BACKGROUND AND OBJECTIVES:

Metabolic monitoring is important for children taking antipsychotic medication, given the risk for increased BMI, impaired glucose metabolism, and hyperlipidemia. The purpose was to examine the influence of provider specialty on the receipt of metabolic monitoring. Specifically, differences in the receipt of recommended care when a child receives outpatient care from a primary care provider (PCP), a mental health provider with prescribing privileges, or both was examined.

METHODS:

Medicaid enrollment and health care and pharmacy claims data from 2 states were used in the analyses. Providers were assigned to specialties by using a crosswalk of the National Provider Identifier numbers to specialty type. A total of 41 078 children were included.

RESULTS:

For both states, 61% of children saw ≥1 provider type and had adjusted odds ratios for receiving metabolic monitoring that were significantly higher than those of children seeing PCPs only. For example, children seeing a PCP and a mental health provider with prescribing privileges during the year had adjusted odds of receiving metabolic monitoring that were 42% higher than those seeing a PCP alone (P < .001).

CONCLUSIONS:

Shared care arrangements significantly increased the chances that metabolic monitoring would be done. For states, health plans, and clinicians to develop meaningful quality improvement strategies, identifying the multiple providers caring for the children and potentially responsible for ordering tests consistent with evidence-based care is essential. Provider attribution in the context of shared care arrangements plays a critical role in driving quality improvement efforts.

What’s Known on This Subject:

Children taking antipsychotic medication are at risk for obesity, impaired glucose metabolism, and hyperlipidemia. Yet <40% of children receive recommended metabolic monitoring. The influence of a health care providers’ specialty on the receipt of metabolic monitoring has not been evaluated.

What This Study Adds:

The majority of children taking antipsychotics see ≥1 provider specialty. Shared care arrangements between primary care physicians and mental health specialists significantly increased the chances that metabolic monitoring would be done compared with care delivered by 1 provider.

In 2015, the Centers for Medicare and Medicaid Services (CMS) adopted a measure set entitled “Safe and Judicious Use of Antipsychotics in Children and Adolescents” to quantify the quality of care for children in state Medicaid programs taking antipsychotics.1  The measure set was developed by the National Committee for Quality Assurance (NCQA) through the Pediatric Quality Measures Program,2  led by the Agency for Healthcare Research and Quality, in collaboration with the CMS. Metabolic monitoring is a key component of the measure set because antipsychotic use places children at risk for increased BMI, impaired glucose metabolism, and hyperlipidemia.3,4  The Metabolic Monitoring for Children and Adolescents on Antipsychotics measure remains an important priority nationally and is currently on the CMS Child Core Set,5  which is used to annually assess state-specific performance on pediatric quality measures. Receipt of metabolic monitoring is defined as youth who are 1 to 17 years of age, with ≥2 prescriptions for the same or different antipsychotics during the year in which the care was assessed, and who received 1 diabetes and 1 cholesterol monitoring test.6  Metabolic screening, which is conducted to obtain baseline information before the initiation of antipsychotic medications, is not part of the measure set.

Just 25% of children in the United States taking antipsychotic medications receive metabolic monitoring.7  Of note, enrollment in specialized managed care programs for children in foster care is associated with higher rates of metabolic monitoring relative to those who are not in foster care.7,8  Other factors influencing the receipt of recommended care are not sufficiently explored. For example, the influence of health care providers’ specialty training on the receipt of metabolic monitoring has not been evaluated.9  Nationally, ∼35% of children receiving outpatient care for behavioral health conditions, regardless of insurance type, see primary care providers (PCPs), such as general pediatricians and family physicians, 41% see mental health specialists, and 24% see both.10  The availability of child and adolescent psychiatrists is limited, even among the commercially insured, contributing to more children seeing a PCP for behavioral health care.11 

The purpose of this study is to examine the influence of provider specialty on the receipt of metabolic monitoring. We used the CMS Child Core Set specifications for the Metabolic Monitoring component of the Safe and Judicious Use of Antipsychotics for Children and Adolescents measure set6  in our study. This measure set is used to evaluate the quality of care that children receive in Medicaid and to develop quality improvement programs. Outpatient providers are often the focal point for quality improvement initiatives within Medicaid.12,13  Better understanding the provider factors that influence the receipt of metabolic monitoring can provide important information to drive future quality improvement interventions.

In our study, we examine differences in the receipt of recommended metabolic monitoring on the basis of (1) receipt of outpatient care from a PCP, mental health provider with prescribing privileges (MHP), or both and (2) the provider type (PCP or MHP) prescribing the antipsychotic the majority of the time. Information about the quality of care observed among different provider types as well as shared care provided by combinations of providers can be used to better target and tailor quality improvement interventions for both providers and families.14  The children in this study were enrolled in Florida or Texas Medicaid. Together, these 2 states provide care for 24% of all children enrolled in Medicaid in the United States.15 

In the study, we evaluated health care use for children enrolled in the Texas and Florida Medicaid programs. All children who were (1) prescribed an antipsychotic and (2) eligible for inclusion in the metabolic monitoring measure (Fig 1) were included in the study population.

