Journal of Rural Community Psychology Volume E7 Number 2 Fall 2004
The Impact of Behavioral Healthcare Services on Medical Utilization
for Children with Externalizing Disorders in a Rural Community
Rachel J. Valleley, Jodi Polaha, & Joseph H. Evans
Munroe-Meyer Institute for Genetics and Rehabilitation
University of Nebraska Medical Center
The integration of behavioral health services into primary care offices offers a variety of possible advantages including decreased use of medical services. This is of particular benefit in rural communities. The purpose of this study was to extend previous research in the area of “medical cost offset” to children in a rural primary care pediatric practice. All of the children were diagnosed with an externalizing disorder and were seen at a primary care-integrated Behavioral Health Clinic (BHC) utilizing brief, problem-focused, and empirically-supported behavioral interventions. Data was collected for the year prior to and after the initial BHC visit. Paired Samples t-tests revealed significant decreases in office visits, prescriptions, and certain types of visits (acute, acute/likely to recur, and psychological). Overall, the major implication of these results indicate that addressing co-occurring mental health problems in children with externalizing disorders can decrease contact necessary with a primary care physician. This may be particularly important in rural communities given great distances and diminished resources.
The Advantages of Integrated Care in Rural Areas
A model of integrated behavioral health and primary care has particular promise in rural areas. Schroeder and her colleagues were the first to describe the successful application of this model in rural North Carolina in the 1970’s (Schroeder, 1979). Since then, the integrated model has enjoyed increasing popularity. In this paradigm, the psychologist is located in the primary care setting, taking referrals from the pediatrician and communicating as needed while each operates independently in his or her area of expertise.
Indeed, a shared location has the overall advantage of facilitating communication among professionals, leading to more efficient, seamless health care. In rural areas, however, there may be additional advantages. First, an integrated model may facilitate follow-through with behavioral health treatment. deGruy (1997) points out that, in small communities, attending behavioral health services at the primary care offices offers greater anonymity or confidentiality than would be available to a person whose vehicle was parked at a free-standing mental health agency. Moreover, Strosahl (2000) has suggested that when the psychologist is located in primary care, mental health is “normalized” and families may be more likely to follow through with the physician’s recommendation to attend behavioral health services.
An important and overlooked advantage to this model in rural areas, however, is that it could decrease patients’ use of primary care services. Indeed, there is a large body of literature showing that accessing behavioral health services may reduce general medical use (Chiles, Lambert, & Hatch, 1999). The literature has touted this finding as an advantage for insurance companies because of the commensurate cost savings. In rural settings, however, this advantage could benefit patients directly in that attending fewer medical visits decreases the more substantial costs associated with greater driving time and distance (missed work/school, travel expenses, etc.). Moreover, for the rural physician, fewer medical visits may result in a more manageable caseload. Indeed, data suggest 25% of physicians are likely to leave their jobs due to overwhelming demand in rural primary care (deGruy, 1997).
Medical Use and “Cost-Offset”
It has been established that those with behavioral or emotional concerns use medical services more. In his review, deGruy (1997) stated “patients with mental diagnoses show consistently higher use of medical resources than their unaffected counterparts, generally on the order of twice baseline utilization rates” (p. 10). Likewise, studies of pediatric samples show increased medical utilization by school-age children is related to stress (Wertlieb, Weigel, & Feldstein, 1988), behavioral concerns (Hankin, et al., 1984; Costello, 1988) and psychosocial problems (Kinsman, Wildman, & Smucker, 1999). Moreover, there is evidence that many of these visits are unwarranted. Specifically, Shapiro (1971) reported that 50-80% of medical visits are by persons without any identifiable physical problem.
Naturally, this “overuse” of medical services results in exacerbated medical costs. Research has long focused on the “medical cost offset effect” as an important advantage to better-integrated behavioral health and primary care services. The “cost-offset effect” is the reduction of health care costs associated with treatment for behavioral health concerns. Support for this effect is robust. A meta-analytic review of 58 controlled studies concluded there is strong evidence for the effect, particularly for the cost of inpatient medical services for people over age 55 (Mumford, et al., 1984). In a similar analysis of 91 studies, evidence showed that behavioral health service provision resulted in an average savings of about 20% (Chiles et al., 1999).
