Managed Care Organization Characteristics and Outpatient Specialty Care Use Among Children With Chronic Illness
http://www.100md.com
《小儿科》
Department of Epidemiology and Health Policy Research
Department of Pediatrics
Department of Biostatistics, College of Medicine
Institute for Child Health Policy, University of Florida, Gainesville, Florida
ABSTRACT
Background. Limited information is available about managed care organization (MCO) characteristics that influence outpatient physician specialist use among children with chronic conditions.
Objective. To examine the association between MCO characteristics and outpatient physician specialist use among children with chronic conditions who were receiving care in MCOs in which primary care providers (PCPs) served as gatekeepers for referrals and who were publicly insured.
Design and Methods. A total of 2333 children who had been diagnosed with a chronic condition and had functional limitations, an increased need for or use of health care services beyond what children normally use, and/or dependence on medications or home medical equipment were included in the study. The odds of an outpatient physician specialist visit 1 year after study entry were examined as a function of child health and sociodemographic characteristics, MCO characteristics, the child’s prior specialty care use, and provider availability in the MCO service delivery area.
Results. Children cared for in MCOs with lower percentages of PCPs paid on a fee-for-service basis (odds ratio: 0.95; 95% confidence interval: 0.92–0.98), with higher percentages of pediatricians in the PCP network (odds ratio: 1.17; 95% confidence interval: 1.07–1.29), and offering financial incentives for meeting quality of care standards (odds ratio: 1.71; 95% confidence interval: 1.28–2.29) had greater odds of outpatient physician specialist visits. Black children had odds of specialty care that were approximately one half those of white children. Children with prior physician specialist use were 52% more likely to have a physician specialist visit in the year after study entry. The children’s diagnoses and condition consequences were not related significantly to the odds of a specialty visit.
Conclusions. Specific MCO characteristics were associated with greater specialty care use among a group of low-income children with chronic conditions. Such information should be used to improve the structure of managed care arrangements for these vulnerable children.
Key Words: children chronic conditions pediatric specialty care use managed care
Abbreviations: MCO, managed care organization PCP, primary care provider SCHIP, State Children’s Health Insurance Program FPL, federal poverty level ICD-9, International Classification of Diseases, Ninth Revision ADD, attention-deficit disorder ADHD, attention-deficit/hyperactivity disorder QuICCC, Questionnaire for Identifying Children With Chronic Conditions
Specialty care is important for many children with chronic conditions, but access to such care may be constrained within managed care environments.1 The use of primary care providers (PCPs) as gatekeepers for managed care organizations (MCOs) is one commonly used strategy to control specialty care use.2,3 Studies of the impact of gatekeeping on children’s receipt of specialty care have resulted in mixed findings. Some studies found more specialty care use in gatekeeping MCOs, compared with nongatekeeping MCOs.1,4 Other researchers found that the replacement of a gatekeeping system with an open-access model increased specialty visits among a group of children with chronic conditions.5 Although the focus on gatekeeping in general yields some important information, MCOs use many other strategies concomitantly with their PCP gatekeepers, such as capitated payments, financial incentives, and prior authorization procedures, which were not well addressed in those studies. The use of these concomitant strategies may or may not meet the unique needs of children with chronic conditions, including their need for specialty physician care.6
The purpose of our study was to examine the association between MCO characteristics and outpatient physician specialist use among a cohort of children with chronic conditions who were all receiving care in MCOs in which PCPs served as gatekeepers and who were insured through a State Children’s Health Insurance Program (SCHIP) in which all MCOs were required to provide the same benefit package and copayment structure but could use different organizational strategies to deliver the children’s health care. We selected MCO characteristics for inclusion in this study on the basis of their potential association with the receipt of specialty care, including prior authorization procedures, the availability of PCPs who are pediatricians in the networks, PCP reimbursement, and the use of financial incentives to promote meeting pediatric quality of care standards. Important covariates that were found in other studies to influence physician specialist use, such as the children’s sociodemographic characteristics, health status, and prior use of physician specialty care and provider availability in their geographic areas, also were included.7–9
This study was designed to examine specialty care use among groups of children with chronic conditions and was not designed to be a study of asthma, diabetes mellitus, or any other single chronic condition. This approach allowed us to study children with more rare chronic conditions, such as cystic fibrosis, spina bifida, and others, by grouping them according to their condition consequences. Children with diverse conditions share more commonalities when grouped according to the consequences of their conditions, such as limitations in functioning or increased use of health care services, than when grouped according to diagnoses.10 Most MCOs care for children with common chronic conditions such as asthma, and many studies have been conducted documenting the quality of care these children receive. However, MCOs also care for children with more rare conditions and little is known about their health care quality. Grouping children on the basis of condition consequences provides one option for examining the care these children receive in managed care settings.
METHODS
Study Setting
This study was conducted with enrollees in Florida’s SCHIP; children between the ages of 5 and 19 years are eligible to enroll. At the time the study began in 1999, families below 200% of the federal poverty level (FPL) were offered subsidized premiums of $15.00 per month, regardless of the number of children enrolled. Families above 200% of FPL could enroll their children but paid the full premium price of approximately $85.00 per child per month.
Commercially licensed health plans were selected through a competitive bid process to form provider networks and to deliver care in counties throughout the state. At the time of this study, only 1 MCO was available in each county. However, the MCOs operated in >1 county, so that 8 MCOs covered 24 Florida counties.
The benefit package covered preventive care with no copayment and other outpatient care, inpatient care, rehabilitative services, mental health care, and emergency services with minimal copayments. Once the children were enrolled, families were required to select a PCP. The SCHIP standards required that the PCP be a board-certified pediatrician or board-certified family practitioner. The MCOs used varying reimbursement, prior authorization, and financial incentive strategies with their PCPs. However, none of them withheld payments on the basis of specialty referrals.
Data Sources
Five data sources were used, namely, child-level enrollment information, child-level health care claims/encounter data, parent telephone survey data, MCO administrator interview data, and county-specific information about the number of pediatricians, the presence of children’s hospitals or academic health centers serving children, and the number of children <18 years of age. The enrollment files contained information about the child’s age, gender, and family income and the number of months the child was enrolled in the program. The person-level claims/encounter data contained Physician’s Current Procedural Terminology codes and International Classification of Diseases, Ninth Revision (ICD-9) codes. Health care claims/encounter data from January 1 through December 31, 1999, were used to identify children with chronic conditions for possible inclusion in the study. Telephone surveys were conducted in 2000 to determine whether the children were experiencing any consequences from their conditions. Only those with condition consequences were included in the final study sample.
