OBJECTIVE. The purpose of this work was to examine pediatricians’ responses to behavioral health concerns raised in the context of rural primary care visits with particular focus on time spent.
METHODS. Research assistants directly observed 302 patient visits in 2 rural pediatric primary care offices. The length of the visit, concerns raised, and physicians’ responses were recorded. Interrater reliability, scored for 25% of observed visits, was strong.
RESULTS. Behavioral, emotional, or developmental concerns were raised by either the parent or physician in 23.6% of all primary care visits observed. Approximately 9% of all visits were identified as a psychological consultation before the visit and lasted ∼7 minutes longer than visits for other reasons. Behavioral concerns were raised during ∼18% of visits not originally identified as a psychological consultation. In these cases, visit length increased by >5 minutes on average, a statistically significant difference. In addition, during these visits, discussion of behavioral concerns often consumed more than half of the visit. Data suggested that physicians were responsive to behavioral, developmental, or emotional concerns, engaging in further assessment, supportive statements, treatment, or referral in ∼97% of the visits when such concerns were raised.
CONCLUSIONS. Findings converge with previous research, showing that approximately one quarter of all primary care visits involve a discussion of behavioral, developmental, or emotional concerns. Moreover, this study is the first to document the specific impact of such concerns on pediatricians’ time, often sited as a key reason why they struggle to effectively identify and treat behavioral concerns presenting in primary care. These data provide a starting point for controlled studies, including comparisons of rural versus urban samples and the impact of various collaborative models of care.
It has been 3 decades since behavioral concerns were first described as the “new morbidity” in pediatric primary care practice.1 Today this trend continues with an estimated 15% to 21% of children presenting in primary care with behavioral health concerns.2,3 Survey data show that pediatricians rank behavioral concerns as the most common problem in their practice.4 In addition, a study of clinician-identified psychosocial problems demonstrated an increase from 6.8% in 1979 to 18.7% in 1996,5 results that were corroborated by 2 national studies showing that diagnostic assessments for children with attention-deficit/hyperactivity disorder (ADHD) increased threefold in primary care between 1989 and 1996.6
The demand for increased behavioral health services in primary care has critical implications in rural settings, which are often identified as “mental health profession shortage areas” based on population/provider ratios. The lack of behavioral health resources in these areas may result in a larger volume of behavioral health concerns presenting in primary care than estimated in national samples.3 Moreover, rural physicians are expected to act not only as gatekeepers but also as specialists in treating such concerns. For many rural practices, this scenario takes place in a context already taxed by a shortage of primary care physicians. According to one study, 25% of rural physicians reported they would leave their jobs in the next 2 years because of the overwhelming demands placed on them in their practice.7
Voluminous writing in this area speculates that physicians struggle to respond to behavioral health concerns because of the traditional dichotomy of behavioral and medical health, lack of fiscal incentives, and gaps in training7–9; however, it seems that the albatross, particularly in rural areas, is time. It is well known that primary care physicians operate on a tight schedule. This is evidenced by data showing that the average length of a primary care appointment is 16.3 minutes but much shorter for children, because length of visit was shown to increase with age.10 Indeed, “time constraints” is frequently offered as an explanation for why pediatricians struggle with identification and diagnosis,3,8,11 as well as appropriate treatment.6,12
One solution proposed to address the behavioral health demand on pediatricians’ time is increased collaboration between physicians and psychologists.11,13,14 In fact, there is a growing literature documenting the use of various coordinated service models between behavioral health professionals and physicians across a wide variety of primary care settings,15–17 including pediatrics.18,19 One particularly popular approach consists of collaboration by “colocation,” in which the psychologist works in the primary care setting, taking referrals from the pediatrician and communicating as needed while each professional operates independently in his or her area of expertise.
Bray and Rogers15 identified professional proximity as a key factor in the success of collaborative practice. A shared location has the overall advantage of facilitating communication among professionals, leading to more efficient, seamless health care. More importantly, the ease of access to behavioral health professionals and their services afforded by collaborative, colocated practice may decrease the amount of time physicians spend managing behavioral health concerns. In rural areas, a variety of benefits of a colocated collaboration have been cited, including decreased travel and increased patient privacy.7 However, the potential impact on overwhelmed physicians’ time may be the most salient advantage.
