Accepted Manuscript Modeling the Determinants of Medicaid Home Care Payments for Children with Special Health Care Needs: A Structural Equation Model Approach Omolola E. Adepoju , PhD, MPH Yichen Zhang , MS Charles D. Phillips , PhD, MPH PII:
S1936-6574(14)00076-4
DOI:
10.1016/j.dhjo.2014.05.003
Reference:
DHJO 309
To appear in:
Disability and Health Journal
Received Date: 15 January 2014 Revised Date:
14 April 2014
Accepted Date: 12 May 2014
Please cite this article as: Adepoju OE, Zhang Y, Phillips CD, Modeling the Determinants of Medicaid Home Care Payments for Children with Special Health Care Needs: A Structural Equation Model Approach, Disability and Health Journal (2014), doi: 10.1016/j.dhjo.2014.05.003. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
ACCEPTED MANUSCRIPT
Modeling the Determinants of Medicaid Home Care Payments for Children with Special
RI PT
Health Care Needs: A Structural Equation Model Approach
Running Title: Medicaid Home Care Payments for Children
Authors
M AN U
Yichen Zhang1, MS
SC
Omolola E. Adepoju1§, PhD, MPH
Charles D. Phillips, PhD1, MPH
1
TE D
Texas A&M University, School of Rural Public Health, Department of Policy and Management, College Station, Texas, USA
§
AC C
EP
Corresponding author Omolola E. Adepoju, PhD, MPH Texas A&M University, School of Rural Public Health, Department of Policy and Management, 1266 TAMU, College Station, TX 77843 Email: [email protected] Phone: +1-979-458-1188
Key words: Special needs children, Medicaid PCS, Healthcare expenditures Financial disclosure: The authors have no conflicts of interest to declare. Abstract word count: 249; complete manuscript word count: 3064; number of references: 27 number of figures/tables: 3
ACCEPTED MANUSCRIPT
Abstract Background: The management of children with special needs can be very challenging and
RI PT
expensive Objective: To examine direct and indirect cost drivers of home care expenditures for this vulnerable and expensive population.
SC
Methods: We retrospectively assessed secondary data on children, ages 4–20, receiving
Medicaid Personal Care Services (PCS) (n=2760). A Structural Equation Model assessed direct
on Medicaid home care payments.
M AN U
and indirect effects of several child characteristics, clinical conditions and functional measures
Results: The mean age of children was 12.1 years and approximately 60% were female. Almost half of all subjects reported mild, moderate or severe ID diagnosis. The mean ADL score was
TE D
5.27 and about 60% of subjects received some type of rehabilitation services. Caseworkers authorized an average of 25.5 hours of PCS support per week. The SEM revealed three groups of costs drivers: indirect, direct and direct+indirect. Cognitive problems, health impairments,
EP
and age affect expenditures, but they operate completely through other variables. Other elements accumulate effects (externalizing behaviors, PCS hours, and rehabilitation) and send
AC C
them on a single path to the dependent variable. A few elements exhibit a relatively complex position in the model by having both significant direct and indirect effects on home care expenditures---medical conditions, intellectual disability, region, and ADL function. Conclusions: The most important drivers of home care expenditures are variables that have both meaningful direct and indirect effects. The only one of these factors that may be within the sphere of policy change is the difference among costs in different regions.
1
ACCEPTED MANUSCRIPT
Background The management of children with special needs can be very challenging and expensive.
RI PT
National estimates suggest that over 1 in every 5 households with children care for at least one child with special healthcare needs (1). Almost 1 million children in the state of Texas are
reported to fall in this category. In 2009, children with special healthcare needs receiving Texas
SC
Medicaid’s Personal Care Services (PCS) represented 0.25% of all children served. Despite this very small percentage, 5.06% of all Medicaid payments for children went to healthcare services
M AN U
for this special needs population. On average, the State spent $31, 570 on healthcare services per child receiving PCS versus $1,571 per child not receiving PCS (2). By and large, health care expenditures for these children are significantly higher than those for other children (3-5). Many children with disabilities and their families rely on Medicaid Personal Care
TE D
Services (PCS) for assistance in meeting their functional needs, especially Medicaid programs delivering community-based personal care (6). The Personal Care Services (PCS) program covers aide services in private residences to perform personal care tasks for patients who, due to
daily living (IADLs).