Claims/encounter data from January 1, 1999, through July 31, 2001, were used to determine whether each child had an outpatient physician specialist visit in the year before entering the study and in the year after entering the study. Data from in-depth face-to-face interviews conducted with MCO administrators were also used. Information from the Area Resource File for 2001, a county-specific, health resources information system, was used to quantify the number of pediatricians and the number of children’s hospitals or academic health centers serving children in the counties in which the MCOs were operating.11 The number of children <18 years of age residing in the county was obtained from census data and combined with the information about pediatrician availability to create a rate of pediatricians per 1000 children. This information was used as a measure of pediatrician and pediatric specialist availability in the children’s service delivery areas.
Measures Used
Identifying Children With Chronic Conditions
Children with chronic conditions were identified in 2 steps, with (1) diagnoses found in claims and encounter data and (2) parent reports about the consequences the child was experiencing from his or her condition.
Step 1
Children were identified initially for study participation through searches of the claims/encounter data for ICD-9 codes that reflect chronic conditions. A panel of 3 physicians at an academic health center developed this list, which contained high-prevalence/low-severity conditions such as asthma and low-prevalence/high-severity conditions such as certain cardiac anomalies. An expert at the National Association of Children’s Hospitals and Related Institutions reviewed the list, and refinements were made on the basis of that review. Two occurrences of the condition were required in most instances, to eliminate rule-out diagnoses. The final list contained 2320 conditions and has been used in other pediatric studies.12,13
Step 2
The Questionnaire for Identifying Children With Chronic Conditions (QuICCC) was administered to parents whose children had ICD-9 codes indicative of a chronic condition, to determine whether the children had a biological, psychologic, or cognitive disorder with a duration of 12 months and consequences of the disorder.14 The QuICCC identifies 3 types of consequences, ie, (1) functional limitations, (2) dependence on compensatory mechanisms or assistance, such as medications or home medical equipment, and (3) service use or need beyond routine care. The 39-item QuICCC was tested with different populations and identifies children with conditions traditionally thought of as chronic. Children with 1 consequences were included in the study.
Managed Care Administrator Interview Guide
The Managed Care Administrator Interview Guide was developed to assess strategies that MCOs use to organize care for children with chronic conditions. The following topics were addressed: (1) prior authorization strategies, including whether children with chronic conditions were exempt from prior authorization procedures for outpatient services; (2) provider network characteristics (percentage of PCPs who were pediatricians, as opposed to family practitioners); (3) percentage of PCPs reimbursed on a fee-for-service basis, capitated, or salaried; and (4) use of financial incentives for meeting quality of care indicators, such as well-child visits, immunizations, and other MCO-required standards of care. All interviews were face to face and were conducted over 2 days with a minimum of 5 different administrators, including the chief financial officer, the medical director, the nurse coordinator(s), the director for network development, and the director of utilization review and quality assurance.
The MCOs were given a copy of the interview guide before the face-to-face interviews and were asked to prepare written responses. The face-to-face interview was used to verify and clarify the written responses. Two members of the project team, who were experienced in in-depth interviewing techniques, conducted the interviews, and their notes were transcribed. The responses were coded into categories when possible or used to calculate percentages of pediatricians in the networks and percentages of PCPs receiving different types of reimbursements. For example, respondents were asked if they waived prior authorization requirements for outpatient services for children with chronic conditions, and the response for each MCO was coded as 1 = yes and 0 = no. This same process was used to categorize the use of financial incentives. Information about the number of PCPs and their specialties were tallied and used to calculate the percentages of pediatricians in the PCP networks of the MCOs. Finally, the MCOs provided the percentages of PCPs paid with capitation, fee-for-service, and salary approaches.
Sample Selection Procedures
Health Science Center institutional review board approval was obtained. The enrollment files were examined to ensure that only 1 child per family was considered for possible inclusion in the study. Then, with claims/encounter data from January 1 to December 31, 1999, 4678 children (8%) were identified as potentially having a chronic condition, on the basis of their diagnoses, and were enrolled currently in the program. Of the 4678 children who had chronic condition diagnoses in the claims and encounter data, 3695 (79%) were located and their parents agreed to participate in an interview with the use of the QuICCC. Of these, 2386 children (65%) were identified as experiencing consequences of their conditions and 2333 had complete enrollment and claims/encounter information and were included in these analyses.
Study Variables
The outcome variable was the odds of an outpatient physician specialist visit. Predictor variables included each child’s sociodemographic characteristics, ie, age, gender, family income expressed as a percentage of the FPL (150% of FPL or >150% of FPL), race, and ethnicity, and the number of months the child was enrolled in the program in the year after the QuICCC survey. Children’s health was categorized according to whether they were experiencing 1, 2, or 3 types of consequences as a result of the chronic condition. The following diagnostic groups also were used: asthma, diabetes mellitus, attention-deficit disorder (ADD)/attention-deficit/hyperactivity disorder (ADHD), and all other chronic condition diagnoses. There were sufficient numbers of children with asthma, diabetes, and ADD/ADHD to include these diagnoses in the models. There were insufficient numbers of the other diagnoses to include them individually, however, and they were grouped into a category of other chronic conditions. An indicator of whether the child had a specialty care visit in the year before study entry also was included.
Four MCO characteristics were included in the analyses. Previous studies suggested that children residing in geographic areas with fewer pediatricians have more unmet specialty care needs.15 Therefore, we included in our analysis the percentages of the PCP networks of the MCOs that were composed of pediatricians. In addition, stringent prior authorization procedures may inhibit children’s access to physician specialty services. In our study, some MCOs waived prior authorization requirements for outpatient services for children with chronic conditions, and we examined this strategy. PCP reimbursement is a critical issue related to the use of services in general and may have a specific impact on the receipt of specialty care.16,17 Therefore, we included information about the MCOs’ PCP reimbursement strategies. Finally, some MCOs give their PCPs financial incentives to promote meeting pediatric quality of care standards.18 These standards typically address preventive care and parent satisfaction and are not focused specifically on specialty care. However, the use of this financial strategy may reflect the MCOs’ interest in addressing the needs of its pediatric enrollees, including the need for specialty care, and information about its use was included in our analyses. Finally, the percentage of pediatricians per 1000 children in the county and an indicator of the presence of a children’s hospital or academic health center was included.
Analyses
This study focused on MCO characteristics and not specific MCOs. More than one MCO used a particular strategy, so that no single strategy was confounded by either a MCO or a geographic location. In this analysis, the effects of the MCO characteristics were taken into account by including 5 MCO characteristics in the model as covariates. The effects of selected local characteristics were taken into account by including the percentage of pediatricians per 1000 children at the county level and an indicator for the presence of a children’s hospital or academic health center as covariates. Individual effects were taken into account by including demographic and health variables.