Despite the fact that “time” is hypothesized to play a major role in pediatricians’ responses to behavioral health concerns and used as a rationale for promoting collaborative models of care, no studies to date have examined pediatricians’ use of time directly. Specifically, there are no data that examine the amount of time pediatricians are spending in primary care visits when behavioral concerns arise. A direct examination of time spent in pediatric primary care is needed as the field attempts to develop strategies regarding the use of that time.
In addition, there is a gap in the current literature regarding how that time is spent. In one study, direct observation showed that behavioral, developmental, and emotional concerns were raised in about half of all pediatric primary care visits, and medical residents “ignored” those concerns in 17% of the cases.20 At the same time, studies using physicians’ self-reports indicate that behavioral concerns are raised in 15%21 to 27%22 of all visits, with approximately one third to one half of those concerns receiving no intervention or referral to a mental health provider. Thus, it seems these behavioral health concerns are often overlooked by physicians.
Finally, Riekert et al11 examined physicians’ responses to behavioral concerns when coordinated behavioral health screening and treatment services were provided. In this study, physicians referred children for screening based on their clinical impressions. A medical chart review was used to analyze physicians’ responses subsequent to feedback from the screening. Results showed no documented follow-up for 35% of children screened, although it is noteworthy that a commensurate number of children did not have clinically significant scores on the screening measure. Overall, it seemed physicians showed increased responsiveness to behavioral health concerns after the coordinated services were offered, with 42% referred for behavioral health services.
Whereas these studies examine important aspects of physician behavior in primary care visits (time spent assessing, prescribing, and referring), they leave unanswered critical questions regarding the amount of time spent, as well as other kinds of responses, such as supportive counseling or the direct provision of a therapeutic intervention. An examination of physicians’ responses, including the amount of time spent with patients, is essential as both the behavioral health and medical fields work to develop appropriate resources for the primary care setting. In fact, various authors have called for this kind of empirical work, particularly in rural settings, where the shortage of behavioral health professionals will likely remain static.23
The purpose of the present study was to examine how rural pediatricians responded to behavioral health concerns. In particular, this study focused on the amount of time spent when a behavioral concern was raised, as well as physicians’ responses to those concerns, including their use of a colocated collaborative behavioral health clinic. The current study is the first to examine these issues in a “real-world” rural pediatric primary care setting. In addition, this study improves on previous research by using a direct observation methodology.
Participants and Setting
Participants were 7 pediatricians from 2 pediatric clinics in rural Nebraska. Both clinics were located in, and surrounded by, counties considered shortage areas for mental health professionals. Three of the pediatricians were male, and 4 were female. The average number of years in practice was 14, with a range of 6 to 28 years. Three of the pediatricians worked full time, and the percentage of total visits observed for each was 33.2%, 24.8%, and 16.4%. Four of the pediatricians worked part time and were observed for 10.1%, 5.7%, 5.0%, and 4.7% of the total sessions. For a majority of visits observed (79.3%) the family was meeting with their designated primary care physician.
Each of the 2 practices had a Behavioral Health Outreach Clinic (BHC) at the time of the study. The BHCs are part of a larger a network of clinics, colocated within primary care settings in various rural communities around Nebraska. The BHCs are staffed by faculty, postdoctoral fellows, and predoctoral interns from the Pediatric Psychology Department at the Munroe-Meyer Institute, University of Nebraska Medical Center. Staff from the BHCs take referrals from the physicians in the practice and provide regularly scheduled behavioral health assessment and treatment services on an outpatient basis. Because the BHCs are located in the primary care office, collaboration between professionals occurs on a frequent and as-needed, informal basis.
A total of 302 pediatric visits were observed. Data were collected data in clinic A over 1 full calendar year and in clinic B over 7 consecutive months (February to August). The average patient age was 5.9 years of age (range: newborn to 20.3 years). Fifty-seven percent were male (N = 174), and 43% were female (N = 128). Ethnic demographics matched US Census data of 2000 for these 2 communities. Specifically, 84.2% were white, 13.4% were Hispanic, 1% were Asian American, 1% were black, and 0.7% were American Indian. Most patients were accompanied to the pediatric visit by their mother (76.2%), followed by both parents (11.6%), fathers (8.9%), a relative (2.3%), or a foster parent (0.7%). Comparisons showed these demographic variables to be similar across clinics A and B. Finally, all of the patients carried either private or state-funded insurance, although data were not maintained on this specific distribution.