EP
impairment, require assistance with activities of daily living (ADL) and instrumental activities of
AC C
Children eligible for PCS can vary tremendously in age and in the conditions that qualify them for PCS. These differences result in variation in the amount of PCS authorized for a child. Authorizations may also vary because children reside in diverse environments where “barriers to, and supports for, functional independence may differ to a considerable degree” (7). Consequently, the Personal Care Assessment Form (PCAF) used to assess children in Texas to
ACCEPTED MANUSCRIPT
determine their need for PCS was designed to emphasize issues related to a child’s functional performance and living environment.
RI PT
Researchers have conducted considerable research on samples of children receiving services through the Medicaid PCS program in Texas from 2008 and 2009. These studies
focused largely on the relationship between functional capabilities and the number of approved
SC
PCS hours, providing information on medical conditions, intellectual disabilities, cognitive functioning, ADLs, IADLs, care resources available in household, and PCS hours.
M AN U
For example, Elliot et al (2011) reported that caregivers’ input on the severity of a child’s limitations significantly impacted the level of PCS services provided (3). Another study assessed the validity and reliability of the PCAF in measuring activity limitations and the amount of home care those families with special needs children might require. The scale demonstrated internal
TE D
consistency, concurrent, predictive, discriminant and construct validity (8). Others have investigated the utilization of medical services by children with special needs (9-14). These researchers reported significant variation and disparities in health care
EP
utilization patterns. Evidence abounds on the state variability and financial burden of raising children with special needs (15-17). Nonetheless, few studies have examined the major cost
AC C
drivers for these very expensive and vulnerable Medicaid recipients. This study models the determinants of Medicaid home care payments for children with special health care needs receiving personal care services. Clearly, a better understanding the needs of Medicaid PCS children is necessary to make informed decisions about the allocation of limited financial and staffing resources (18). We use Structural Equation Modeling (SEM) to estimate the effects of children’s strengths and needs on total expenditures for Medicaid home
ACCEPTED MANUSCRIPT
care. Research of this nature will shed light both on direct and indirect cost drivers for this group of children, the potential impacts on health care services usage, and provide justification
RI PT
for the need of these services.
Methods
SC
Data
We retrospectively assessed secondary data on children ages 4–20 receiving PCS
M AN U
through the Medicaid PCS program. Over a 6-month period (September 2008- April 2009), Medicaid case managers in each of Texas’ 11 Health Regions assessed the need for Medicaid PCS among children enrolled in the Texas Medicaid Early and Periodic Screening, Diagnostic and Treatment (EPSDT) Program. All information about a child’s health status came from family and
TE D
caregiver responses that were recorded by a case manager or from a case manager’s observations of the child during the assessment process. Children who were old enough to provide information elaborated on their caregivers’ responses. A total of 2,760 assessments
AC C
Measurement
EP
were completed and transmitted to the research team.
The Personal Care Assessment Form (PCAF) was used to determine the amount of Medicaid home care that families with special needs children might require. The PCAF assesses care for children 4–20 years of age and has been used in previous studies (2, 8, 19-22). The PCAF was developed by an interdisciplinary project team (including experts in health policy, public health, special education, school psychology, rehabilitation psychology, and health care
ACCEPTED MANUSCRIPT
management) and with input from representatives and staff in the Texas Department of State Health Services (2, 3, 8, 17, 19-22).