FIGURE 1

Children included in the analyses. HbA1c, hemoglobin A1c; LDL, low-density lipoprotein; LDL-C, low-density lipoprotein cholesterol.

FIGURE 1

Children included in the analyses. HbA1c, hemoglobin A1c; LDL, low-density lipoprotein; LDL-C, low-density lipoprotein cholesterol.

Child-level Medicaid enrollment and health care and pharmacy claims data from 2014 to 2017 were used to illustrate trends in the metabolic monitoring rates. Data from 2017 were used in the multivariable analyses because these were the most recent data available from the states. The enrollment files contain information about the child’s date of birth, sex, race and/or ethnicity, place of residence, enrollment in foster care, and number of months enrolled in Medicaid. The health care and pharmacy claims contain the children’s diagnoses, recorded as International Classification of Diseases, 10th Revision codes, and filled prescriptions, using National Drug Codes. National Provider Identifiers (NPIs) are also in the claims data and were used to match providers rendering care and prescribing providers to the children and to identify the provider type.

We began by identifying the providers’ specialties. Providers’ NPIs were mapped to a provider specialty taxonomy by using Healthcare Provider Taxonomy Codes to identify PCPs and MHPs.16  PCPs were further assigned to the categories of pediatrician, family medicine, internal medicine, physician assistant, advanced registered nurse practitioner, or federally qualified health center. A NCQA-certified auditor reviewed and certified the taxonomy and the crosswalk of the NPIs to the provider specialty.

Two approaches were then used to attribute the children’s care to ≥1 providers. The first approach involved using the NCQA provider taxonomy, in combination with the presence of ≥1 Current Procedural Terminology code outpatient evaluation or management visit, Healthcare Common Procedural Coding System or UB Revenue Codes, to determine if the child had seen a PCP, MHP, or both at least once in the measurement year.

Second, children also were assigned by using the NCQA provider taxonomy in combination with the pharmacy data, which indicate the NPI of the provider writing the prescription, to the PCP or MHP prescribing the antipsychotic the majority of the time during the year. Across both states, 6996 unique PCPs and 1685 MHPs were identified as rendering and/or prescribing providers and mapped to the NCQA provider taxonomy.

The outcome of interest was receipt of care for metabolic monitoring (yes or no). The measure was calculated by using an NCQA-certified software program.1,17  Two sets of regression models were prepared for each state, along with a combined model. The predictor variable in the first set of models used provider specialty for the providers the child saw during the year in which the outcome was measured by using the previously described attribution strategies. The second set of regression models used provider specialty for the person who prescribed the antipsychotic most frequently during the year as the predictor variable. The covariates in all models were as follows: children’s sex, age, cardiometabolic diagnoses, enrollment in foster care, place of residence categorized by using the rural-urban commuting area codes (RUCCs),18  and the social vulnerability index (SVI). The SVI ranks census tracts on 15 social factors, including poverty, lack of vehicle access, and crowded housing.19  Place of residence and SVI are both associated with barriers to accessing appropriate care.20,21  ArcGIS software was used with the child’s address information to calculate the geospatial variables.

Cardiometabolic diagnoses included obesity, morbid obesity, overweight, abnormal weight gain, metabolic syndrome, hypertension, elevated blood pressure without hypertension, hyperlipidemia, acquired acanthosis nigricans, hyperinsulinism, impaired fasting glucose, hypercholesteremia, impaired glucose tolerance, hypertriglyceridemia, mixed hyperlipidemia, type 2 diabetes, and nonalcoholic fatty liver disease.

For the child to have a cardiometabolic diagnosis, the diagnosis had to appear at least once for an inpatient admission or a minimum of twice for an outpatient evaluation and management visit to improve the specificity of classification. Diagnoses associated with ancillary services (eg, laboratory and radiology) were not used. If a child had ≥1 of the diagnoses indicating a cardiometabolic condition, he or she was assigned to the cardiometabolic condition category.

The preceding criteria were also used to assign children to a behavioral health diagnosis. Children were assigned to ≥1 category if they had co-occurring conditions. The children’s behavioral health diagnoses are described but were not used in the multivariate modeling because all children taking antipsychotics should receive metabolic monitoring regardless of diagnosis.

