Medical Visits and Dental Exams for Children in Medicaid
Medical Visits and Dental Exams for Children in Medicaid
This study was a retrospective cohort study based on Iowa Medicaid enrollment and claims data (2000–2008) and was approved by the University of Iowa institutional review board.
In calendar year 2000, 14 364 children were born and were enrolled in Iowa Medicaid at birth. We excluded 7611 children who were enrolled less than 41 continuous months to allow for complete identification of each of the 10 recommended WBVs (see Main Predictor Variables section). In addition, we excluded 133 children who received dental care from a community health center to focus on children seen in private practice clinics.
Because our interest was on prevention-oriented dental examinations rather than treatment-driven visits, we also excluded children who received any restorative dental care before the first examination (n = 170) and those who received restorative dental care but no examination (n = 128). The final study population consisted of children born in calendar year 2000 who were enrolled for 41 or more continuous months starting from birth and who received dental care from private practice dental clinics (n = 6,322).
We organized model covariates into 5 domains (Figure 1): ascribed factors (immutable individual-level variables), proximal factors (modifiable individual-level variables), immediate factors (household-level mediators between proximal and intermediate variables), intermediate factors (community-level variables), and distal factors (system-level variables).
(Enlarge Image)
Figure 1.
Conceptual model of relationship between well baby visits and timing of first dental examinations for Medicaid-enrolled children.
Note. Dotted arrow indicates potential mediating pathway not evaluated in current study.
Source. The 5 model domains under which covariates were organized are based on Patrick et al.
We conceptualized the predictor variables and the outcome measure as proximal factors, and they reflect the hypothesized link between medical and dental care use.
The 2 sets of predictor variables were both proximal factors: (1) the total number of WBVs from birth to age 41 months (WBV frequency) by period and (2) the age at which the first WBV took place (timing of the first WBV). WBVs were identified from claims files using the following International Classification of Diseases,Version 9, Clinical Modification (ICD-9-CM) and Current Procedural Terminology (CPT) codes: V20.2, V70.0, V70.3, V70.5, V70.6, V70.8, V70.9, 99381, 99382, 99391, 99392, and 99432.
We used the 2000 American Academy of Pediatrics WBV schedule to assess whether a child received each of the following 10 recommended WBVs (no or yes): 1 month, 2 months, 4 months, 6 months, 9 months, 12 months, 15 months, 18 months, 24 months, and 36 months. We calculated the age at which each WBV took place by subtracting the child's date of birth from the WBV date of service and applying previously published age ranges around each WBV to allow flexibility in when WBVs occurred. For example, we classified a child who had a WBV between age 7 days and younger than 1 month as having had the 1-month WBV (Table 1).
To measure WBV frequency, we classified each WBV into 1 of 3 periods on the basis of age ranges relevant in dentistry and summed them (Table 1): number of WBVs before age 1 year (period 1, age at which primary incisors begin to erupt; as many as 5 visits possible), number of WBVs between ages 1 and 2 years (period 2, age at which primary molars begin to erupt; as many as 3 possible visits), and number of WBVs between ages 2 and 3 years (period 3, age at which all primary teeth have erupted; as many as 2 visits possible).
The timing of the first WBV consisted of 2 variables: (1) whether the first WBV took place by time t (a time-dependent indicator variable) and (2) the interaction between the first variable and the exact age (in days) at which the first WBV took place.
The main outcome measure, a proximal factor, was the timing of the child's first dental examination measured as the first prevention-oriented dental examination. We used the following American Dental Association Current Dental Terminology Codes to identify dental examinations from the claims files: D0150 (comprehensive dental examination) or D0120 (periodic dental examination). We included the D0120 code because some dentists use this to code young children's first dental examination rather than the more appropriate D0150.
There were 8 additional model covariates organized into the following domains:
We used survival analysis to test our study hypotheses. Data were censored for children with no examination by age 41 months, the end of the study period. For the survival analyses, children were part of the risk set until their first dental examination. Before running our regression models, we evaluated the proportional hazards assumption using timedependent covariates in the form of {covariate * [log(time) – mean log(time)]}. For any variable that violated this assumption, we included both the main effect and the timedependent covariate in the model. We ran 3 multiple-variable Cox proportional hazards regression models (a = .05) that included the following as time-dependent variables: (1) WBV frequency across 3 periods, (2) timing of the first WBV, or (3) both WBV frequency and the timing of the first WBV. Because we found no differences across the models, we reported hazard ratios and 95% confidence intervals from model 3. We completed all analyses using SAS version 9.3 (SAS Institute, Cary, NC).
Methods
This study was a retrospective cohort study based on Iowa Medicaid enrollment and claims data (2000–2008) and was approved by the University of Iowa institutional review board.
Study Participants
In calendar year 2000, 14 364 children were born and were enrolled in Iowa Medicaid at birth. We excluded 7611 children who were enrolled less than 41 continuous months to allow for complete identification of each of the 10 recommended WBVs (see Main Predictor Variables section). In addition, we excluded 133 children who received dental care from a community health center to focus on children seen in private practice clinics.
Because our interest was on prevention-oriented dental examinations rather than treatment-driven visits, we also excluded children who received any restorative dental care before the first examination (n = 170) and those who received restorative dental care but no examination (n = 128). The final study population consisted of children born in calendar year 2000 who were enrolled for 41 or more continuous months starting from birth and who received dental care from private practice dental clinics (n = 6,322).