RI PT
The major endogenous variable in our analyses was home care payments, defined as the sum of Texas Medicaid PCS payments over the 6-months period after the PCAF data was collected. These data were obtained from the Texas Medicaid claims data provided by the
SC
Texas Medicaid & Healthcare Partnership, a state Medicaid contractor under ACS State
Healthcare LLC. The study sample (n=2760) represented more than half of all children receiving
M AN U
and being evaluated for PCS in 2009 (2, 5). The use of home care payments, as opposed to charges, is significant as it provides a much precise representation of the home care reimbursements for Medicaid PCS.
The exogenous variables used in the model predicting total expenditures for home care
•
TE D
were:
Age: Subject date of birth and age were obtained from the PCAF instrument and included in the SEM model as a continuous variable. All children in the study were
Region: Based on the eleven Texas Department of State Health Services (DSHS) Health Service Regions (Region 1: Lubbock; Region 2/3: Arlington; Region 4/5 North: Tyler;
AC C
•
EP
between ages 4-20.
Region 6/5 South: Houston; Region 7: Temple; Region 8: San Antonio; Region 9/10 El Paso; Region 11: Harlingen), initial results of our multivariate analysis was used to develop a dichotomous variable indicative of high and low cost regions. This variable was included in the SEM to summarize regional effects. Regions 3,4,6,10,11 were identified as high cost regions.
ACCEPTED MANUSCRIPT
•
Medical conditions: These were determined based on the PCAF classification of conditions that affect the subject’s functional, cognitive, or behavioral status or require
RI PT
treatment, therapy, or medication AND were diagnosed by a licensed or certified health care professional. Identified conditions include anemia, apnea, arthritis,
asthma/respiratory disorder, cancer, cerebral palsy, cleft palate, congenital heart
SC
disorder, cystic fibrosis, diabetes, epilepsy/chronic seizure disorder, hemophilia,
macro/microcephaly, metabolic disorder (e.g., PKU), muscular dystrophy, any paralysis,
M AN U
pathological bone fracture, renal failure, spinal cord dysfunction, substance abuserelated problems at birth, and traumatic brain injury. A sum of the number of medical conditions each child had was generated, with 0 representing the absence of any of these conditions.
Health impairments: An additive scale was created by summing items identified as
TE D
•
health conditions/problems that currently affect the subject’s functional, cognitive, or behavioral status or requires treatment, therapy or medication. These conditions
EP
include allergies, bed-bound/chair-fast, contractures, falls, fractures, hearing problems, limitations in range of motion, pressure ulcers, recurrent aspirations, shortness of
•
AC C
breath during normal activities, speech problems, and vision problems. Cognitive problems: For children who were not comatose, cognitive function was determined using a six-item, additive scale assessing short-term memory, longterm memory, procedural memory, daily decision-making, making oneself understood, and the ability to understand others.
ACCEPTED MANUSCRIPT
•
Activities of Daily Living (ADL): Several researchers agree that the most important subject characteristics in understanding care costs is ADL function (23-25). A summary
RI PT
scale based on the number of ADLs in which a child needed or received hands-on assistance (Hands-On ADL Scale) was used. Ten ADLs were assessed including bed
mobility, positioning, eating, transfers, locomotion inside the dwelling, locomotion
SC
outside the dwelling, toilet use, dressing, personal hygiene, and bathing. The summary scale was chosen because it resulted in a high r-squared value and exhibited a good
M AN U
measure of transparency so that each level in the scale had a clear meaning(3). The Hands-On ADL Scale has been used by other studies analyzing PCAF outcomes (3, 8).
Endogenous variables included in the SEM are:
Level of intellectual disability. Based on the PCAF’s delineation of intellectual disability
TE D
•
(ID) (0=No, 1=Mild ID, 2=Moderate, 3=Severe, 4=Profound), we created a dichotomous measure of ID (0=no ID diagnosis; 1=ID diagnosis) due to the wide variation in the
•
EP
prevalence of different levels of ID.