Statistical analyses were performed by using SAS, version 9.4 (SAS Institute, Inc, Cary, NC); statistical significance was defined as P < .05. All analyses, including calculation of metabolic monitoring rates, accounted for clustering of patients by provider by using generalized estimating equations (GEEs). In all analyses, the provider clustering variable was defined by the most frequently seen outpatient provider in that year. Under this general approach, estimates of metabolic monitoring rates (and 95% confidence intervals) were calculated by state and year. Given stability in trends over time, all subsequent analyses were focused on 2017 data only. Rates were calculated by using the aforementioned patient and provider type characteristics (Table 1). Odds ratios comparing the provider types were calculated via GEEs. This was done in a univariate fashion (each provider variable separately) and in a multivariate model that included the provider variable and all covariates of interest (Table 2).

TABLE 1

Characteristics of Florida and Texas Medicaid Children and Adolescents Prescribed an Antipsychotic with Metabolic Monitoring Classification Calendar Year 2017

CharacteristicFlorida (N = 13 078)Texas (N = 28 000)Florida and Texas (N = 41 078)
Behavioral health condition (groups not mutually exclusive), n (%)    
 Schizophrenia 913 (7.0) 1993 (7.1) 2906 (7.1) 
 Autism without irritability 2712 (20.7) 4088 (14.6) 6800 (16.6) 
 Autism with irritability 443 (3.4) 679 (2.4) 1122 (2.7) 
 Bipolar 2395 (18.3) 6973 (24.9) 9368 (22.8) 
 ADHD without conduct disorder 7183 (54.9) 16 854 (60.2) 24 037 (58.5) 
 ADHD with conduct or disruptive behavioral disorder 1051 (8.0) 2970 (10.6) 4021 (9.8) 
 Conduct or disruptive behavioral disorder, no ADHD 407 (3.1) 816 (2.9) 1223 (3.0) 
 Anxiety or depression 4359 (33.3) 13 395 (47.8) 17 754 (43.2) 
 Trauma and stressor and/or adjustment related disorder 2692 (20.6) 6396 (22.8) 9088 (22.1) 
 Other mental health disorder 5204 (39.8) 13 974 (49.9) 19 178 (46.7) 
 None of the above diagnoses 784 (6.0) 1559 (5.6) 2343 (5.7) 
Behavioral health diagnosis is accepted condition for use of antipsychotics, n (%) 5961 (45.6) 12 991 (46.4) 18 952 (46.1) 
Any cardiometabolic diagnosis,an (%) 2175 (16.6) 5611 (20.0) 7786 (19.0) 
Enrolled in foster care, n (%) 2021 (15.5) 3811 (13.6) 5832 (14.2) 
During the year, ≥1 visit, n (%)   
 PCP 4006 (30.6) 8890 (31.8) 12 896 (31.4) 
 MHP 1564 (12.0) 1765 (6.3) 3329 (8.1) 
 PCP and MHP 7508 (57.4) 17 345 (62.0) 24 853 (60.5) 
Majority prescribing provider, n (%)    
 PCPb 3986 (30.5) 9366 (33.5) 13 352 (32.5) 
  Pediatrician 1052 (26.4) 3135 (33.4) 4187 (31.3) 
  Family medicine 110 (2.8) 878 (9.4) 988 (7.4) 
  Internal medicine 50 (1.3) 122 (1.3) 172 (1.3) 
  ARNP 2179 (54.7) 3822 (40.8) 6001 (44.9) 
  PA 563 (14.1) 1381 (14.7) 1944 (14.6) 
 MHP 7801 (59.7) 16 708 (59.7) 24 509 (59.7) 
 Other 1291 (9.9) 1926 (6.9) 3217 (7.8) 
Age, mean (SD; range) 12.6 (3.2; 3–17) 12.0 (3.4; 3–17) 12.2 (3.3; 3–17) 
Female sex, n (%) 4613 (35.2) 9752 (34.8) 14 365 (35.0) 
Race and/or ethnicity, n (%)    
 Black (Non-Hispanic) 1727 (13.2) 4224 (15.1) 5951 (14.5) 
 Hispanic 1677 (12.8) 11 080 (39.6) 12 757 (31.1) 
 Other 521 (4.0) 120 (0.4) 641 (1.6) 
 Unknown 5008 (38.3) 6575 (23.5) 11 583 (28.2) 
 White (Non-Hispanic) 4145 (31.7) 6001 (21.4) 10 146 (24.7) 
RUCC, n (%)    
 Rural (4–9) 484 (3.7) 3430 (12.3) 3914 (9.5) 
 Small or medium metropolitan (2–3) 4893 (37.4) 8559 (30.6) 13 452 (32.8) 
 Urban (1) 7701 (58.9) 16 011 (57.2) 23 712 (57.8) 