Before each observed pediatric visit, research assistants used office notes on the physician’s schedule to code the anticipated reason for the visit according to 1 of 4 categories: acute, chronic, well-child, and psychological consultation. This classification system is a modified version of one developed by Starfield et al.24 Examples of “acute visits” include bronchitis, eye irritation, poison ivy, and urinary tract infection. Examples of “chronic conditions” include obesity, diabetes, and asthma. “Well-child” visits were, for example, school physicals and immunizations. “Psychological consultation” was coded, for example, when children attended for medication checks for ADHD/depression or ADHD diagnostic workups. Acute and well-child visits made up the majority of sessions observed, accounting for 54.6% and 30.8% of the sessions, respectively. A proportionately smaller number of the sessions were categorized as psychological (8.9%) and chronic (5.6%).
Data for the current article were obtained in the context of a broader study examining a variety of factors occurring in primary care visits. For the purposes of this article, observers collected data on the length of each visit in minutes. Session times began when the physician walked into the examination room and ended when he or she left. Second, observers used a coding sheet to count the number of medical and behavioral health concerns that were raised, who raised them (caregiver or physician), and the physician’s response(s). Behavioral health concerns were defined as any that fit into a list of 30 problem areas, including a range of psychiatric diagnoses (eg, ADHD and anxiety), related symptoms (eg, tantrums and tics), and social problems (eg, social skills and sibling rivalry). Physician response(s) were coded as: (1) no action taken; (2) prescribed medication; (3) provided a relevant informational handout; (4) referred to BHC; (5) referred to outside mental health professional; (6) provided specific intervention recommendations; (7) provided supportive statements; and (8) gathered further information/assessed problem. This coding scheme was adapted from a variety of sources.25–27 Finally, observers rated the proportion of the visit spent on behavioral concerns on a 5-point scale (1 = none, 2 = less than half of the visit, 3 = half of the visit, 4 = majority of the visit, 5 = entire visit).
Undergraduate and graduate research assistants were trained as observers by first familiarizing themselves with the study definitions and the coding process. They then coded a sample audiotape, and their ratings were compared with the primary investigator’s ratings. Finally, each research assistant observed 3 to 4 pediatric visits. Their coding forms and audiotapes were reviewed by the primary investigator to ensure that they met the 90% reliability criterion before independently coding sessions.
At the beginning of each observation day, research assistants were assigned to 1 physician, whose primary care visits they would observe during a 1- to 4-hour time period, depending on their availability for that day. This assignment was made based on the availability of physicians and an attempt to distribute observations across all of the physicians in the practice. Research assistants first reviewed the physician’s list of scheduled patients and coded the anticipated reason for the visit as acute, chronic, well-child, or psychological (as described above). After each patient was brought back to the room and was waiting for the physician, the assistant entered the room and obtained consent. Although data were not specifically coded for families who declined participation, we estimate that <5 families declined participation. Assistants collected data throughout the visit and did not interact with the physician or patient.
All of the observed sessions were audio taped, and 25.2% were randomly selected and independently coded. Reliability was calculated separately for each of the primary variables described above. An agreement was scored for length of visit if the total times coded were within 1 minute of each other. Agreements were counted for the rest of the variables if they matched exactly (ie, the specific number of behavior concerns was the same across both coders). Reliability was then calculated as the number of agreements over the total number of agreements plus disagreements, multiplied by 100%.
Overall, agreement across independent coders indicates adequate reliability for each of the variables. Percentage of agreement was as follows: length of visit (97.4%); number of medical concerns raised (81.6%); how much of the visit was spent on behavior-related concerns (94.7%); and total number of behavior concerns raised (94.7%). For physician responses during each session, the following agreements were obtained according to response: no action taken (100%); prescribed medication (97.4%); provided a handout (100%); referred to BHC (100%); referred to outside mental health professional (98.7%); provided specific intervention recommendations (96.1%); provided supportive statements (97.4%); and gathered further information/assessed problem (100%). Intercoder agreement for the observer ratings regarding the length of time spent discussing behavioral concerns was 94.7%.