Externalizing behaviors: An additive scale was created by summing items identified as
AC C
externalizing behaviors (e.g., verbally abusive, bullying/menacing, socially inappropriate behaviors, resisting ADL care, damage to property and injury to animals). •
PCS hours: The number of approved PCS hours was determined by Medicaid case managers assigned to complete a 7-day record on the PCAF to determine the number of PCS hours a child required. Elliot et al (2011) described the process of determination of
ACCEPTED MANUSCRIPT
the number of hours as a dynamic process, involving both the case manager and the primary care giver of the child (3). Rehabilitation: A dichotomous variable representing those who received any type of
RI PT
•
rehabilitation services (1) versus those who did not (0), was created. •
Home payments- Medicaid payments for PCS services received by these children were
SC
obtained from the 2008-2009 Medicaid Claims dataset. Because of the skewed
distribution issues with Medicaid cost data, we used a log transformation of the cost
M AN U
variable to create a normal distribution that does not violate any statistical assumptions.
Analysis
All data management and analyses were performed using STATA 12. A conceptual model was created to show direct and indirect relationships between the exogenous and
TE D
endogenous variables (Table 1). Descriptive analysis employing frequencies, means and standard deviation are used to describe child characteristics, clinical conditions and functional
EP
measures. A SEM was designed to estimate the effects of medical conditions, health impairments, cognitive problems, Activities of Daily Living, Behaviors, PCS hours, and
AC C
Rehabilitation on Medicaid home payments for children receiving Medicaid PCS services. The SEM is a useful tool for this type of study because it provides a statistical technique for testing and estimating relations using a combination of statistical data and qualitative assumptions (26). Importantly, unlike the ordinary least squares (OLS) approach that quantifies the total effect for each predictor, SEM separately estimates the direct and indirect effects of
ACCEPTED MANUSCRIPT
each predictor variable. This attribute allows for the development of more “realistic” models of complex processes.
RI PT
To assess the goodness of fit for the SEM model, we report tests of Comparative Fit Index (CFI), Root Mean Square Error of Approximating (RMSEA), and Chi-square tests. We focus on CFI and RMSEA as the major goodness of fit evaluation criteria since the chi-square test for
SC
large samples are often statistically significant. The CFI compares the fit of a target model to the fit of the null model; a value ≥ 0.95 CFI indicates goodness of fit (27). RMSEA is an estimate of
M AN U
the average size of misfit with adjustment for the degree of freedom, which shows how far the estimated model is from observed data. A value of RMSEA ≤ 0.05 indicates a good fit, between 0.05 and 0.08 fair fit, from 0.08 to 0.1 mediocre fit and larger than 0.1 means the model need
TE D
improvement (27). For the SEM model, p>|z||z|
(0.02) (0.02) (0.02) (0.02) (0.02) (0.02) (0.02)
0.08 -0.02 0.02 0.14 0.07 0.09 0.11
0.11 (0.02) 0.26 (0.02) -0.24 (0.02) 0.15 (0.02) 0.39 (0.02) -0.08 (0.02) -0.09 -0.04 0.24 -0.33 -0.19
(0.02) (0.02) (0.02) (0.02) (0.02)
M AN U
0.41 (0.01) 0.28 (0.02) -0.12 (0.02)
SC
0.36
EP
Payment (home) Intellectual disability Externalizing behaviors PCS Hours Rehabilitation Medical conditions ADL Limitation scale Region PCS Hours ADL Limitation scale Age Region Rehabilitation Level of intellectual disability ADL Limitation scale Age Intellectual Disability Medical conditions Cognitive problems ADL Limitation scale Externalizing behaviors Medical conditions Health impairments Cognitive problems ADL Limitation scale Age
Estimate(S.E.) Standardized Non- Standardized
TE D
Path
0.11 (0.02) 0.04 (0.00) -0.02 (0.00)
0.00 0.00 0.00
0.06 (0.01) 0.05 (0.02) -0.02 (0.01)
0.00 0.00 0.00
-0.18 -0.06 0.17 -0.28 -0.11
(0.04) (0.04) (0.01) (0.02) (0.01)
0.28
0.19
0.00 0.11 0.00 0.00 0.00 Overall R2
0.22
0.62
AC C
EP
TE D
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
ACCEPTED MANUSCRIPT
SEM Model ***Non-standadized*******
(PCS Hours
S1936-6574(14)00076-4
DOI:
10.1016/j.dhjo.2014.05.003
Reference:
DHJO 309
To appear in:
Disability and Health Journal