SVI, n (%)    
 Fourth quartile (Most vulnerable) 2796 (21.4) 7599 (27.1) 10 395 (25.3) 
 Third quartile 3076 (23.4) 7206 (25.7) 10 282 (25.0) 
 Second quartile 3666 (28.0) 6605 (23.6) 10 271 (25.0) 
 First quartile (least vulnerable) 3540 (27.1) 6590 (23.5) 10 130 (24.7) 
CharacteristicFlorida (N = 13 078)Texas (N = 28 000)Florida and Texas (N = 41 078)
Behavioral health condition (groups not mutually exclusive), n (%)    
 Schizophrenia 913 (7.0) 1993 (7.1) 2906 (7.1) 
 Autism without irritability 2712 (20.7) 4088 (14.6) 6800 (16.6) 
 Autism with irritability 443 (3.4) 679 (2.4) 1122 (2.7) 
 Bipolar 2395 (18.3) 6973 (24.9) 9368 (22.8) 
 ADHD without conduct disorder 7183 (54.9) 16 854 (60.2) 24 037 (58.5) 
 ADHD with conduct or disruptive behavioral disorder 1051 (8.0) 2970 (10.6) 4021 (9.8) 
 Conduct or disruptive behavioral disorder, no ADHD 407 (3.1) 816 (2.9) 1223 (3.0) 
 Anxiety or depression 4359 (33.3) 13 395 (47.8) 17 754 (43.2) 
 Trauma and stressor and/or adjustment related disorder 2692 (20.6) 6396 (22.8) 9088 (22.1) 
 Other mental health disorder 5204 (39.8) 13 974 (49.9) 19 178 (46.7) 
 None of the above diagnoses 784 (6.0) 1559 (5.6) 2343 (5.7) 
Behavioral health diagnosis is accepted condition for use of antipsychotics, n (%) 5961 (45.6) 12 991 (46.4) 18 952 (46.1) 
Any cardiometabolic diagnosis,an (%) 2175 (16.6) 5611 (20.0) 7786 (19.0) 
Enrolled in foster care, n (%) 2021 (15.5) 3811 (13.6) 5832 (14.2) 
During the year, ≥1 visit, n (%)   
 PCP 4006 (30.6) 8890 (31.8) 12 896 (31.4) 
 MHP 1564 (12.0) 1765 (6.3) 3329 (8.1) 
 PCP and MHP 7508 (57.4) 17 345 (62.0) 24 853 (60.5) 
Majority prescribing provider, n (%)    
 PCPb 3986 (30.5) 9366 (33.5) 13 352 (32.5) 
  Pediatrician 1052 (26.4) 3135 (33.4) 4187 (31.3) 
  Family medicine 110 (2.8) 878 (9.4) 988 (7.4) 
  Internal medicine 50 (1.3) 122 (1.3) 172 (1.3) 
  ARNP 2179 (54.7) 3822 (40.8) 6001 (44.9) 
  PA 563 (14.1) 1381 (14.7) 1944 (14.6) 
 MHP 7801 (59.7) 16 708 (59.7) 24 509 (59.7) 
 Other 1291 (9.9) 1926 (6.9) 3217 (7.8) 
Age, mean (SD; range) 12.6 (3.2; 3–17) 12.0 (3.4; 3–17) 12.2 (3.3; 3–17) 
Female sex, n (%) 4613 (35.2) 9752 (34.8) 14 365 (35.0) 
Race and/or ethnicity, n (%)    
 Black (Non-Hispanic) 1727 (13.2) 4224 (15.1) 5951 (14.5) 
 Hispanic 1677 (12.8) 11 080 (39.6) 12 757 (31.1) 
 Other 521 (4.0) 120 (0.4) 641 (1.6) 
 Unknown 5008 (38.3) 6575 (23.5) 11 583 (28.2) 
 White (Non-Hispanic) 4145 (31.7) 6001 (21.4) 10 146 (24.7) 
RUCC, n (%)    
 Rural (4–9) 484 (3.7) 3430 (12.3) 3914 (9.5) 
 Small or medium metropolitan (2–3) 4893 (37.4) 8559 (30.6) 13 452 (32.8) 
 Urban (1) 7701 (58.9) 16 011 (57.2) 23 712 (57.8) 
SVI, n (%)    
 Fourth quartile (Most vulnerable) 2796 (21.4) 7599 (27.1) 10 395 (25.3) 
 Third quartile 3076 (23.4) 7206 (25.7) 10 282 (25.0) 
 Second quartile 3666 (28.0) 6605 (23.6) 10 271 (25.0) 
 First quartile (least vulnerable) 3540 (27.1) 6590 (23.5) 10 130 (24.7) 

ADHD, attention-deficit/hyperactivity disorder; ARNP, advanced registered nurse practitioner; PA, physician’s assistant.

a

≥1 of the following diagnoses: obesity, morbid obesity, overweight, abnormal weight gain, metabolic syndrome, hypertension, elevated blood pressure without hypertension, hyperlipidemia, acquired acanthosis nigricans, hyperinsulinism, impaired fasting glucose, hypercholesteremia, impaired glucose tolerance, hypertriglyceridemia, mixed hyperlipidemia, type 2 diabetes, and nonalcoholic fatty liver disease.

b

PCP subspecialty percentages are among those whose majority prescribing provider is a PCP.