Logistic-regression analysis is used widely to investigate the relationship between a binary outcome and a set of explanatory variables.19 For binary response models, the response, Y, of an individual can take 1 of 2 possible values, denoted for convenience as 1 and 0 (for example, Y = 1 if there was an outpatient physician specialist visit for the child and otherwise Y = 0). If x is a vector of explanatory variables and p = Pr(Y = 1|x) is the response probability to be modeled, then the linear logistic model has the form logit(p) = log[p/(1 – p)] = + 'x, where is the intercept parameter and is the vector of slope parameters. We used SAS procedure PROC LOGISTIC (SAS software release 9.0; SAS Institute, Cary, NC) to fit this model with Y as the indicator for an outpatient specialist visit for the child and MCO characteristics as the explanatory variables, controlling for the child’s sociodemographic features, health characteristics, and specialty care use in the year before study entry and provider availability in the MCOs’ service areas.
We used the stepwise model selection method to produce the reduced model. In stepwise selection, an attempt is made to remove any insignificant variables from the model before a significant variable is added to the model, and this process is terminated if no additional effect can be added to the model or if the effect just entered into the model is the only effect removed in the subsequent backward elimination. A significance level of .1 is required to allow a variable into the model, and a significance level of .08 is required for a variable to remain in the model. The reduced model is not significantly different from the full model at the significance level of .05.
RESULTS
Sample Characteristics
The final study sample contained 2333 children, who were 11.9 ± 3.53 years of age. Sixty percent of the children were male and 40% were female (Table 1). Most children were white (76%) and non-Hispanic (75%). However, substantial proportions were black (13%) or of mixed race (11%) and 25% were Hispanic. Sixty-six percent of the children were below 150% of the FPL and 34% were above. Twenty-five percent of the study sample experienced only 1 consequence from their conditions, 37% had 2 consequences, and 38% had all 3 possible consequences (need for or actual elevated service use, dependence on compensatory devices, and functional limitations). Twenty-two percent of the children had asthma, 12% had ADD/ADHD, and 4% had diabetes. The remaining children (62%) had different chronic conditions, with no single condition being present for >3% of the children. These conditions included physical health conditions such as cystic fibrosis, epilepsy, leukemia, and spina bifida and behavioral health conditions such as depression, schizophrenia, and oppositional disorders. Fifty-eight percent of children had 1 chronic condition diagnosis, and 42% had 2 chronic condition diagnoses. Finally, 15% of the children had an outpatient specialist visit in the year after the interview. However, the percentages of children with a specialty visit after the interview varied from 6% to 21%, depending on the MCO characteristics.
Multivariate Results
The children’s diagnoses were not related significantly to the odds of an outpatient physician specialist visit and were not retained in the reduced model. In the full model, children with 3 condition consequences were 1.4 times more likely to have an outpatient physician specialist visit than were children with 1 condition consequence. However, no significant differences in outpatient physician specialist visits were observed between those with 2 condition consequences and those with 1, and the number of condition consequences was not retained in the reduced model.
Three of the 5 MCO characteristics were related significantly to the odds of a child having an outpatient physician specialist visit and were retained in the reduced model. The higher the percentage of pediatricians in the PCP network, the greater were the odds of a child having an outpatient physician specialist visit (P = .001). For each 10% increase in the percentage of pediatricians in the PCP network, the child’s odds of having an outpatient physician specialist visit increased by 17.1%. Similarly, the higher the percentage of PCPs in the network paid on a fee-for-service basis, the lower were the odds of a child having an outpatient physician specialist visit (P = .003). For example, for each 10% increase in the percentage of PCPs receiving fee-for-service payments, the odds of an outpatient physician specialist visit decreased by 5%. Children cared for in MCOs that offered a bonus to PCPs for meeting quality of care standards were 71% more likely to have an outpatient physician specialist visit than were those not cared for in such settings (P = .0003).
The presence of a children’s hospital or an academic health center in the MCOs’ service delivery areas was not significant in the full model and was not retained in the reduced model. Similarly, the number of pediatricians per 1000 children in the counties in which the MCOs were operating was not significant and was not included in the final model.
Finally, children’s use of physician outpatient specialist services in the year before study entry was significantly related to the use of such services in the year after study entry (P = .005). Children who saw a physician specialist in the year before the interview were 52% more likely than those who did not to see a physician specialist in the subsequent year.
DISCUSSION
This study examined the association between MCO characteristics and the receipt of outpatient physician specialist care among a group of children with chronic conditions who were experiencing consequences from those conditions. The uniformity in our study, in terms of the required use of PCPs as gatekeepers, the benefit package design, the copayment structure, and the lack of financial withholds for specialty referrals, allowed us to focus on the MCOs’ provider network composition and varying uses of primary reimbursement strategies, financial incentives for quality standards, and outpatient prior authorization procedures.
In our study, a higher percentage of PCPs receiving fee-for-service payments, compared with capitation, was associated with lower odds of physician specialty use. Conversely, this finding indicated that greater use of capitation was associated with greater odds of physician specialty use. The findings from other researchers regarding capitated payments to PCPs and specialty referral rates were mixed, with some reporting increased referrals and others finding no difference in referrals between PCPs who were paid on a capitated basis and those who were not.17,20 It is possible that PCPs paid on a capitated basis are more aware of financial constraints than are those paid on a fee-for-service basis and are less willing to treat a particular health problem when discretion exists regarding the necessity for a referral.
Children cared for in MCOs with higher percentages of pediatricians, compared with family practitioners, in the network also had greater odds of specialty referrals. A higher concentration of pediatricians in a network could reflect better PCP and specialist availability. This is unlikely, however, because we included measures of provider availability in the children’s counties, and these variables were not significant. Another possible explanation for this finding is selection bias. At the time of this study, families could not choose between MCOs on the basis of the provider network or other factors of interest because only 1 MCO was available in each county or group of counties representing a service delivery area. However, families could choose whether to enroll their children in the SCHIP. It is possible that families whose children had greater health care needs were more likely to enroll them in the SCHIP if the MCOs provided good access to pediatricians. Although all of the children in this study had chronic conditions and we controlled for condition consequences and specific conditions in our analyses, unmeasured health status characteristics could partially explain this finding.
Finally, other studies demonstrated that pediatricians’ and family practitioners’ behaviors vary with respect to the resources used when caring for children. For example, pediatricians are more likely than family practitioners to order a sepsis evaluation for a 6-week-old infant with a temperature of 38.3°C21 or to make referrals for depression or other psychiatric disorders.22 Pediatricians may have a higher index of suspicion when working with chronically ill children, because of their in-depth pediatric training, which makes them more likely to order specialty referrals, compared with family practitioners.