Received Date: 15 January 2014 Revised Date:
14 April 2014
Accepted Date: 12 May 2014
Please cite this article as: Adepoju OE, Zhang Y, Phillips CD, Modeling the Determinants of Medicaid Home Care Payments for Children with Special Health Care Needs: A Structural Equation Model Approach, Disability and Health Journal (2014), doi: 10.1016/j.dhjo.2014.05.003. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
ACCEPTED MANUSCRIPT
Modeling the Determinants of Medicaid Home Care Payments for Children with Special
RI PT
Health Care Needs: A Structural Equation Model Approach
Running Title: Medicaid Home Care Payments for Children
Authors
M AN U
Yichen Zhang1, MS
SC
Omolola E. Adepoju1§, PhD, MPH
Charles D. Phillips, PhD1, MPH
1
TE D
Texas A&M University, School of Rural Public Health, Department of Policy and Management, College Station, Texas, USA
§
AC C
EP
Corresponding author Omolola E. Adepoju, PhD, MPH Texas A&M University, School of Rural Public Health, Department of Policy and Management, 1266 TAMU, College Station, TX 77843 Email: [email protected] Phone: +1-979-458-1188
Key words: Special needs children, Medicaid PCS, Healthcare expenditures Financial disclosure: The authors have no conflicts of interest to declare. Abstract word count: 249; complete manuscript word count: 3064; number of references: 27 number of figures/tables: 3
ACCEPTED MANUSCRIPT
Abstract Background: The management of children with special needs can be very challenging and
RI PT
expensive Objective: To examine direct and indirect cost drivers of home care expenditures for this vulnerable and expensive population.
SC
Methods: We retrospectively assessed secondary data on children, ages 4–20, receiving
Medicaid Personal Care Services (PCS) (n=2760). A Structural Equation Model assessed direct
on Medicaid home care payments.
M AN U
and indirect effects of several child characteristics, clinical conditions and functional measures
Results: The mean age of children was 12.1 years and approximately 60% were female. Almost half of all subjects reported mild, moderate or severe ID diagnosis. The mean ADL score was
TE D
5.27 and about 60% of subjects received some type of rehabilitation services. Caseworkers authorized an average of 25.5 hours of PCS support per week. The SEM revealed three groups of costs drivers: indirect, direct and direct+indirect. Cognitive problems, health impairments,
EP
and age affect expenditures, but they operate completely through other variables. Other elements accumulate effects (externalizing behaviors, PCS hours, and rehabilitation) and send
AC C
them on a single path to the dependent variable. A few elements exhibit a relatively complex position in the model by having both significant direct and indirect effects on home care expenditures---medical conditions, intellectual disability, region, and ADL function. Conclusions: The most important drivers of home care expenditures are variables that have both meaningful direct and indirect effects. The only one of these factors that may be within the sphere of policy change is the difference among costs in different regions.
1
ACCEPTED MANUSCRIPT
Background The management of children with special needs can be very challenging and expensive.
RI PT
National estimates suggest that over 1 in every 5 households with children care for at least one child with special healthcare needs (1). Almost 1 million children in the state of Texas are
reported to fall in this category. In 2009, children with special healthcare needs receiving Texas
SC
Medicaid’s Personal Care Services (PCS) represented 0.25% of all children served. Despite this very small percentage, 5.06% of all Medicaid payments for children went to healthcare services
M AN U
for this special needs population. On average, the State spent $31, 570 on healthcare services per child receiving PCS versus $1,571 per child not receiving PCS (2). By and large, health care expenditures for these children are significantly higher than those for other children (3-5). Many children with disabilities and their families rely on Medicaid Personal Care
TE D
Services (PCS) for assistance in meeting their functional needs, especially Medicaid programs delivering community-based personal care (6). The Personal Care Services (PCS) program covers aide services in private residences to perform personal care tasks for patients who, due to
daily living (IADLs).