Frequency of Behavioral Health Concerns Raised
Across clinics, the average number of medical concerns raised during a visit was 1.39 (range: 0–6), and the average number of behavioral concerns raised was 0.50 (range: 0–9). Behavioral concerns were raised by the parent and/or physician in 23.6% of all visits (clinic A: 21.5%; clinic B: 29.3%). When a behavioral health concern was raised, it was raised by the parent 74% of the time and by the pediatrician 26% of the time. When concerns were raised, the average number of concerns discussed was 2.10 (range: 1–9). The percentage of behavioral concerns raised in well-child visits was 22.5%; in acute visits, 12.1%; in chronic visits, 18.8%; and in psychological consultations, 100%.
Results showed that visits coded in advance as a psychological consultation were the longest in comparison with those identified as well-child, chronic, and acute (see Table 1). Specifically, visits identified before attendance as psychological consultations were 6.9 minutes longer on average than visits for other reasons. An analysis of variance indicated that these differences in length were statistically significant (F3,293 = 19.26; P < .001). Using the Scheffe method for posthoc comparisons, analyses revealed that visits coded as psychological consultations were significantly longer than both acute (P < .001) and chronic (P = .006) visits.
In addition, results showed that when behavioral concerns were raised spontaneously in the context of an acute, chronic, or well-child visit, the average length of that visit was 16.57 minutes (SD = 8.55) as compared with 11.36 minutes (SD = 5.93) when no behavioral concerns were raised, a difference of >5 minutes. An independent-samples t test showed that this difference is statistically significant (P < .001). Table 1 depicts minutes spent by visit type when behavioral concerns were not raised and when they were.
An analysis of observer-based ratings of time spent discussing behavioral concerns showed that, as expected, a vast majority of sessions preidentified as psychological consultation were spent discussing behavioral concerns. These data showed that, across all of the visits observed, pediatricians spent up to one half of the session discussing behavioral concerns in 10.9% of all acute care visits, 18.8% of chronic visits, and 19.5% of well-child visits. In examining only those visits in which behavioral concerns were raised, estimates of time spent discussing such concerns were significantly higher, as depicted in Table 2.
Type of Response
Table 3 illustrates physicians’ responses when behavioral concerns were presented. The most common response was to gather information about the concern (70.4%), followed by making supportive statements (56.3%). The frequency of response types was generally comparable across the 2 clinics with the exception supportive statements. According to a χ2 analysis of each response (α was set at .01), clinic A provided significantly more supportive statements when behavioral concerns were presented. Whereas these data do not indicate how referrals related to specific concerns, data show that referrals were common across a wide variety of concerns raised, including internalizing problems (eg, anxiety and depression), externalizing problems (eg, ADHD, aggression, noncompliance, and tics/habits), and academic and behavioral concerns at the child’s school.
The present study was the first to examine rural pediatricians’ responses to patients’ behavioral health concerns in primary care. Results showed that behavioral health concerns were raised by either parents or physicians in almost one quarter (23.6%) of all of the visits observed. This frequency is generally commensurate with estimates from large-scale studies with broader samples showing that 15% to 21% of patients presenting in pediatric primary care have behavioral health concerns.3,5 In addition, these data, generated by direct observation, are consistent with studies using clinicians’ self-reports estimating that 15% to 27% of sessions involve behavioral health issues.21,22
The general convergence of these data across studies using varied methods is noteworthy. Both the present study and that of 1 previous study21 had no age restrictions, whereas other studies reported on children ages 4 to 8 years22 and 4 to 15 years.5 In addition, the present study sampled a rural population, whereas the others did not. As described above, data collection approaches also varied across these studies. Taken together, findings from these studies have begun to provide a well-established expectation for patient demand.
At the same time, it is surprising that this study did not document a higher rate of behavioral concerns, as hypothesized, given the use of a rural setting with greater provider shortages. One potential explanation for this may be this study’s inclusion of data from infant visits, which may have suppressed the overall frequency of behavior problems presented. In fact, the percentage of behavioral concerns presented was greater (33% of all visits) when only children ages ≥4 years were included in analyses. Future research should be designed to make direct comparisons across urban-rural or shortage and nonshortage area samples, keeping in mind the potential impact of sample age.