TABLE 2

Odds Ratios of Metabolic Monitoring by Provider Type, Florida and Texas Combined (2017)

Provider VariablesOverall (Texas and Florida Combined)Adjusted Model by StateState and Provider Interaction, P
UnadjustedaAdjustedbFloridaTexas
Odds Ratio (95% CI)POdds Ratio (95% CI)POdds Ratio (95% CI)POdds Ratio (95% CI)P
During the year, ≥1 visit         .0319 
 PCP Reference — Reference — Reference — Reference — — 
 MHP 0.80 (0.73–0.89) <.001 0.77 (0.69–0.86) <.001 0.64 (0.54–0.76) <.001 0.87 (0.75–1.00) .055 — 
 PCP and MHP 1.72 (1.62–1.82) <.001 1.42 (1.32–1.52) <.001 1.31 (1.16–1.47) <.001 1.47 (1.35–1.60) <.001 — 
Majority prescribing provider         .2709 
 PCP Reference — Reference — Reference — Reference — — 
 MHP 1.45 (1.36–1.54) <.001 1.25 (1.17–1.33) <.001 1.19 (1.05–1.34) .006 1.27 (1.18–1.38) <.001 — 
 Other 1.21 (1.11–1.33) <.001 1.20 (1.08–1.32) <.001 1.25 (1.07–1.45) .005 1.14 (1.00–1.30) .048 — 
Provider VariablesOverall (Texas and Florida Combined)Adjusted Model by StateState and Provider Interaction, P
UnadjustedaAdjustedbFloridaTexas
Odds Ratio (95% CI)POdds Ratio (95% CI)POdds Ratio (95% CI)POdds Ratio (95% CI)P
During the year, ≥1 visit         .0319 
 PCP Reference — Reference — Reference — Reference — — 
 MHP 0.80 (0.73–0.89) <.001 0.77 (0.69–0.86) <.001 0.64 (0.54–0.76) <.001 0.87 (0.75–1.00) .055 — 
 PCP and MHP 1.72 (1.62–1.82) <.001 1.42 (1.32–1.52) <.001 1.31 (1.16–1.47) <.001 1.47 (1.35–1.60) <.001 — 
Majority prescribing provider         .2709 
 PCP Reference — Reference — Reference — Reference — — 
 MHP 1.45 (1.36–1.54) <.001 1.25 (1.17–1.33) <.001 1.19 (1.05–1.34) .006 1.27 (1.18–1.38) <.001 — 
 Other 1.21 (1.11–1.33) <.001 1.20 (1.08–1.32) <.001 1.25 (1.07–1.45) .005 1.14 (1.00–1.30) .048 — 

CI, confidence interval; —, not applicable.

a

Unadjusted odds ratios are derived from the GEE model (accounting for most seen NPIs during the year) of only the individual provider variable and including only state as a covariate.

b

Adjusted odds ratios are derived from GEE models described above with both provider variables but also include as covariates the following patient characteristics: indicator of whether the behavioral health diagnosis is an accepted condition for use of antipsychotic, any cardiometabolic diagnosis (described in a footnote for Table 1), enrolled in foster care, age, sex, race and/or ethnicity, RUCC category, and SVI quartile.

Less than 40% of children received metabolic monitoring in 2017, with trends from 2014 to 2017 shown in Fig 2. The majority of the children were boys (∼65%) in both states, with an average age of 12.2 years (SD 3.3; Table 1). Approximately 40% of the children in Texas and 13% in Florida were Hispanic. Approximately one-quarter of the children resided in neighborhoods with the highest social vulnerability. Overall, 19% of children had ≥1 cardiometabolic condition. In Florida, 57% of children were seeing a PCP and MHP, and 62% in Texas were seeing both provider types, with 61% seeing both provider types in the combined state analysis. In both states, approximately one-third of children had a PCP and 60% had an MHP as the majority prescriber for their medication.

FIGURE 2

Trends in metabolic monitoring in Florida and Texas, 2014–2017. CI, confidence interval.

FIGURE 2

Trends in metabolic monitoring in Florida and Texas, 2014–2017. CI, confidence interval.

State-specific results are contained in Fig 3 and Table 2. Table 2 also shows the combined results for both states. Children seeing combinations of providers had adjusted odds ratios for receiving metabolic monitoring that were significantly higher than those for children seeing PCPs only. For example, children seeing a PCP and an MHP during the year had adjusted odds of receiving metabolic monitoring that were 31% and 47% higher in Florida and Texas, respectively, and 42% higher in the combined state analysis than those seeing a PCP alone (P < .001). Those seeing an MHP alone had adjusted odds of metabolic monitoring that were 36% and 13% lower in Florida and Texas, respectively, and 23% lower in combined state analysis than those seeing a PCP alone.