EP
impairment, require assistance with activities of daily living (ADL) and instrumental activities of
AC C
Children eligible for PCS can vary tremendously in age and in the conditions that qualify them for PCS. These differences result in variation in the amount of PCS authorized for a child. Authorizations may also vary because children reside in diverse environments where “barriers to, and supports for, functional independence may differ to a considerable degree” (7). Consequently, the Personal Care Assessment Form (PCAF) used to assess children in Texas to
ACCEPTED MANUSCRIPT
determine their need for PCS was designed to emphasize issues related to a child’s functional performance and living environment.
RI PT
Researchers have conducted considerable research on samples of children receiving services through the Medicaid PCS program in Texas from 2008 and 2009. These studies
focused largely on the relationship between functional capabilities and the number of approved
SC
PCS hours, providing information on medical conditions, intellectual disabilities, cognitive functioning, ADLs, IADLs, care resources available in household, and PCS hours.
M AN U
For example, Elliot et al (2011) reported that caregivers’ input on the severity of a child’s limitations significantly impacted the level of PCS services provided (3). Another study assessed the validity and reliability of the PCAF in measuring activity limitations and the amount of home care those families with special needs children might require. The scale demonstrated internal
TE D
consistency, concurrent, predictive, discriminant and construct validity (8). Others have investigated the utilization of medical services by children with special needs (9-14). These researchers reported significant variation and disparities in health care
EP
utilization patterns. Evidence abounds on the state variability and financial burden of raising children with special needs (15-17). Nonetheless, few studies have examined the major cost
AC C
drivers for these very expensive and vulnerable Medicaid recipients. This study models the determinants of Medicaid home care payments for children with special health care needs receiving personal care services. Clearly, a better understanding the needs of Medicaid PCS children is necessary to make informed decisions about the allocation of limited financial and staffing resources (18). We use Structural Equation Modeling (SEM) to estimate the effects of children’s strengths and needs on total expenditures for Medicaid home
ACCEPTED MANUSCRIPT
care. Research of this nature will shed light both on direct and indirect cost drivers for this group of children, the potential impacts on health care services usage, and provide justification
RI PT
for the need of these services.
Methods
SC
Data
We retrospectively assessed secondary data on children ages 4–20 receiving PCS
M AN U
through the Medicaid PCS program. Over a 6-month period (September 2008- April 2009), Medicaid case managers in each of Texas’ 11 Health Regions assessed the need for Medicaid PCS among children enrolled in the Texas Medicaid Early and Periodic Screening, Diagnostic and Treatment (EPSDT) Program. All information about a child’s health status came from family and
TE D
caregiver responses that were recorded by a case manager or from a case manager’s observations of the child during the assessment process. Children who were old enough to provide information elaborated on their caregivers’ responses. A total of 2,760 assessments
AC C
Measurement
EP
were completed and transmitted to the research team.
The Personal Care Assessment Form (PCAF) was used to determine the amount of Medicaid home care that families with special needs children might require. The PCAF assesses care for children 4–20 years of age and has been used in previous studies (2, 8, 19-22). The PCAF was developed by an interdisciplinary project team (including experts in health policy, public health, special education, school psychology, rehabilitation psychology, and health care
ACCEPTED MANUSCRIPT
management) and with input from representatives and staff in the Texas Department of State Health Services (2, 3, 8, 17, 19-22).
RI PT
The major endogenous variable in our analyses was home care payments, defined as the sum of Texas Medicaid PCS payments over the 6-months period after the PCAF data was collected. These data were obtained from the Texas Medicaid claims data provided by the
SC
Texas Medicaid & Healthcare Partnership, a state Medicaid contractor under ACS State
Healthcare LLC. The study sample (n=2760) represented more than half of all children receiving
M AN U
and being evaluated for PCS in 2009 (2, 5). The use of home care payments, as opposed to charges, is significant as it provides a much precise representation of the home care reimbursements for Medicaid PCS.