The primary focus of this study was physicians’ time spent dealing with behavioral concerns in day-to-day practice. Results showed that during visits identified in advance as psychological consultations, pediatricians spent an average of ∼7 minutes longer than in visits for other reasons (ie, well-child, acute, or chronic concerns). Because the physicians in this sample often “budget” for this by scheduling more time for psychological consultations, the extra 7 minutes may not impact their ability to stay on schedule; but this time expenditure does decrease the number of patients that can be seen daily. For example, according to these data, in a 35-patient daily caseload, ∼3 (∼9% of all visits) would be identified as needing a 20-minute psychological consultation. Thus, during the hour a physician would spend in psychological consultations, ∼6 acute, 5 chronic, or 4 well-child visits could be addressed.
More importantly, when behavioral concerns were raised spontaneously during visits identified as well-child, acute, or chronic in nature, pediatricians spent >5 minutes longer than when such concerns were not raised. This extra time is unplanned for and could result in the physician getting behind schedule, a situation that could certainly impact a physician’s job performance and stress level, as well as patient satisfaction. Again, extrapolating from a daily caseload of 35 patients, ∼32 would be scheduled for acute, chronic, or well-child care. Of these, ∼8 would involve the spontaneous discussion of behavioral, developmental, or emotional concerns. A physician who, according to these data, spent ∼5 extra minutes with each of these cases would be behind by an average of 42 minutes at the end of the day.
Whereas the present study does not provide data regarding the specific number of minutes spent discussing behavioral concerns, ratings with high interrater reliability showed that when a behavioral concern was raised, discussion about those concerns absorbed at least half of the visit for ∼35% of acute visits, 67% of chronic visits, and 43% of well-child visits. These data suggest that when behavioral concerns are raised, increased visit length results from physicians taking the time to discuss those concerns.
These findings are further supported by results showing that when a behavioral concern was identified, physicians offered some form of response 97% of the time. Few studies have examined physicians’ responses to behavioral concerns, and it is difficult to draw comparisons across studies because of differences in sampling and methodology. Tentative comparisons suggest that physicians’ responses to behavioral concerns in this study are higher than that reported in previous studies, which show responses in only 69%22 to 83%20 of all cases. Moreover, physicians in this study engaged in more frequent assessment (∼70% of all responses) and supportive statements (50%). Previous work in this area shows that these rates at 43% and 3%, respectively.20 If the physicians in this study demonstrated greater responsiveness to patients’ behavioral health needs, this may be because, given the lack of community resources, physicians in rural areas are serving as both gatekeepers and behavioral health providers. An alternate explanation for this finding may be the existence of a colocated, collaborative behavioral health clinic, because it is possible that physicians working within this model experience greater confidence and skill in dealing with such concerns.28
Although the current study adds to the much-needed empirical literature regarding behavioral health care in primary care settings, it is not without limitations, and future research is encouraged. For instance, methodologic weaknesses in the current study limit its external validity. Specifically, using a small number of physicians in a strictly rural and health care-insured sample restricts its potential application to other practices, communities, and demographic groups. Future research can build on these findings by broadening sample characteristics. In addition, whereas direct observation has many advantages over other data collection methods (eg, physician-report and medical-chart review), it does present the potential for observer effects that could alter both patient and physician behavior.
Future research should also provide a more detailed account of physician behavior. Whereas the current study contributes a broad perspective on physicians’ responses to behavioral concerns, follow-up studies could examine more specific responses, such as the manner in which physicians recruit information regarding behavioral health concerns (ie, general questions versus questions targeted at specific areas of concern). Moreover, ongoing study should examine differences between physicians (eg, training in behavioral health care, approach to behavioral concerns, etc), explain why they exist, and delineate ways to capitalize on these when searching for systemwide solutions for dealing with behavioral health care in primary care settings.
Findings converge with previous research, showing that approximately one quarter of all primary care visits involve a discussion of behavioral, developmental, or emotional concerns. Moreover, this study is the first to document the specific impact of such concerns on pediatricians’ time, often sited as a key reason why they struggle to effectively identify and treat behavioral concerns presenting in primary care. These data provide a starting point for future controlled studies of time, including comparisons of rural versus urban samples, the impact of various collaborative models of care, and specific physician responses to behavioral concerns raised in the context of primary care visits.
The following behavioral pediatric trainees, supported by a grant from the Nebraska Healthcare Cash Fund, contributed to the collection of data for this study: Torri Smith, University of Nebraska-Omaha; Stephanie Cole, Hastings College; and Natalie Anderson, Tammi Beckman, and Jessica Mack, University of Nebraska-Kearney.
We are especially grateful to the physicians and staff at the 2 rural pediatric clinics for their participation in this study.