MCOs’ use of financial incentives for meeting quality of care standards was associated with increased odds of outpatient physician specialty use. The quality of care standards focused on preventive care and family satisfaction, not specialty care. However, the use of financial incentives focused on pediatric quality of care standards might have fostered an increased emphasis on meeting children’s health care needs, including the need for specialty referrals for this cohort of chronically ill children.
Factors other than the MCO characteristics were associated with the children’s odds of specialty care. Black children were approximately one half as likely as white children to receive an outpatient physician specialist visit, even after consideration of other important covariates in the model. This finding has been documented in other studies. For example, black and Hispanic children use less health care than white children, in part because of a lack of health insurance.7 However, insured black and Hispanic children receive less care overall, including specialty care, than do insured white children, even when income and health status are considered.8,9
Children’s prior use of specialty services was an important predictor of their future use of such services. The influence of prior health care use on future health care use was expected and was well documented in other studies.23 The number of condition consequences the children had and their specific conditions were not associated significantly with the odds of outpatient physician specialist use. However, this study focused on a cohort of children who had chronic conditions that were identified through diagnoses found in claims and encounter data and who had at least 1 consequence from their conditions. It is possible that, after the initial diagnostic and condition consequence screening was conducted, the additional health status information we had available to us was not refined sufficiently to explain their use of outpatient specialist care.
There are some study limitations. First, MCOs have complex organizational structures that were addressed only partially through our study variables. For example, the MCOs were able to provide only limited information about their specialty networks, such as the overall numbers of specialists for both adult and child enrollees. Because more specific information about the number of specialists available to pediatric enrollees was not available, it was not used in our analyses. Such information should be used in future studies.
Second, although studying children grouped by condition consequences provides important information that otherwise would be difficult to obtain about those with rare conditions, it does not provide clear benchmarks for assessing whether the specialty care was needed. Although we were able to determine that children cared for by pediatricians had more specialty use and black children used fewer specialty services, even after consideration of diagnoses and condition consequences, we could not determine whether the use was appropriate. It is possible that pediatricians in this study made unnecessary referrals and white children overused services, but this cannot be determined from our study.
Third, we included some relevant, county-level, health care delivery system measures in our analyses. Although these factors were not significant, it is possible that unmeasured community factors influenced the outcomes.
Despite these limitations, this study provides useful information about child-specific and MCO-related factors that contribute to outpatient physician specialist use among a group of low-income children with chronic conditions. The findings regarding fee-for-service and capitated payments and specialty referrals support additional exploration of blended payment systems, which use capitation for some services and fee-for-service supplements for others.24 Blended payments may encourage capitated PCPs to manage chronic conditions more completely, rather than making specialty referrals that may be discretionary. Providing financial incentives for meeting MCO-specific quality of care standards and ensuring good access to pediatricians in the PCP network are also important for the receipt of specialty care. Our study suggests that MCOs using these strategies would provide better access to specialty care for chronically ill children. However, additional studies examining MCO characteristics and specialty care use in the context of specific clinical situations are needed.
ACKNOWLEDGMENTS
This work was funded by the Agency for Healthcare Research and Quality, the American Association of Health Plans Foundation, and the Health Resources and Services Administration, Maternal and Child Health Bureau (grant U01 HS09949-02).
FOOTNOTES
Accepted Sep 24, 2004.
No conflict of interest declared.
REFERENCES
Forrest CB, Glade GB, Starfield B, et al. Gatekeeping and referral of children and adolescents to specialty care. Pediatrics. 1999;104 :28 –34
Curtis D, Kaye N, Riley T. Transitioning to Medicaid Managed Care: Children With Special Health Care Needs. Portland, ME: National Academy for State Health Policy; 1999
Szilagyi PG, Roghmann KJ, Foye HR, et al. The effect of independent practice association plans on use of pediatric ambulatory medical care in one group practice. JAMA. 1990;263 :2198 –2203
Ferris TG, Chang Y, Perrin JM, Blumenthal D, Pearson SD. Effects of removing gatekeeping on specialist utilization by children in a health maintenance organization. Arch Pediatr Adolesc Med. 2002;156 :574 –579
Simpson L, Fraser I. Children and managed care: what research can, can’t, and should tell us about impact. Med Care Res Rev. 1999;56 :13 –36
Shenkman E, Wu SS, Nackashi J, Sherman J. Managed care organizational characteristics and health care use among children with special health care needs. Health Serv Res. 2003;38 :1599 –1624
Shenkman E, Pendergast J, Reiss J, Walther E, Bucciarelli R, Freedman S. The school enrollment-based health insurance program: socioeconomic factors in enrollees’ use of health services. Am J Public Health. 1996;86 :1791 –1793
Stein REK, Jessop DJ. What diagnosis does not tell: the case for a noncategorical approach to chronic illness in childhood. Soc Sci Med. 1989;29 :769 –778
Shenkman E, Vogel B, Boyett J, Naff R. Disenrollment and re-enrollment patterns in a SCHIP. Health Care Financ Rev. 2002;23 :47 –63
Youngblade LE, Col J, Shenkman E. Adolescents’ health care use in a Title XXI program. J Adolesc Health. 2002;30 :262 –273
Stein RE, Westbrook LE, Bauman LJ. The Questionnaire for Identifying Children With Chronic Conditions: a measure based on a noncategorical approach. Pediatrics. 1997;99 :513 –521
Perrin JM, Kuhlthau K, Walker DK, Stein REK, Newacheck PW, Gortmaker SL. Monitoring health care for children with chronic conditions in a managed care environment. Matern Child Health J. 1997;1 :15 –23
Forrest CB, Nutting P, Werner JJ, et al. Managed health plans effects on the specialty referral process: results from the Ambulatory Sentinel Practice Network referral study. Med Care. 2003;41 :242 –253
Stern AL. Modeling the future of health care. Health Plan. 2001;Sep/Oct:15–18
Agresti A. Categorical Data Analysis. New York, NY: John Wiley & Sons; 2002
Iversen T, Luras H. The effect of capitation on GPs’ referral decisions. Health Econ. 2000;9 :199 –210
Boulis AK, Long J. Variation in the treatment of children by primary care physician specialty. Arch Pediatr Adolesc Med. 2002:156 :1210 –1215
Rushton JL, Clark SJ, Freed GL. Primary care role in the management of childhood depression: a comparison of pediatricians and family physicians. Pediatrics. 2000;105 (4 pt 2):957 –962
Newhouse JP. Pricing the Priceless: A Health Care Conundrum. Cambridge, MA: Massachusetts Institute of Technology; 2002
Robinson JC. Theory and practice in the design of physician payment incentives. Milbank Q. 2001;79 :149 –177(Elizabeth Shenkman, PhD, )
Department of Pediatrics
Department of Biostatistics, College of Medicine
Institute for Child Health Policy, University of Florida, Gainesville, Florida
ABSTRACT
Background. Limited information is available about managed care organization (MCO) characteristics that influence outpatient physician specialist use among children with chronic conditions.