A majority of research in the medical use area has focused on adults; however, a few studies have demonstrated this effect with pediatric samples. Graves and Hastrup (1981) examined medical visits over a two-year period for 63 low-income children in a subsidized health program. The children were divided evenly into three groups: a group receiving psychological services, a group matched by key demographic variables, and a group matched by key demographics as well as frequency of medical use and psychological distress. This study replicated findings in the adult literature showing that the groups referred for mental health services used services more than those who were not referred. Moreover, results showed reductions in medical utilization only for the group who received the psychological services.
A similar study evaluated the effects of brief, problem-focused behavioral protocols on the medical use of 93 children in a health maintenance organization (Finney, Riley & Cataldo, 1991). Results showed a subsequent reduction in primary care visits for acute illnesses (e.g., upper respiratory infection, ear infection) and nonmorbidity issues (e.g., health supervision, school physicals). A matched comparison group’s use was unchanged. Results showed the children referred for behavioral concerns (e.g., noncompliance, disruptive behavior) and toileting problems accounted for this reduction.
Medical Use in Rural, Pediatric Primary Care
As stated above, the results of such work may have particular clinical significance in rural areas, where great distances and diminished resources drive up the “cost” for accessing medical services. Along these lines, it is noteworthy that both of the studies described above used brief, problem-focused behavioral interventions. Graves and Hastrup (1981) reported patients attended an average of 4 mental health visits and Finney et al. (1991) reported treatments to occur over 1-6 sessions. Thus, there is some evidence that the reduction in services can occur with the limited additional “cost” for the behavioral health services themselves.
While previous research links behavioral health treatment with decreased medical use, there are few studies with children, and none that examine this outcome in the context of integrated care. As stated above, in rural areas, the possibility that the integrated model might decrease medical use has particularly salient implications, however, there is no research in this area. The current study was designed to assess the effects of rural, pediatric primary care-based behavioral health services on the medical use of patients who received those services. The study focused on children presenting with an externalizing behavior disorders. These children were selected based upon evidence that this is the most common problem raised by parents to pediatricians (Arndorfer, Allen, & Aljazireh, 1999).
Participants and Setting
Participants were children ages 2- 16 (x = 8) who were seen in an integrated behavioral health clinic between 1998 and 2001. The behavioral health clinic was operated in a private pediatric primary care practice in a small, rural Midwestern community (population=20,000). Only children diagnosed as having an externalizing behavior disorder (i.e., oppositional defiant disorder, attention-deficit/hyperactivity disorder, disruptive behavior disorder: not otherwise specified) at their first behavioral health visit were included. Only behavior disordered children were included in the sample to try to have a more homogenous sample. In order to ensure participants were actively receiving medical care in this clinic, children were only included if they had a medical record at the primary care office with at least one non-behavioral health entry prior to the initial BHC intake. One hundred fifty-one children met these criteria (48 females and 103 males).
The behavioral health clinic was staffed by a psychologist and graduate students in psychology. Over 90% of referrals to the clinic were made by the pediatricians in the practice. Clinic staff used brief, problem-focused, and empirically-supported behavioral interventions to address the externalizing behavior problems (for a review of the most common, see Christophersen & Mortweet, 2001). Recommendations were tailored for each participant based upon referral concerns, functional assessment, and family need. The participants in this study attended a mean of 4.94 behavioral health sessions (range 1-19).