FIGURE 3

Metabolic monitoring rates by provider type based on outpatient visits and majority prescribing provider, 2017. A, During the year, at least 1 visit. B, Majority prescribing provider. ARNP, advanced registered nurse practitioner; CI, confidence interval; PA, physician’s assistant.

FIGURE 3

Metabolic monitoring rates by provider type based on outpatient visits and majority prescribing provider, 2017. A, During the year, at least 1 visit. B, Majority prescribing provider. ARNP, advanced registered nurse practitioner; CI, confidence interval; PA, physician’s assistant.

We also examined provider differences in receipt of metabolic monitoring on the basis of the provider being the majority prescriber for the antipsychotic medication. Children who had an MHP majority prescriber were 19% to 27% more likely in Florida and Texas, respectively, and 25% more likely in combined state analysis to receive metabolic monitoring than those who had a PCP as their majority prescriber.

Understanding the interface of PCPs with mental health specialists is important and can be a strategic lever in improving metabolic monitoring practices. In our study, we found that 61% of children taking antipsychotic medications in Florida and Texas Medicaid programs combined saw both a PCP and an MHP, which is consistent with national trends and care recommendeation.10,2229  Provision of care by both PCPs and MHPs significantly increased the chances that metabolic monitoring would be done compared with care delivered by PCPs only. We also found that when an MHP was the majority prescriber for antipsychotics, children had significantly higher odds of receiving metabolic monitoring relative to those whose majority prescriber was a PCP. These findings are consistent with previous studies30  and suggest that children taking antipsychotics should probably be monitored by PCPs and MHPs in collaboration.

Overall, 60% of children in these 2 states did not receive recommended metabolic monitoring. Our findings about the high percentage of children seeing 2 provider types and the low percentage receiving metabolic monitoring raise important questions about the role of provider attribution in impacting quality of care. When a child is attributed to a provider, the assumption is that the provider has some level of responsibility for providing the recommended care.31  Information about provider attribution can be used to better monitor quality of care, target and tailor quality improvement initiatives, and administer value-based purchasing programs to incentivize high quality care. In the case of children taking antipsychotic medication, attribution is complex. As seen in this study, multiple providers care for and/or prescribe medications for these children. For states, health plans, and clinicians to develop meaningful quality measurement and improvement strategies, identifying the providers caring for the children and potentially responsible for ordering tests, medications, and/or procedures consistent with evidence-based care is essential. Thus, attribution plays a critical role in engaging PCPs and MHPs in driving quality improvement efforts.

Rather than thinking about provider attribution as a single provider, in the case of children with complex chronic conditions, there often are multiple providers who comanage care in shared arrangements.32  Attributing care back to the multiple providers is important for care coordination and development of strategies to ensure that the evidence-based care is delivered and there is appropriate follow-up with the family and child to be sure care is received.

State Medicaid programs and Medicaid managed care plans can use the information from this study and apply the attribution logic to better understand the multiple providers seeing the children. In addition, Medicaid managed care plans can share information from the attribution logic with PCPs about other providers engaged in the care of children in their practice who are taking antipsychotic medication to facilitate communication and care coordination. This information sharing can be used to engage PCPs, MHPs, and others caring for the children in the development of quality improvement efforts across provider types and to promote care coordination to ensure that children receive recommended metabolic monitoring. Formalized shared care arrangements and adaptation of existing care delivery models to support integrated care, which can vary in degree from external coordination to on-site intervention and collaboration, are effective methods to promote partnership between primary and mental health providers.33 

This study has limitations that can also serve as directions for future research. Claims data only contain information about the care that was received. In the case of metabolic monitoring, it is not known whether health care providers ordered the metabolic monitoring and families did not take the children for the testing or if the testing was not ordered. Understanding both clinician and family factors that may influence the quality of care will allow for more targeted and tailored intervention strategies. It also is possible that metabolic monitoring tests were not recorded in the claims data for children who were hospitalized if laboratory testing were part of an inpatient bundled payment. We descriptively examined the percentage of children receiving metabolic monitoring among those with an inpatient stay in 2017 relative to those who were not hospitalized. In both states, 45% of children with an inpatient stay for nonbehavioral and behavioral health reasons received metabolic monitoring, compared with 28% of children not hospitalized. Therefore, we think it is unlikely that metabolic monitoring was undercounted related to inpatient stays.