Objective. To examine the association between MCO characteristics and outpatient physician specialist use among children with chronic conditions who were receiving care in MCOs in which primary care providers (PCPs) served as gatekeepers for referrals and who were publicly insured.
Design and Methods. A total of 2333 children who had been diagnosed with a chronic condition and had functional limitations, an increased need for or use of health care services beyond what children normally use, and/or dependence on medications or home medical equipment were included in the study. The odds of an outpatient physician specialist visit 1 year after study entry were examined as a function of child health and sociodemographic characteristics, MCO characteristics, the child’s prior specialty care use, and provider availability in the MCO service delivery area.
Results. Children cared for in MCOs with lower percentages of PCPs paid on a fee-for-service basis (odds ratio: 0.95; 95% confidence interval: 0.92–0.98), with higher percentages of pediatricians in the PCP network (odds ratio: 1.17; 95% confidence interval: 1.07–1.29), and offering financial incentives for meeting quality of care standards (odds ratio: 1.71; 95% confidence interval: 1.28–2.29) had greater odds of outpatient physician specialist visits. Black children had odds of specialty care that were approximately one half those of white children. Children with prior physician specialist use were 52% more likely to have a physician specialist visit in the year after study entry. The children’s diagnoses and condition consequences were not related significantly to the odds of a specialty visit.
Conclusions. Specific MCO characteristics were associated with greater specialty care use among a group of low-income children with chronic conditions. Such information should be used to improve the structure of managed care arrangements for these vulnerable children.
Key Words: children chronic conditions pediatric specialty care use managed care
Abbreviations: MCO, managed care organization PCP, primary care provider SCHIP, State Children’s Health Insurance Program FPL, federal poverty level ICD-9, International Classification of Diseases, Ninth Revision ADD, attention-deficit disorder ADHD, attention-deficit/hyperactivity disorder QuICCC, Questionnaire for Identifying Children With Chronic Conditions
Specialty care is important for many children with chronic conditions, but access to such care may be constrained within managed care environments.1 The use of primary care providers (PCPs) as gatekeepers for managed care organizations (MCOs) is one commonly used strategy to control specialty care use.2,3 Studies of the impact of gatekeeping on children’s receipt of specialty care have resulted in mixed findings. Some studies found more specialty care use in gatekeeping MCOs, compared with nongatekeeping MCOs.1,4 Other researchers found that the replacement of a gatekeeping system with an open-access model increased specialty visits among a group of children with chronic conditions.5 Although the focus on gatekeeping in general yields some important information, MCOs use many other strategies concomitantly with their PCP gatekeepers, such as capitated payments, financial incentives, and prior authorization procedures, which were not well addressed in those studies. The use of these concomitant strategies may or may not meet the unique needs of children with chronic conditions, including their need for specialty physician care.6
The purpose of our study was to examine the association between MCO characteristics and outpatient physician specialist use among a cohort of children with chronic conditions who were all receiving care in MCOs in which PCPs served as gatekeepers and who were insured through a State Children’s Health Insurance Program (SCHIP) in which all MCOs were required to provide the same benefit package and copayment structure but could use different organizational strategies to deliver the children’s health care. We selected MCO characteristics for inclusion in this study on the basis of their potential association with the receipt of specialty care, including prior authorization procedures, the availability of PCPs who are pediatricians in the networks, PCP reimbursement, and the use of financial incentives to promote meeting pediatric quality of care standards. Important covariates that were found in other studies to influence physician specialist use, such as the children’s sociodemographic characteristics, health status, and prior use of physician specialty care and provider availability in their geographic areas, also were included.7–9
This study was designed to examine specialty care use among groups of children with chronic conditions and was not designed to be a study of asthma, diabetes mellitus, or any other single chronic condition. This approach allowed us to study children with more rare chronic conditions, such as cystic fibrosis, spina bifida, and others, by grouping them according to their condition consequences. Children with diverse conditions share more commonalities when grouped according to the consequences of their conditions, such as limitations in functioning or increased use of health care services, than when grouped according to diagnoses.10 Most MCOs care for children with common chronic conditions such as asthma, and many studies have been conducted documenting the quality of care these children receive. However, MCOs also care for children with more rare conditions and little is known about their health care quality. Grouping children on the basis of condition consequences provides one option for examining the care these children receive in managed care settings.
METHODS
Study Setting
This study was conducted with enrollees in Florida’s SCHIP; children between the ages of 5 and 19 years are eligible to enroll. At the time the study began in 1999, families below 200% of the federal poverty level (FPL) were offered subsidized premiums of $15.00 per month, regardless of the number of children enrolled. Families above 200% of FPL could enroll their children but paid the full premium price of approximately $85.00 per child per month.
Commercially licensed health plans were selected through a competitive bid process to form provider networks and to deliver care in counties throughout the state. At the time of this study, only 1 MCO was available in each county. However, the MCOs operated in >1 county, so that 8 MCOs covered 24 Florida counties.
The benefit package covered preventive care with no copayment and other outpatient care, inpatient care, rehabilitative services, mental health care, and emergency services with minimal copayments. Once the children were enrolled, families were required to select a PCP. The SCHIP standards required that the PCP be a board-certified pediatrician or board-certified family practitioner. The MCOs used varying reimbursement, prior authorization, and financial incentive strategies with their PCPs. However, none of them withheld payments on the basis of specialty referrals.
Data Sources
Five data sources were used, namely, child-level enrollment information, child-level health care claims/encounter data, parent telephone survey data, MCO administrator interview data, and county-specific information about the number of pediatricians, the presence of children’s hospitals or academic health centers serving children, and the number of children <18 years of age. The enrollment files contained information about the child’s age, gender, and family income and the number of months the child was enrolled in the program. The person-level claims/encounter data contained Physician’s Current Procedural Terminology codes and International Classification of Diseases, Ninth Revision (ICD-9) codes. Health care claims/encounter data from January 1 through December 31, 1999, were used to identify children with chronic conditions for possible inclusion in the study. Telephone surveys were conducted in 2000 to determine whether the children were experiencing any consequences from their conditions. Only those with condition consequences were included in the final study sample.