- Accepted January 27, 2006.
- Address correspondence to Rachel J. Valleley, PhD, Pediatrics, Munroe-Meyer Institute, University of Nebraska Medical Center, Omaha, NE. E-mail:
The authors have indicated they have no financial relationships relevant to this article to disclose.
- ↵Haggerty RJ, Roughman KJ, Pless IB. Child Health and the Community. New York, NY: John Wiley and Sons; 1975
- ↵Lavigne JV, Gibbons RD, Arend R, Rosenbaum D, Binns HJ, Christoffel KK. Rational service planning in pediatric primary care: Continuity and change in psychopathology among children enrolled in pediatric practices. J Pediatr Psychol.1999;24 :393– 403
- ↵Kelleher KJ, McInerny TK, Gardner WP, Childs GE, Wasserman RC. Increasing identification of psychosocial problems: 1979–1996. Pediatrics.2000;105 :1313– 1321
- ↵Perrin E, Stancin T. A continuing dilemma: whether and how to screen for concerns about children’s behavior in primary care settings. Pediatr Rev.2002;23 :264– 275
- ↵Wolraich ML. The referral process: the pediatrician as gatekeeper. In: Brown RT ed. Cognitive Aspects to Chronic Illness in Children. New York, NY: Guilford Press; 1999:15– 24
- ↵Riekert KA, Stancin T, Palermo TM Drotar D. A psychological behavioral screening service: use, feasibility, and impact in a primary care setting. J Pediatr Psychol.1999;24 :405– 414
- ↵Drotar D. Consulting With Pediatricians: Psychological Perspectives. New York, NY: Plenum Press; 1995
- ↵Strosahl K. The psychologist in primary health care. In: Kent AJ, Hersen M, eds. A Psychologists’ Proactive Guide to Managed Mental Health Care. Mahwah, NJ: Lawrence Erlbaum Associates; 2000:87– 112
- Getler AA, Joseph RC, Grunebaum J, Grunebaum H, Eder G. Collaboration in action: Innovations in primary care and mental health. Fam Syst Health.2001;19 :459– 466
- ↵Clay DL, Stern M. Pediatric psychology in primary care. In: James LE, Folen RA, eds. Primary Care Clinical Health Psychology: A Model for the Next Frontier. Washington, DC: American Psychological Association; 2005
- ↵Drotar D. Psychological research in pediatric settings: lessons from the field. J Pediatr Psychol.1989;16 :1– 11
- ↵Finney JW, Riley AW, Cataldo MF. Psychology in primary health care: effects of brief targeted therapy on children’s medical care utilization. J Pediatr Psychol.1991;16 :447– 461
- ↵Sharp L, Pantell RH, Murphy LO, Lewis CC. Psychosocial problems during child health supervision visits: eliciting, then what? Pediatrics.1992;89 :619– 623
- ↵Williams J, Klinepeter K, Palmes G, Pulley A, Foy JM. Diagnosis and treatment of behavioral health disorders in pediatric practice. Pediatrics.2004;114 :601– 606
- ↵Horwitz SM, Leaf PJ, Leventhal JM, Forsyth B, Speechley KN. Identification and management of psychosocial and developmental problems in community-based primary care pediatric practices. Pediatrics.1992;89 :480– 485
- ↵Kelleher KJ, Long N. Barriers and new directions in mental health services research in the primary care setting. J Clin Child Psychol.1994;23 :132– 142
- ↵Garrison WT, Bailey EN, Garb J, Ecker B, Spencer P, Sigelman D. Interactions between parents and pediatric primary care physicians about children’s mental health. Hosp Commun Psych.1992;43 :489– 493
- Goldberg ID, Roghmann KJ, McInerny TK, Burke JD. Mental health problems among children seen in pediatric management: prevalence management. Pediatrics.1984;73 :278– 293
- ↵Wissow LS, Roter DL, Wilson MEH. Pediatrician interview style and mothers’ disclosure of psychosocial issues. Pediatrics.1994;93 :289– 295
- ↵Robinson P. Improving care for a primary care population: depression as an example. In Maruish M, ed. Handbook of Psychological Assessment in Primary Care Settings. Mahwah, NJ: Lawrence Erlbaum Associates Publishers; 2000
- Copyright © 2006 by the American Academy of Pediatrics