It is also not known the extent to which health care providers have access to information about other providers the children are seeing. For example, it is not known whether the PCPs have information about whether the children are seeing MHPs and vice versa. In addition, Medicaid claims data do not contain information about supplemental services that children may receive through behavioral health welfare agencies, for example. Developing strategies to ensure that there is knowledge about and communication with other providers involved in the children’s care is an important step toward improving quality. In addition, because the percentages of children receiving metabolic monitoring were stable across the 3 years of data available for this study, we did not conduct a longitudinal analysis to determine patterns in children’s receipt of annual monitoring, which is an area for future research. Finally, this study was focused on the types of providers seeing the children. Other provider characteristics may be related to the receipt of metabolic monitoring that could be the focus for quality improvement initiatives and future research. Nonetheless, an important first step is identifying the different types of providers that children taking antipsychotics see and/or receive their prescriptions from as a first step in planning improvements in care.

Children taking antipsychotic medications often require care from multiple providers. In 2 of the 3 largest states in the United States, the majority of children taking antipsychotics were cared for by PCPs and MHPs. Yet in both states, consistent with national trends, ∼60% of children did not receive recommended metabolic monitoring.34  Quality of care measures do not take into consideration multiple providers who potentially could order the recommended care for children. The results of our study point to the importance of state Medicaid agencies and Medicaid managed care plans in identifying all providers caring for the children taking antipsychotic medication and using this information to engage the providers in quality improvement efforts to improve metabolic monitoring rates.

Dr Shenkman conceptualized and designed the study, drafted the initial manuscript, and reviewed and revised the manuscript; Drs Thompson, Bussing, and Forrest drafted the initial manuscript and contributed critical clinical content; Dr Gurka led the data analysis and drafted the data analysis section and the initial draft of the Results section in addition to participating in overall manuscript development, review, and revision; Dr Sun and Ms Mack conducted the data analysis and participated in drafting the data analysis and Results sections; Ms Woodard coordinated organizing the data and critically reviewed the manuscript for intellectual content; Dr Mistry critically reviewed the manuscript and provided the conceptualization related to the Pediatric Quality Measures Program; and all authors approved the final manuscript as submitted and agree to be accountable for all aspects of the work.

FUNDING: Supported by grant 1U18HS025298-03 from the Agency for Healthcare Research and Quality. Research reported in this publication also was supported in part by the University of Florida Clinical and Translational Science Institute, which is supported in part by the National Institutes of Health National Center for Advancing Translational Sciences under award number UL1TR001427. The content is solely the responsibility of the authors and does not necessarily represent the official views of the University of Florida’s Clinical and Translational Science Institute or the National Institutes of Health. Funded by the National Institutes of Health (NIH).