Claims/encounter data from January 1, 1999, through July 31, 2001, were used to determine whether each child had an outpatient physician specialist visit in the year before entering the study and in the year after entering the study. Data from in-depth face-to-face interviews conducted with MCO administrators were also used. Information from the Area Resource File for 2001, a county-specific, health resources information system, was used to quantify the number of pediatricians and the number of children’s hospitals or academic health centers serving children in the counties in which the MCOs were operating.11 The number of children <18 years of age residing in the county was obtained from census data and combined with the information about pediatrician availability to create a rate of pediatricians per 1000 children. This information was used as a measure of pediatrician and pediatric specialist availability in the children’s service delivery areas.
Measures Used
Identifying Children With Chronic Conditions
Children with chronic conditions were identified in 2 steps, with (1) diagnoses found in claims and encounter data and (2) parent reports about the consequences the child was experiencing from his or her condition.
Step 1
Children were identified initially for study participation through searches of the claims/encounter data for ICD-9 codes that reflect chronic conditions. A panel of 3 physicians at an academic health center developed this list, which contained high-prevalence/low-severity conditions such as asthma and low-prevalence/high-severity conditions such as certain cardiac anomalies. An expert at the National Association of Children’s Hospitals and Related Institutions reviewed the list, and refinements were made on the basis of that review. Two occurrences of the condition were required in most instances, to eliminate rule-out diagnoses. The final list contained 2320 conditions and has been used in other pediatric studies.12,13
Step 2
The Questionnaire for Identifying Children With Chronic Conditions (QuICCC) was administered to parents whose children had ICD-9 codes indicative of a chronic condition, to determine whether the children had a biological, psychologic, or cognitive disorder with a duration of 12 months and consequences of the disorder.14 The QuICCC identifies 3 types of consequences, ie, (1) functional limitations, (2) dependence on compensatory mechanisms or assistance, such as medications or home medical equipment, and (3) service use or need beyond routine care. The 39-item QuICCC was tested with different populations and identifies children with conditions traditionally thought of as chronic. Children with 1 consequences were included in the study.
Managed Care Administrator Interview Guide
The Managed Care Administrator Interview Guide was developed to assess strategies that MCOs use to organize care for children with chronic conditions. The following topics were addressed: (1) prior authorization strategies, including whether children with chronic conditions were exempt from prior authorization procedures for outpatient services; (2) provider network characteristics (percentage of PCPs who were pediatricians, as opposed to family practitioners); (3) percentage of PCPs reimbursed on a fee-for-service basis, capitated, or salaried; and (4) use of financial incentives for meeting quality of care indicators, such as well-child visits, immunizations, and other MCO-required standards of care. All interviews were face to face and were conducted over 2 days with a minimum of 5 different administrators, including the chief financial officer, the medical director, the nurse coordinator(s), the director for network development, and the director of utilization review and quality assurance.
The MCOs were given a copy of the interview guide before the face-to-face interviews and were asked to prepare written responses. The face-to-face interview was used to verify and clarify the written responses. Two members of the project team, who were experienced in in-depth interviewing techniques, conducted the interviews, and their notes were transcribed. The responses were coded into categories when possible or used to calculate percentages of pediatricians in the networks and percentages of PCPs receiving different types of reimbursements. For example, respondents were asked if they waived prior authorization requirements for outpatient services for children with chronic conditions, and the response for each MCO was coded as 1 = yes and 0 = no. This same process was used to categorize the use of financial incentives. Information about the number of PCPs and their specialties were tallied and used to calculate the percentages of pediatricians in the PCP networks of the MCOs. Finally, the MCOs provided the percentages of PCPs paid with capitation, fee-for-service, and salary approaches.
Sample Selection Procedures
Health Science Center institutional review board approval was obtained. The enrollment files were examined to ensure that only 1 child per family was considered for possible inclusion in the study. Then, with claims/encounter data from January 1 to December 31, 1999, 4678 children (8%) were identified as potentially having a chronic condition, on the basis of their diagnoses, and were enrolled currently in the program. Of the 4678 children who had chronic condition diagnoses in the claims and encounter data, 3695 (79%) were located and their parents agreed to participate in an interview with the use of the QuICCC. Of these, 2386 children (65%) were identified as experiencing consequences of their conditions and 2333 had complete enrollment and claims/encounter information and were included in these analyses.
Study Variables
The outcome variable was the odds of an outpatient physician specialist visit. Predictor variables included each child’s sociodemographic characteristics, ie, age, gender, family income expressed as a percentage of the FPL (150% of FPL or >150% of FPL), race, and ethnicity, and the number of months the child was enrolled in the program in the year after the QuICCC survey. Children’s health was categorized according to whether they were experiencing 1, 2, or 3 types of consequences as a result of the chronic condition. The following diagnostic groups also were used: asthma, diabetes mellitus, attention-deficit disorder (ADD)/attention-deficit/hyperactivity disorder (ADHD), and all other chronic condition diagnoses. There were sufficient numbers of children with asthma, diabetes, and ADD/ADHD to include these diagnoses in the models. There were insufficient numbers of the other diagnoses to include them individually, however, and they were grouped into a category of other chronic conditions. An indicator of whether the child had a specialty care visit in the year before study entry also was included.
Four MCO characteristics were included in the analyses. Previous studies suggested that children residing in geographic areas with fewer pediatricians have more unmet specialty care needs.15 Therefore, we included in our analysis the percentages of the PCP networks of the MCOs that were composed of pediatricians. In addition, stringent prior authorization procedures may inhibit children’s access to physician specialty services. In our study, some MCOs waived prior authorization requirements for outpatient services for children with chronic conditions, and we examined this strategy. PCP reimbursement is a critical issue related to the use of services in general and may have a specific impact on the receipt of specialty care.16,17 Therefore, we included information about the MCOs’ PCP reimbursement strategies. Finally, some MCOs give their PCPs financial incentives to promote meeting pediatric quality of care standards.18 These standards typically address preventive care and parent satisfaction and are not focused specifically on specialty care. However, the use of this financial strategy may reflect the MCOs’ interest in addressing the needs of its pediatric enrollees, including the need for specialty care, and information about its use was included in our analyses. Finally, the percentage of pediatricians per 1000 children in the county and an indicator of the presence of a children’s hospital or academic health center was included.
Analyses
This study focused on MCO characteristics and not specific MCOs. More than one MCO used a particular strategy, so that no single strategy was confounded by either a MCO or a geographic location. In this analysis, the effects of the MCO characteristics were taken into account by including 5 MCO characteristics in the model as covariates. The effects of selected local characteristics were taken into account by including the percentage of pediatricians per 1000 children at the county level and an indicator for the presence of a children’s hospital or academic health center as covariates. Individual effects were taken into account by including demographic and health variables.