Data were obtained from the participant’s medical record in the pediatric office. Information collected included the number of pediatric contacts (telephone and office), and medications prescribed. In addition, the types of concerns raised during the pediatric visit were categorized using an encounter classification system developed by Starfield et al. (1984). Types of concerns were categorized into the following areas; acute (e.g., bronchitis, epistaxis, eye irritation, poison ivy), acute/likely to recur (e.g., pharyngitis/strep, asthma, otitis media, urinary tract infection), injuries (e.g., bone fracture, contusion, dislocated shoulder, concussion, sprains), chronic (e.g., diabetes, epilepsy, acne, hearing loss, migraine headache), psychosocial/psychosomatic (e.g., school/family problems, headache, any DSM diagnosis), and nonmorbidity (e.g., well-child visits, school physical, sports physicals).
The participant’s medical record was reviewed and coded by a research assistant to include the year prior to behavioral health services (Year One) and the year following the initiation of those services (Year Two). Research assistants recorded the number of contacts, whether medications were recommended, and what concerns were raised during the visit. The concerns raised during the visit were determined by reading the physician’s dictation describing the visit. Medical staff were consulted when it was unclear which coding category was most appropriate and a list was maintained regarding decision rules. Only primary care visits were reviewed, thus, the coded contacts did not include behavioral health visits that may have been ongoing during Year Two.
A research assistant reviewed 20% of the files for reliability purposes. The coders agreed 89.8% of the time that a contact occurred. Inter-rater agreement was 98.9% for type of visit (office vs. telephone), 95% for over-the-counter medications, and 85% for prescriptions. All ambulatory visit encounter categories were above 90% for visit type (range, acute-likely to recur = 91% and Injury = 99%).
Year One - Year Two Comparisons
Paired Samples t-tests were conducted to compare the frequency of primary care contacts during the Year One and Year Two for the entire sample. The results of these analyses can be found in Table 1. Significant decreases across the two years were noted for the total number of office visits, prescriptions given, and over-the-counter medications recommended. The specific types of visits showing a decrease from Year One to Year Two included: acute visits, acute/likely to recur visits, and psychosocial visits. A decrease in the number of chronic visits approached significance. There were no significant differences in telephone contacts made during Year One and Year Two with the exception of an increase in telephone calls to fill a psychotropic medication (t= -2.010, p=.046).
Mean Office Contacts for Year One and Year Two
Variable Year One Mean Year Two Mean t Sig.
Total Office Visits 4.87 3.48 5.254*** .000
Prescriptions 3.09 2.31 2.682** .008
OTC Medications .98 .66 2.403* .017
Psych. Prescriptions .56 .62 -.534 .594
Acute Visits .42 .25 2.651** .009
Acute/Likely to Recur Visits 2.77 1.92 4.119*** .000
Injury Visits .23 .21 .310 .757
Chronic Visits .36 .18 1.902a .059
Psychosocial Visits 1.55 1.03 3.527** .001
Nonmorbidity Visits .38 .34 .749 .455
Procedure Visits .29 .32 -.535 .594
Note. N = 151. ap =.059. *p< .05. **p < .01. ***p < .001
Twenty-three percent of the sample reported their child had a chronic illness during the behavioral health clinic intake visit. Children with chronic health conditions require increased medical care, thus, additional analyses were completed to more closely examine this subgroup (see Table 2). Paired Samples t-tests were conducted for patients who were diagnosed with a chronic condition (e.g., diabetes, asthma) as well as those children who did not have a chronic medical condition. Much like the sample as a whole, children with chronic medical conditions showed significant decreases for the total number of office visits, prescriptions given, OTC medications recommended, and acute/likely to recur visits. There were no significant differences noted between Year One and Year Two telephone contacts.
Mean Office Contacts for Year 1 and Year 2 for Patients with Chronic Medical Conditions and those without Chronic Medical Conditions
Variable Year 1 Mean Year 2 Mean t Sig.
