in our sample, which reflects the ethnic distribution of the geographic area in which the study was conducted. Second, the participants' conditions are characterized by the need for ongoing self-care and by episodic flares of disease activity. We purposely chose this somewhat homogeneous group to reduce the disease-related variability in the study. Adolescents with other illness trajectories, such as trauma or cancer, and those being treated for acute illnesses may have different priorities. Third, funding limitations precluded inclusion of a comparison group of healthy adolescents. Therefore, although we could compare our findings with those in the literature, we could not directly determine whether healthy adolescents would respond to our questionnaire differently, compared with adolescents with chronic conditions. This is an area for future research. Although the psychometric properties of the measure were fairly strong, the sample size was relatively small; therefore, conclusions about the psychometric properties, especially the factor structure, should be considered preliminary. Additional methodologic development in larger and more diverse populations of adolescents would allow us to draw firmer conclusions, especially regarding individual characteristics of adolescents associated with priorities and preferences.
Despite the limitations identified above, data from this study support several conclusions. First, for adolescents with chronic illnesses, technical competence is important but interpersonal aspects of care are key determinants of quality. In particular, adolescents want their physicians to be honest with them. Second, they want to participate in their own care and to have their viewpoints and concerns taken seriously. Third, they want physicians to pay attention to their pain, including procedural pain. The Institute of Medicine report1 provides a framework for incorporating these adolescent priorities into health care. Improvements can occur at the level of the team caring for the patient (the “microsystem”), at the organizational level, and at the environmental or policy level. First, communication with adolescents is an area in which physicians feel poorly prepared28–30 and is thus an area in which significant improvement could occur. Rigorous studies have demonstrated that skill-building interventions can improve the effectiveness of physician-patient communication and that the effects are lasting.31,32 Second, providers should be careful to avoid the appearance of dishonesty. Third, providers should increase their attentiveness to pain, particularly pain related to common procedures such as venipuncture. At the organizational level, systems such as routine use of previsit questionnaires and scheduled time alone with the physicians for adolescents who desire it can help ensure that concerns are heard.
At the environmental or policy level, quality of care measures for adolescents, such as the Foundation for Accountability Young Adult Health Care Survey,33 focus on preventive care. New measures aimed at managing chronic illnesses should consider addressing pain, because our findings suggest that pain management is a key determinant of quality. Incorporation of this dimension into quality measures or into future studies of adolescents' perceptions of quality may help us to bridge the chasm between the care these adolescents receive and the care they desire and deserve.
This work was supported by a Generalist Physician Faculty Scholars Program Grant (to M.T.B.) from the Robert Wood Johnson Foundation.
We thank Dawn Nebrig, Kristin Rudolph, Max Cote, and Michael Yi for assistance with data collection.
- ↵Institute of Medicine. Crossing the Quality Chasm: A New Health System for the 21st Century. Washington, DC: National Academy Press; 2001
- ↵Stewart MA. Effective physician-patient communication and health outcomes: a review. CMAJ.1995;152 :1423– 1433
- ↵Safran DG. Defining the future of primary care: what can we learn from patients? Ann Intern Med.2003;38 :248– 255
- ↵Ginsburg KR, Menapace AS, Slap GB. Factors affecting the decision to seek health care: the voice of adolescents. Pediatrics.1997;100 :922– 930
- ↵Litt IF, Cuskey WR. Satisfaction with health care: a predictor of adolescents' appointment keeping. J Adolesc Health.1984;5 :196– 200
- ↵Newacheck PW, Strickland B, Shonkoff JP, et al. An epidemiologic profile of children with special health care needs. Pediatrics.1998;102 :117– 123
- ↵US Department of Health and Human Services, Maternal and Child Health Bureau. The National Survey of Children with Special Health Care Needs Chartbook. Washington, DC: US Department of Health and Human Services; 2004
- ↵US Congress, Office of Technology Assessment. Adolescent Health: Volume II: Background and the Effectiveness of Selected Prevention and Treatment Services. Washington, DC: US Government Printing Office; 1991
- ↵Tivorsak TL, Britto MT, Klostermann BK, Nebrig DM, Slap GB. Are pediatric practice settings adolescent friendly? An exploration of attitudes and preferences. Clin Pediatr.2004;43 :55– 61
- ↵Duke PM, Litt IF, Gross RT. Adolescents' self-assessment of sexual maturation. Pediatrics.1980;66 :918– 920
- ↵Gray WM. How Is Your Logic? Experimental Edition, Form A. Boulder, CO: Biological Sciences Curriculum Study; 1976
- ↵Landgraf JM, Abetz L, Ware JE. The Child Health Questionnaire User's Manual. 1st ed. Boston, MA: Integrated Therapeutics Group; 1996
- ↵Sawyer SM, Phelan PD, Bowes G. Reproductive health in young women with cystic fibrosis: knowledge, behavior and attitudes. J Adolesc Health Care.1995;17 :46– 50
- ↵Kolbe LJ, Kann L, Collins JL. Overview of the youth risk behavior surveillance system. Public Health Rep.1993;108(suppl 11) :2– 10
- ↵Brener ND, Collins JL, Kann L, Warren CW, Williams BI. Reliability of the youth risk behavior survey questionnaire. Am J Epidemiol.1995;141 :575– 580
- ↵Grol R, Wensing M, Mainz J, et al. Patients' priorities with respect to general practice care: an international comparison. European Task Force on Patient Evaluations of General Practice. Fam Pract.1999;16 :4– 11
- ↵Edgman-Levitan S, Cleary PD. What information do consumers want and need? Health Aff.1996;15 :42– 56
- ↵Lewin SA, Skea ZC, Entwistle V, Zwarestein M, Dick J. Interventions for providers to promote a patient-centered approach in clinical consultations. Cochrane Database Syst Rev.1995;4 :C003267
- Copyright © 2004 by the American Academy of Pediatrics