Logistic-regression analysis is used widely to investigate the relationship between a binary outcome and a set of explanatory variables.19 For binary response models, the response, Y, of an individual can take 1 of 2 possible values, denoted for convenience as 1 and 0 (for example, Y = 1 if there was an outpatient physician specialist visit for the child and otherwise Y = 0). If x is a vector of explanatory variables and p = Pr(Y = 1|x) is the response probability to be modeled, then the linear logistic model has the form logit(p) = log[p/(1 – p)] = + 'x, where is the intercept parameter and is the vector of slope parameters. We used SAS procedure PROC LOGISTIC (SAS software release 9.0; SAS Institute, Cary, NC) to fit this model with Y as the indicator for an outpatient specialist visit for the child and MCO characteristics as the explanatory variables, controlling for the child’s sociodemographic features, health characteristics, and specialty care use in the year before study entry and provider availability in the MCOs’ service areas.
We used the stepwise model selection method to produce the reduced model. In stepwise selection, an attempt is made to remove any insignificant variables from the model before a significant variable is added to the model, and this process is terminated if no additional effect can be added to the model or if the effect just entered into the model is the only effect removed in the subsequent backward elimination. A significance level of .1 is required to allow a variable into the model, and a significance level of .08 is required for a variable to remain in the model. The reduced model is not significantly different from the full model at the significance level of .05.
RESULTS
Sample Characteristics
The final study sample contained 2333 children, who were 11.9 ± 3.53 years of age. Sixty percent of the children were male and 40% were female (Table 1). Most children were white (76%) and non-Hispanic (75%). However, substantial proportions were black (13%) or of mixed race (11%) and 25% were Hispanic. Sixty-six percent of the children were below 150% of the FPL and 34% were above. Twenty-five percent of the study sample experienced only 1 consequence from their conditions, 37% had 2 consequences, and 38% had all 3 possible consequences (need for or actual elevated service use, dependence on compensatory devices, and functional limitations). Twenty-two percent of the children had asthma, 12% had ADD/ADHD, and 4% had diabetes. The remaining children (62%) had different chronic conditions, with no single condition being present for >3% of the children. These conditions included physical health conditions such as cystic fibrosis, epilepsy, leukemia, and spina bifida and behavioral health conditions such as depression, schizophrenia, and oppositional disorders. Fifty-eight percent of children had 1 chronic condition diagnosis, and 42% had 2 chronic condition diagnoses. Finally, 15% of the children had an outpatient specialist visit in the year after the interview. However, the percentages of children with a specialty visit after the interview varied from 6% to 21%, depending on the MCO characteristics.
Multivariate Results
The children’s diagnoses were not related significantly to the odds of an outpatient physician specialist visit and were not retained in the reduced model. In the full model, children with 3 condition consequences were 1.4 times more likely to have an outpatient physician specialist visit than were children with 1 condition consequence. However, no significant differences in outpatient physician specialist visits were observed between those with 2 condition consequences and those with 1, and the number of condition consequences was not retained in the reduced model.
Three of the 5 MCO characteristics were related significantly to the odds of a child having an outpatient physician specialist visit and were retained in the reduced model. The higher the percentage of pediatricians in the PCP network, the greater were the odds of a child having an outpatient physician specialist visit (P = .001). For each 10% increase in the percentage of pediatricians in the PCP network, the child’s odds of having an outpatient physician specialist visit increased by 17.1%. Similarly, the higher the percentage of PCPs in the network paid on a fee-for-service basis, the lower were the odds of a child having an outpatient physician specialist visit (P = .003). For example, for each 10% increase in the percentage of PCPs receiving fee-for-service payments, the odds of an outpatient physician specialist visit decreased by 5%. Children cared for in MCOs that offered a bonus to PCPs for meeting quality of care standards were 71% more likely to have an outpatient physician specialist visit than were those not cared for in such settings (P = .0003).
The presence of a children’s hospital or an academic health center in the MCOs’ service delivery areas was not significant in the full model and was not retained in the reduced model. Similarly, the number of pediatricians per 1000 children in the counties in which the MCOs were operating was not significant and was not included in the final model.
Finally, children’s use of physician outpatient specialist services in the year before study entry was significantly related to the use of such services in the year after study entry (P = .005). Children who saw a physician specialist in the year before the interview were 52% more likely than those who did not to see a physician specialist in the subsequent year.
DISCUSSION
This study examined the association between MCO characteristics and the receipt of outpatient physician specialist care among a group of children with chronic conditions who were experiencing consequences from those conditions. The uniformity in our study, in terms of the required use of PCPs as gatekeepers, the benefit package design, the copayment structure, and the lack of financial withholds for specialty referrals, allowed us to focus on the MCOs’ provider network composition and varying uses of primary reimbursement strategies, financial incentives for quality standards, and outpatient prior authorization procedures.
In our study, a higher percentage of PCPs receiving fee-for-service payments, compared with capitation, was associated with lower odds of physician specialty use. Conversely, this finding indicated that greater use of capitation was associated with greater odds of physician specialty use. The findings from other researchers regarding capitated payments to PCPs and specialty referral rates were mixed, with some reporting increased referrals and others finding no difference in referrals between PCPs who were paid on a capitated basis and those who were not.17,20 It is possible that PCPs paid on a capitated basis are more aware of financial constraints than are those paid on a fee-for-service basis and are less willing to treat a particular health problem when discretion exists regarding the necessity for a referral.
Children cared for in MCOs with higher percentages of pediatricians, compared with family practitioners, in the network also had greater odds of specialty referrals. A higher concentration of pediatricians in a network could reflect better PCP and specialist availability. This is unlikely, however, because we included measures of provider availability in the children’s counties, and these variables were not significant. Another possible explanation for this finding is selection bias. At the time of this study, families could not choose between MCOs on the basis of the provider network or other factors of interest because only 1 MCO was available in each county or group of counties representing a service delivery area. However, families could choose whether to enroll their children in the SCHIP. It is possible that families whose children had greater health care needs were more likely to enroll them in the SCHIP if the MCOs provided good access to pediatricians. Although all of the children in this study had chronic conditions and we controlled for condition consequences and specific conditions in our analyses, unmeasured health status characteristics could partially explain this finding.
Finally, other studies demonstrated that pediatricians’ and family practitioners’ behaviors vary with respect to the resources used when caring for children. For example, pediatricians are more likely than family practitioners to order a sepsis evaluation for a 6-week-old infant with a temperature of 38.3°C21 or to make referrals for depression or other psychiatric disorders.22 Pediatricians may have a higher index of suspicion when working with chronically ill children, because of their in-depth pediatric training, which makes them more likely to order specialty referrals, compared with family practitioners.