Total Office Visits
Chronic 6.47 4.13 4.068*** .000
No Chronic 4.52 3.39 3.652*** .000
Chronic 4.91 2.78 2.594* .014
No Chronic 2.54 2.17 1.226 .223
Chronic 1.38 .59 2.693* .011
No Chronic =.90 .68 1.342 .182
Chronic .50 .47 .150 .882
No Chronic .63 .71 -.580 .563
Chronic .56 .28 1.958a .059
No Chronic .36 .24 1.640 .104
Acute/Likely to Recur Visits
Chronic 4.19 2.69 3.623*** .001
No Chronic 2.39 1.73 2.711** .008
Chronic .34 .25 .533 .598
No Chronic .21 .21 .000 1.00
Chronic 1.22 .63 1.462 .154
No Chronic .13 .05 1.685 .095
Chronic 1.53 1.06 1.371 .180
No Chronic 1.63 1.10 3.020** .003
Chronic .38 .28 .77 .447
No Chronic .39 .37 .287 .774
Chronic .44 .38 .387 .701
No Chronic .25 .34 -1.174 .243
Note. N = 32. ap=.059. *p< .05. **p < .01. ***p < .001
For those children without a chronic medical condition, results were again similar to the sample as a whole and can be seen in Table 2. Significant decreases were found for total office visits, acute/likely to recur visits, and psychosocial visits. No significant differences were found for telephone contacts with the exception of psychotropic medications prescribed (t = -2.086, p = .039), which increased in the year after children had received BHC services.
Number of BHC Sessions
The sample in this study attended a range from 1-19 BHC visits, and it was hypothesized that those who attended more BHC sessions may have benefited more from those services, resulting in more marked decrease in medical use. Thus, a correlation was computed for the number of sessions attended and the change score in primary care use from Year One to Year Two (amount of decrease in medical services). Given this hypothesis, a positive correlation would be expected, with more BHC visits associated with a greater change score across years. In fact, no relationship was found (r=.081, p=.325).
As children age, they often necessarily have fewer medical needs and requirements for well-child visits. To ensure that decreases noted in visits from Year One to Year Two were not a function of children aging, the average change score (primary care visits during Year One – primary care visits during Year Two) for each age was calculated. The average change across years was -.11, indicating that overall decreases in office visits did not occur as children got older (Table 3). Thus, decreases observed in the primary analyses were not due to children aging.
Mean Office Contacts for Year One by Age
Age Year One Mean Change N Minimum Maximum
Two-year old 6.50 8 3 14
Three-year old 5.42 -1.08 12 0 11
Four-year old 5.60 .18 20 1 18
Five-year old 5.00 -.6 22 0 16
Six-year old 3.57 -1.43 7 1 7
Seven-year old 6.45 2.88 11 0 21
Eight-year old 3.57 -2.88 14 0 8
Nine-year old 3.27 -.3 15 0 9
Ten-year old 3.86 .59 7 1 9
Eleven-year old 4.71 .85 7 1 8
Twelve-year old 5.11 .4 9 0 18
Thirteen-year old 3.50 -1.61 4 0 7
Fourteen-year old 3.83 .33 6 1 7
Fifteen-year old 8.25 4.42 4 4 10
Sixteen-year old 5.00 -3.25 5 2 7
Average change -.11
Note. Change was calculated by subtracting Year One visits from Year Two visits.
Office Contacts Changes
Consistent with previous research, decreases were noted in the overall number of contacts children had with their pediatrician after receiving behavioral health services. The average decrease in primary care visits from Year One to Year Two was about 1.5 visits. For rural physicians facing a caseload heavy with behavioral health concerns, this decrease across patients could significantly impact rural physicians’ workload. First, it could save time, since it is likely that behavioral health problems are more time-consuming than a majority of medical concerns presenting in primary care.
Second, it could eliminate visits focused on issues physicians feel least prepared to address including questions about noncompliance, hyperactivity, or poor academic performance. Hence, in the context of a 40-patient daily caseload, this small decrease in visits could decrease physician stress, increasing the likelihood that they remain in rural communities. This potential outcome is supported by data from the present study showing that significant decreases were specifically noted for visits regarding psychological concerns.