The exogenous variables used in the model predicting total expenditures for home care
•
TE D
were:
Age: Subject date of birth and age were obtained from the PCAF instrument and included in the SEM model as a continuous variable. All children in the study were
Region: Based on the eleven Texas Department of State Health Services (DSHS) Health Service Regions (Region 1: Lubbock; Region 2/3: Arlington; Region 4/5 North: Tyler;
AC C
•
EP
between ages 4-20.
Region 6/5 South: Houston; Region 7: Temple; Region 8: San Antonio; Region 9/10 El Paso; Region 11: Harlingen), initial results of our multivariate analysis was used to develop a dichotomous variable indicative of high and low cost regions. This variable was included in the SEM to summarize regional effects. Regions 3,4,6,10,11 were identified as high cost regions.
ACCEPTED MANUSCRIPT
•
Medical conditions: These were determined based on the PCAF classification of conditions that affect the subject’s functional, cognitive, or behavioral status or require
RI PT
treatment, therapy, or medication AND were diagnosed by a licensed or certified health care professional. Identified conditions include anemia, apnea, arthritis,
asthma/respiratory disorder, cancer, cerebral palsy, cleft palate, congenital heart
SC
disorder, cystic fibrosis, diabetes, epilepsy/chronic seizure disorder, hemophilia,
macro/microcephaly, metabolic disorder (e.g., PKU), muscular dystrophy, any paralysis,
M AN U
pathological bone fracture, renal failure, spinal cord dysfunction, substance abuserelated problems at birth, and traumatic brain injury. A sum of the number of medical conditions each child had was generated, with 0 representing the absence of any of these conditions.
Health impairments: An additive scale was created by summing items identified as
TE D
•
health conditions/problems that currently affect the subject’s functional, cognitive, or behavioral status or requires treatment, therapy or medication. These conditions
EP
include allergies, bed-bound/chair-fast, contractures, falls, fractures, hearing problems, limitations in range of motion, pressure ulcers, recurrent aspirations, shortness of
•
AC C
breath during normal activities, speech problems, and vision problems. Cognitive problems: For children who were not comatose, cognitive function was determined using a six-item, additive scale assessing short-term memory, longterm memory, procedural memory, daily decision-making, making oneself understood, and the ability to understand others.
ACCEPTED MANUSCRIPT
•
Activities of Daily Living (ADL): Several researchers agree that the most important subject characteristics in understanding care costs is ADL function (23-25). A summary
RI PT
scale based on the number of ADLs in which a child needed or received hands-on assistance (Hands-On ADL Scale) was used. Ten ADLs were assessed including bed
mobility, positioning, eating, transfers, locomotion inside the dwelling, locomotion
SC
outside the dwelling, toilet use, dressing, personal hygiene, and bathing. The summary scale was chosen because it resulted in a high r-squared value and exhibited a good
M AN U
measure of transparency so that each level in the scale had a clear meaning(3). The Hands-On ADL Scale has been used by other studies analyzing PCAF outcomes (3, 8).
Endogenous variables included in the SEM are:
Level of intellectual disability. Based on the PCAF’s delineation of intellectual disability
TE D
•
(ID) (0=No, 1=Mild ID, 2=Moderate, 3=Severe, 4=Profound), we created a dichotomous measure of ID (0=no ID diagnosis; 1=ID diagnosis) due to the wide variation in the
•
EP
prevalence of different levels of ID.