MCOs’ use of financial incentives for meeting quality of care standards was associated with increased odds of outpatient physician specialty use. The quality of care standards focused on preventive care and family satisfaction, not specialty care. However, the use of financial incentives focused on pediatric quality of care standards might have fostered an increased emphasis on meeting children’s health care needs, including the need for specialty referrals for this cohort of chronically ill children.
Factors other than the MCO characteristics were associated with the children’s odds of specialty care. Black children were approximately one half as likely as white children to receive an outpatient physician specialist visit, even after consideration of other important covariates in the model. This finding has been documented in other studies. For example, black and Hispanic children use less health care than white children, in part because of a lack of health insurance.7 However, insured black and Hispanic children receive less care overall, including specialty care, than do insured white children, even when income and health status are considered.8,9
Children’s prior use of specialty services was an important predictor of their future use of such services. The influence of prior health care use on future health care use was expected and was well documented in other studies.23 The number of condition consequences the children had and their specific conditions were not associated significantly with the odds of outpatient physician specialist use. However, this study focused on a cohort of children who had chronic conditions that were identified through diagnoses found in claims and encounter data and who had at least 1 consequence from their conditions. It is possible that, after the initial diagnostic and condition consequence screening was conducted, the additional health status information we had available to us was not refined sufficiently to explain their use of outpatient specialist care.
There are some study limitations. First, MCOs have complex organizational structures that were addressed only partially through our study variables. For example, the MCOs were able to provide only limited information about their specialty networks, such as the overall numbers of specialists for both adult and child enrollees. Because more specific information about the number of specialists available to pediatric enrollees was not available, it was not used in our analyses. Such information should be used in future studies.
Second, although studying children grouped by condition consequences provides important information that otherwise would be difficult to obtain about those with rare conditions, it does not provide clear benchmarks for assessing whether the specialty care was needed. Although we were able to determine that children cared for by pediatricians had more specialty use and black children used fewer specialty services, even after consideration of diagnoses and condition consequences, we could not determine whether the use was appropriate. It is possible that pediatricians in this study made unnecessary referrals and white children overused services, but this cannot be determined from our study.
Third, we included some relevant, county-level, health care delivery system measures in our analyses. Although these factors were not significant, it is possible that unmeasured community factors influenced the outcomes.
Despite these limitations, this study provides useful information about child-specific and MCO-related factors that contribute to outpatient physician specialist use among a group of low-income children with chronic conditions. The findings regarding fee-for-service and capitated payments and specialty referrals support additional exploration of blended payment systems, which use capitation for some services and fee-for-service supplements for others.24 Blended payments may encourage capitated PCPs to manage chronic conditions more completely, rather than making specialty referrals that may be discretionary. Providing financial incentives for meeting MCO-specific quality of care standards and ensuring good access to pediatricians in the PCP network are also important for the receipt of specialty care. Our study suggests that MCOs using these strategies would provide better access to specialty care for chronically ill children. However, additional studies examining MCO characteristics and specialty care use in the context of specific clinical situations are needed.
ACKNOWLEDGMENTS
This work was funded by the Agency for Healthcare Research and Quality, the American Association of Health Plans Foundation, and the Health Resources and Services Administration, Maternal and Child Health Bureau (grant U01 HS09949-02).
FOOTNOTES
Accepted Sep 24, 2004.
No conflict of interest declared.
REFERENCES
Forrest CB, Glade GB, Starfield B, et al. Gatekeeping and referral of children and adolescents to specialty care. Pediatrics. 1999;104 :28 –34
Curtis D, Kaye N, Riley T. Transitioning to Medicaid Managed Care: Children With Special Health Care Needs. Portland, ME: National Academy for State Health Policy; 1999
Szilagyi PG, Roghmann KJ, Foye HR, et al. The effect of independent practice association plans on use of pediatric ambulatory medical care in one group practice. JAMA. 1990;263 :2198 –2203
Ferris TG, Chang Y, Perrin JM, Blumenthal D, Pearson SD. Effects of removing gatekeeping on specialist utilization by children in a health maintenance organization. Arch Pediatr Adolesc Med. 2002;156 :574 –579
Simpson L, Fraser I. Children and managed care: what research can, can’t, and should tell us about impact. Med Care Res Rev. 1999;56 :13 –36
Shenkman E, Wu SS, Nackashi J, Sherman J. Managed care organizational characteristics and health care use among children with special health care needs. Health Serv Res. 2003;38 :1599 –1624
Shenkman E, Pendergast J, Reiss J, Walther E, Bucciarelli R, Freedman S. The school enrollment-based health insurance program: socioeconomic factors in enrollees’ use of health services. Am J Public Health. 1996;86 :1791 –1793
Stein REK, Jessop DJ. What diagnosis does not tell: the case for a noncategorical approach to chronic illness in childhood. Soc Sci Med. 1989;29 :769 –778
Shenkman E, Vogel B, Boyett J, Naff R. Disenrollment and re-enrollment patterns in a SCHIP. Health Care Financ Rev. 2002;23 :47 –63
Youngblade LE, Col J, Shenkman E. Adolescents’ health care use in a Title XXI program. J Adolesc Health. 2002;30 :262 –273
Stein RE, Westbrook LE, Bauman LJ. The Questionnaire for Identifying Children With Chronic Conditions: a measure based on a noncategorical approach. Pediatrics. 1997;99 :513 –521
Perrin JM, Kuhlthau K, Walker DK, Stein REK, Newacheck PW, Gortmaker SL. Monitoring health care for children with chronic conditions in a managed care environment. Matern Child Health J. 1997;1 :15 –23
Forrest CB, Nutting P, Werner JJ, et al. Managed health plans effects on the specialty referral process: results from the Ambulatory Sentinel Practice Network referral study. Med Care. 2003;41 :242 –253
Stern AL. Modeling the future of health care. Health Plan. 2001;Sep/Oct:15–18
Agresti A. Categorical Data Analysis. New York, NY: John Wiley & Sons; 2002
Iversen T, Luras H. The effect of capitation on GPs’ referral decisions. Health Econ. 2000;9 :199 –210
Boulis AK, Long J. Variation in the treatment of children by primary care physician specialty. Arch Pediatr Adolesc Med. 2002:156 :1210 –1215
Rushton JL, Clark SJ, Freed GL. Primary care role in the management of childhood depression: a comparison of pediatricians and family physicians. Pediatrics. 2000;105 (4 pt 2):957 –962
Newhouse JP. Pricing the Priceless: A Health Care Conundrum. Cambridge, MA: Massachusetts Institute of Technology; 2002
Robinson JC. Theory and practice in the design of physician payment incentives. Milbank Q. 2001;79 :149 –177(Elizabeth Shenkman, PhD, )