Whether a small number of decreased medical visits would result in a significant savings for patients is less clear. While patients attended 1.5 fewer primary care visits on average during Year Two, they attended an average of approximately 5 behavioral health visits that year. Thus, the addition of these services resulted in an increase in clinic attendance of 3.5 visits. An analysis of 467 patients attending the behavioral health clinic at the primary care clinic from which the sample was drawn showed 52% traveled from outside the community where the clinic was located. The average round trip driving distance for this group was approximately 50 miles. Thus, the average increased travel mileage for this group during Year Two may have been about 175 miles. These kinds of specific figures highlight the importance of cost savings in terms of travel time/distance for rural populations.
There are, of course, other possible “cost-savings” associated with behavioral health visits not addressed in this study. For example, a parent whose child misbehaves at school may miss more work due to disruptive phone calls, team meetings, or other changes. Behavioral health clinic visits could decrease these more difficult to measure “costs.” Furthermore, it is possible that over subsequent years the behavioral health treatment would continue to “pay off” in reduced visits to the primary care provider.
In addition to decreases for overall contacts, significant decreases were noted for acute and acute/likely-to-recur visits. These findings are consistent with Finney and his colleagues, who asserted that parents are more likely to bring their child in to see their pediatrician for acute problems if they also have concerns about their child’s behavior (Finney et al., 1991). Specifically, the externalizing behavior problem could exacerbate an acute illness. For example, a sore throat may be more frustrating for a parent if the child is also noncompliant with taking medication.
Telephone Contacts Changes
Overall, differences were not found for telephone contacts from Year One to Year Two with the exception of psychotropic medications. Significantly more psychotropic medications were prescribed via the telephone in Year Two. This could be due to the fact that once behavioral interventions were combined with medication, children’s behavioral problems stabilized. This would not require additional visits to the physician to try to adjust psychotropic medications, only a phone call to refill the prescription. For families in rural communities, the management of concerns over the telephone is significantly more efficient than face-to-face contacts.
For the 22.5% of children who had a chronic medical condition, overall pediatric office visits were decreased by about 2 ½ visits for Year Two. Once again, a possible explanation for this decrease could be that chronic conditions may be more problematic when a child has a co-occurring externalizing disorder due to decreased medical adherence on the part of the child. Additional support for this hypothesis comes from significant decreases in prescriptions and OTC medications for children with chronic conditions. Furthermore, children without chronic conditions did not have significant changes in medications prescribed or recommended. Thus, it is possible that a child with a chronic condition may be following the overall medical regimen better when their behavioral problems have been addressed resulting in less “flare-ups” of the chronic condition and fewer additional medications needed.
In the present study, no correlation was found between the number of behavioral health clinic visits attended and the number of primary care visits attended in Year Two. One possible explanation for the lack of relationship could be that children who attended more behavioral health sessions did so because of more severe externalizing behavior problems or because their parent struggled to follow recommendations with adequate integrity. Thus, the family could have attended a large number of behavioral health sessions with little improvement in child behavior. In this case, the family may have continued to use primary care visits to address behavioral health concerns.
Limitations and Future Research
A primary limitation to this study is the lack of matched control group to lend further support that changes in from Year One to Year Two were due to behavioral health clinic contact. While age did not appear to function as a confound in this study, it is possible that other factors could have contributed to the decrease. Ideally, future research would include a comparison group matched for age, gender, and number of primary care visits in Year One who were never referred for behavioral health services. In addition, a third group of similarly matched children who were referred for behavioral health services at an agency outside the primary care clinic would provide information about the specific benefit of the integrated care model described in this study.
In addition, another limitation to this study was the lack of outcome data on treatment effects. It is difficult to conclude with great certainty that decreased medical visits in Year Two was due to improvement in behavior problem without having information related to patient outcomes (e.g., symptoms and functioning). Furthermore, it is impossible to determine subject attrition rates due to the retrospective nature of this study. While attrition is a possible explanation for the observed decreases in medical visits, it is unlikely that a significant number of parents changed primary care settings for their children given that no other pediatric offices were in the area.