Study Variables
We organized model covariates into 5 domains (Figure 1): ascribed factors (immutable individual-level variables), proximal factors (modifiable individual-level variables), immediate factors (household-level mediators between proximal and intermediate variables), intermediate factors (community-level variables), and distal factors (system-level variables).
(Enlarge Image)
Figure 1.
Conceptual model of relationship between well baby visits and timing of first dental examinations for Medicaid-enrolled children.
Note. Dotted arrow indicates potential mediating pathway not evaluated in current study.
Source. The 5 model domains under which covariates were organized are based on Patrick et al.
We conceptualized the predictor variables and the outcome measure as proximal factors, and they reflect the hypothesized link between medical and dental care use.
Main Predictor Variables
The 2 sets of predictor variables were both proximal factors: (1) the total number of WBVs from birth to age 41 months (WBV frequency) by period and (2) the age at which the first WBV took place (timing of the first WBV). WBVs were identified from claims files using the following International Classification of Diseases,Version 9, Clinical Modification (ICD-9-CM) and Current Procedural Terminology (CPT) codes: V20.2, V70.0, V70.3, V70.5, V70.6, V70.8, V70.9, 99381, 99382, 99391, 99392, and 99432.
We used the 2000 American Academy of Pediatrics WBV schedule to assess whether a child received each of the following 10 recommended WBVs (no or yes): 1 month, 2 months, 4 months, 6 months, 9 months, 12 months, 15 months, 18 months, 24 months, and 36 months. We calculated the age at which each WBV took place by subtracting the child's date of birth from the WBV date of service and applying previously published age ranges around each WBV to allow flexibility in when WBVs occurred. For example, we classified a child who had a WBV between age 7 days and younger than 1 month as having had the 1-month WBV (Table 1).
To measure WBV frequency, we classified each WBV into 1 of 3 periods on the basis of age ranges relevant in dentistry and summed them (Table 1): number of WBVs before age 1 year (period 1, age at which primary incisors begin to erupt; as many as 5 visits possible), number of WBVs between ages 1 and 2 years (period 2, age at which primary molars begin to erupt; as many as 3 possible visits), and number of WBVs between ages 2 and 3 years (period 3, age at which all primary teeth have erupted; as many as 2 visits possible).
The timing of the first WBV consisted of 2 variables: (1) whether the first WBV took place by time t (a time-dependent indicator variable) and (2) the interaction between the first variable and the exact age (in days) at which the first WBV took place.
Main Outcome Measure
The main outcome measure, a proximal factor, was the timing of the child's first dental examination measured as the first prevention-oriented dental examination. We used the following American Dental Association Current Dental Terminology Codes to identify dental examinations from the claims files: D0150 (comprehensive dental examination) or D0120 (periodic dental examination). We included the D0120 code because some dentists use this to code young children's first dental examination rather than the more appropriate D0150.
Model Covariates
There were 8 additional model covariates organized into the following domains:
Ascribed factors, modeled as confounders: child's sex (male or female); child's race, a factor related to the timing of dental visits for children, as reported by the child's caregiver (White, non-White, or missing); whether the child was at risk for developing a chronic condition (no or yes), a measure developed in consultation with a pediatrician with expertise in chronic conditions (John Neff, MD, personal communication, November 18, 2010; defined as an ICD-9-CM, CPT, or Healthcare Common Procedure Coding System codes indicating ventilator use, gastrostomy, tracheotomy, premature birth, low birth weight, infantile seizures, or newborn apnea during the first 16 months of life); and whether the child was eligible for Medicaid through the Supplemental Security Income Program for 6 or more months during the first year of life (no or yes), a measure of chronic condition severity.
Proximal factor: whether the child saw different medical providers for WBVs (no or yes), a measure of the caregiver's preference for or ability to obtain child health care services consistent with the medical home concept.
Immediate factor: whether the child's mother used any preventive dental care 12 months before the child was born (no or yes), a proxy for the caregiver's preferences for preventive dental care.
Intermediate factor: rurality, a 4-level variable based on the 2003 US Department of Agriculture Rural and Urban Continuum Codes associated with the child's residence (rural, urban nonadjacent to metropolitan, urban adjacent to metropolitan, metropolitan), which measures the physical, social, and economic resources available within the community.
Distal factor: whether the child lived in a dental Health Professional Shortage Area (no or yes), a measure of the dentist resources available at the system level.
Statistical Analyses
We used survival analysis to test our study hypotheses. Data were censored for children with no examination by age 41 months, the end of the study period. For the survival analyses, children were part of the risk set until their first dental examination. Before running our regression models, we evaluated the proportional hazards assumption using timedependent covariates in the form of {covariate * [log(time) – mean log(time)]}. For any variable that violated this assumption, we included both the main effect and the timedependent covariate in the model. We ran 3 multiple-variable Cox proportional hazards regression models (a = .05) that included the following as time-dependent variables: (1) WBV frequency across 3 periods, (2) timing of the first WBV, or (3) both WBV frequency and the timing of the first WBV. Because we found no differences across the models, we reported hazard ratios and 95% confidence intervals from model 3. We completed all analyses using SAS version 9.3 (SAS Institute, Cary, NC).
Source...