Externalizing behaviors: An additive scale was created by summing items identified as
AC C
externalizing behaviors (e.g., verbally abusive, bullying/menacing, socially inappropriate behaviors, resisting ADL care, damage to property and injury to animals). •
PCS hours: The number of approved PCS hours was determined by Medicaid case managers assigned to complete a 7-day record on the PCAF to determine the number of PCS hours a child required. Elliot et al (2011) described the process of determination of
ACCEPTED MANUSCRIPT
the number of hours as a dynamic process, involving both the case manager and the primary care giver of the child (3). Rehabilitation: A dichotomous variable representing those who received any type of
RI PT
•
rehabilitation services (1) versus those who did not (0), was created. •
Home payments- Medicaid payments for PCS services received by these children were
SC
obtained from the 2008-2009 Medicaid Claims dataset. Because of the skewed
distribution issues with Medicaid cost data, we used a log transformation of the cost
M AN U
variable to create a normal distribution that does not violate any statistical assumptions.
Analysis
All data management and analyses were performed using STATA 12. A conceptual model was created to show direct and indirect relationships between the exogenous and
TE D
endogenous variables (Table 1). Descriptive analysis employing frequencies, means and standard deviation are used to describe child characteristics, clinical conditions and functional
EP
measures. A SEM was designed to estimate the effects of medical conditions, health impairments, cognitive problems, Activities of Daily Living, Behaviors, PCS hours, and
AC C
Rehabilitation on Medicaid home payments for children receiving Medicaid PCS services. The SEM is a useful tool for this type of study because it provides a statistical technique for testing and estimating relations using a combination of statistical data and qualitative assumptions (26). Importantly, unlike the ordinary least squares (OLS) approach that quantifies the total effect for each predictor, SEM separately estimates the direct and indirect effects of
ACCEPTED MANUSCRIPT
each predictor variable. This attribute allows for the development of more “realistic” models of complex processes.
RI PT
To assess the goodness of fit for the SEM model, we report tests of Comparative Fit Index (CFI), Root Mean Square Error of Approximating (RMSEA), and Chi-square tests. We focus on CFI and RMSEA as the major goodness of fit evaluation criteria since the chi-square test for
SC
large samples are often statistically significant. The CFI compares the fit of a target model to the fit of the null model; a value ≥ 0.95 CFI indicates goodness of fit (27). RMSEA is an estimate of
M AN U
the average size of misfit with adjustment for the degree of freedom, which shows how far the estimated model is from observed data. A value of RMSEA ≤ 0.05 indicates a good fit, between 0.05 and 0.08 fair fit, from 0.08 to 0.1 mediocre fit and larger than 0.1 means the model need
TE D
improvement (27). For the SEM model, p>|z||z|
(0.02) (0.02) (0.02) (0.02) (0.02) (0.02) (0.02)
0.08 -0.02 0.02 0.14 0.07 0.09 0.11
0.11 (0.02) 0.26 (0.02) -0.24 (0.02) 0.15 (0.02) 0.39 (0.02) -0.08 (0.02) -0.09 -0.04 0.24 -0.33 -0.19
(0.02) (0.02) (0.02) (0.02) (0.02)
M AN U
0.41 (0.01) 0.28 (0.02) -0.12 (0.02)
SC
0.36
EP
Payment (home) Intellectual disability Externalizing behaviors PCS Hours Rehabilitation Medical conditions ADL Limitation scale Region PCS Hours ADL Limitation scale Age Region Rehabilitation Level of intellectual disability ADL Limitation scale Age Intellectual Disability Medical conditions Cognitive problems ADL Limitation scale Externalizing behaviors Medical conditions Health impairments Cognitive problems ADL Limitation scale Age
Estimate(S.E.) Standardized Non- Standardized
TE D
Path
0.11 (0.02) 0.04 (0.00) -0.02 (0.00)
0.00 0.00 0.00
0.06 (0.01) 0.05 (0.02) -0.02 (0.01)
0.00 0.00 0.00
-0.18 -0.06 0.17 -0.28 -0.11
(0.04) (0.04) (0.01) (0.02) (0.01)
0.28
0.19
0.00 0.11 0.00 0.00 0.00 Overall R2
0.22
0.62
AC C
EP
TE D
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
ACCEPTED MANUSCRIPT
SEM Model ***Non-standadized*******
(PCS Hours