Further research could examine whether the reduction in medical use continues for years following behavioral health services, thus adding to overall decreased “cost” for both the patient and the physician. Additional research could also explore further factors that result in decreased medical use after behavioral health services are delivered. For example, behavior change in children or decreased parental stress may predict this phenomenon. In addition, it is possible that the decrease is more robust for a subgroup of patients (e.g., less severe behavior problems, responsive/adherent to treatment recommendations). Finally, further research about integration of behavioral health into rural areas need to be conducted.
This study provides further support to previous results showing that that treating behavioral health problems results in fewer office visits to physicians. This study further extended the previous research to a rural community where resources are limited and decreased medical use may have significant implications for patient quality of life.
Arndorfer, R. E., Allen, K.D., & Aljazireh, L. (1999). Behavioral health needs in pediatric
medicine and the acceptability of behavioral solutions: Implications for behavioral
psychologists. Behavior Therapy, 30, 137-148.
Chiles, J. A., Lambert, M. J., & Hatch, A. L. (1999). The impact of psychological interventions
on medical cost offset: A meta-analytic review. Clinical Psychology: Science and
Practice, 6, 204-220.
Christophersen, E. R. & Mortweet, S. L. (2001). Treatments that Work With Children:
Empirically Supported Strategies for Managing Childhood Problems. Washington, DC:
American Psychological Association.
Costello, E. J. (1988). Primary care pediatrics and child psychopathology: A review of
diagnostic, treatment, and referral practices. Pediatrics, 78, 1044-1051.
deGruy, F. V. (1997). Mental healthcare in the primary care setting: A paradigm problem.
Families, Systems, & Health, 15, 3-26.
Finney, J. W., Riley, A. W., & Cataldo, M. F., (1991). Psychology in primary health care:
Effects of brief targeted therapy on children’s medical care utilization. Journal of
Pediatric Psychology, 16, 447-461.
Graves, R. L., & Hastrup, J. L. (1981). Psychological intervention and medical utilization in
children and adolescents of low-income families. Professional Psychology, 12, 426-433.
Hankin, J. R., Starfield, B. H., Steinwachs, D. M., Benson, P., Livingston, G., & Katz, H. P.
(1984). The relationship between specialized mental health care and patterns of primary
care use among children enrolled in a prepaid group practice. Research in Community
and Mental Health, 4, 203-220.
Kinsman, A. M., Wildman, B. G., & Smucker, W. D. (1999). Brief report: Parent report about
health care use: Relationship to child’s and parent’s psychosocial problems. Journal of
Pediatric Psychology, 24, 435-439.
Mumford, E., Schlesinger, H. J., Glass, G. V., Patrick, C. & Cuerdon , T. (1984). A new look at
evidence about reduced cost of medical utilization following mental health treatment.
Psychiatry, 141, 1145-1156.
Schroeder, C. S. (1979). Psychologists in a private pediatric practice. Journal of Pediatric
Psychology, 4, 5-18.
Shapiro, A.K. (1971). Placebo effects in medicine, psychotherapy, and psycho-analysis. In S. L.
Garfield, & A. E. Bergin (Eds.), Handbook of Psychotherapy and Behavior Change: An
Empirical Analysis (pp. 439-473). New York: John Wiley.
Starfield, B., Katz, H., Gabriel, A., Livingston, G., Benson, P., Hankin, J., Horn, S., &
Steinwachs, D. (1984). Morbidity in children: A longitudinal view. New England
Journal of Medicine, 310, 824-829.
Strosahl, K. (2000). The psychologist in primary health care. In A. J. Kent & M. Hersen (Eds),
A Psychologists’ Proactive Guide to Managed Mental Health Care. Mahwah, NJ:
Lawrence Erlbaum Associates, 87-112.
Wertlieb, D., Weigel, C. Feldstein, M. (1988). The impact of stress and temperament on medical
utilization by school-age children. Journal of Pediatric Psychology, 13, 409-421.