     
  • CMS

    Centers for Medicare and Medicaid Services

  •  
  • GEE

    generalized estimating equation

  •  
  • MHP

    mental health provider with prescribing privileges

  •  
  • NCQA

    National Committee for Quality Assurance

  •  
  • NPI

    National Provider Identifier

  •  
  • PCP

    primary care provider

  •  
  • RUCC

    rural-urban commuting area code

  •  
  • SVI

    social vulnerability index

1
National Committee on Quality Assurance
.
HEDIS measures for the safe & judicious use of antipsychotic medications in children and adolescents. Available at: https://www.ncqa.org/hedis/reports-and-research/national-collaborative-for-innovation-in-quality-measurement/hedis-measures-for-the-safe-judicious-use-of-antipsychotic-medications-in-children-and-adolescents/. Accessed February 10, 2020
2
Agency for Healthcare Research and Quality
.
PQMP measures. Available at: https://www.ahrq.gov/pqmp/measures/index.html. Accessed August 29, 2020
3
Agency for Health Care Research and Quality
.
CHIPRA pediatric quality measures program. Available at: https://archive.ahrq.gov/policymakers/chipra/pqmpback.html. Accessed February 13, 2020
4
Maayan
L
,
Correll
CU
.
Weight gain and metabolic risks associated with antipsychotic medications in children and adolescents
.
J Child Adolesc Psychopharmacol
.
2011
;
21
(
6
):
517
535
5
Centers for Medicare & Medicaid Services
.
6
National Committee for Quality Assurance
.
Metabolic Monitoring for Children and Adolescents on Antipsychotics (APM)
.
Technical specifications for health plans, HEDIS
.
2019
;
2
:
227
230
7
Leckman-Westin
E
,
Finnerty
M
,
Scholle
SH
, et al
.
Differences in Medicaid antipsychotic medication measures among children with SSI, foster care, and income-based aid
.
J Manag Care Spec Pharm
.
2018
;
24
(
3
):
238
246
8
Crystal
S
,
Mackie
T
,
Fenton
MC
, et al
.
Rapid growth of antipsychotic prescriptions for children who are publicly insured has ceased, but concerns remain
.
Health Aff (Millwood)
.
2016
;
35
(
6
):
974
982
9
Olfson
M
,
Blanco
C
,
Wang
S
,
Laje
G
,
Correll
CU
.
National trends in the mental health care of children, adolescents, and adults by office-based physicians
.
JAMA Psychiatry
.
2014
;
71
(
1
):
81
90
10
Petterson
S
,
Miller
BF
,
Payne-Murphy
JC
,
Phillips
RL
 Jr.
.
Mental health treatment in the primary care setting: patterns and pathways
.
Fam Syst Health
.
2014
;
32
(
2
):
157
166
11
Cama
S
,
Malowney
M
,
Smith
AJB
, et al
.
Availability of outpatient mental health care by pediatricians and child psychiatrists in five U.S. cities
.
Int J Health Serv
.
2017
;
47
(
4
):
621
635
12
Beck
A
,
Boggs
JM
,
Alem
A
, et al
.
Large-scale implementation of collaborative care management for depression and diabetes and/or cardiovascular disease
.
J Am Board Fam Med
.
2018
;
31
(
5
):
702
711
13
Butz
C
,
Valleru
J
,
Castillo
A
,
Butter
EM
.
Implementation of an outcome measure in pediatric behavioral health: a process improvement initiative
.
Pediatr Qual Saf
.
2017
;
2
(
6
):
e043
14
Sultan
MA
,
Pastrana
CS
,
Pajer
KA
.
Shared care models in the treatment of pediatric attention-deficit/hyperactivity disorder (ADHD): are they effective? [published online ahead of print March 26, 2018]
.
Health Serv Res Manag Epidemiol
. doi:
15
Kaiser Family Foundation
.
Monthly child enrollment in Medicaid and CHIP. Available at: https://www.kff.org/medicaid/state-indicator/total-medicaid-and-chip-child-enrollment. Accessed August 18, 2020
16
National Uniform Claim Committee
.
Health care provider taxonomy. Available at: www.nucc.org/index.php?option=com_content&view=article&id=14&Itemid=125. Accessed February 10, 2020
17
Inovalon
.
The Inovalon ONE® Platform. Available at: www.inovalon.com/howwehelp/hedis-advantage. Accessed November 4, 2018
18
United States Department of Agriculture
.
Rural-urban commuting area codes. Available at: https://www.ers.usda.gov/data-products/rural-urban-commuting-area-codes.aspx. Accessed February 10, 2020
19
Agency for Toxic Substances and Disease Registry
.
CDC SVI Publications & Materials. Available at: https://svi.cdc.gov/publications.html. Accessed February 10, 2020
20
Arthur
KC
,
Lucenko
BA
,
Sharkova
IV
,
Xing
J
,
Mangione-Smith
R
.
Using state administrative data to identify social complexity risk factors for children
.
Ann Fam Med
.
2018
;
16
(
1
):
62
69
21
Pullmann
MD
,
VanHooser
S
,
Hoffman
C
,
Heflinger
CA
.
Barriers to and supports of family participation in a rural system of care for children with serious emotional problems
.
Community Ment Health J
.
2010
;
46
(
3
):
211
220
22
Committee on Psychosocial Aspects of Child and Family Health and Task Force on Mental Health
.
Policy statement–The future of pediatrics: mental health competencies for pediatric primary care
.
Pediatrics
.
2009
;
124
(
1
):
410
421
23
Anderson
LE
,
Chen
ML
,
Perrin
JM
,
Van Cleave
J
.
Outpatient visits and medication prescribing for US children with mental health conditions
.
Pediatrics
.
2015
;
136
(
5
).
24
Kuo
DZ
,
McAllister
JW
,
Rossignol
L
,
Turchi
RM
,
Stille
CJ
.
Care coordination for children with medical complexity: whose care is it, anyway?
Pediatrics
.
2018
;
141
(
suppl 3
):
S224
S232
25
de Voursney
D
,
Huang
LN
.
Meeting the mental health needs of children and youth through integrated care: a systems and policy perspective
.
Psychol Serv
.
2016
;
13
(
1
):
77
91
26
Langer
CS
,
Antonelli
RC
,
Chamberlain
L
,
Pan
RJ
,
Keller
D
.
Evolving federal and state health care policy: toward a more integrated and comprehensive care-delivery system for children with medical complexity
.
Pediatrics
.
2018
;
141
(
suppl 3
):
S259
S265
27
Croghan
TW
,
Brown
JD
.
Integrating Mental Health Treatment into the Patient Centered Medical Home
.
Rockville, MD
:
Agency for Healthcare Research and Quality
;
2010
28
Pidano
AE
,
Arora
P
,
Gipson
PY
,
Hudson
BO
,
Schellinger
KB
.
Psychologists and pediatricians in the primary care sandbox: communication is key to cooperative play
.
J Clin Psychol Med Settings
.
2018
;
25
(
1
):
32
42
29
Chun
TH
,
Mace
SE
,
Katz
ER
;
American Academy of Pediatrics
;
Committee on Pediatric Emergency Medicine, and American College of Emergency Physicians
;
Pediatric Emergency Medicine Committee
.
Evaluation and management of children and adolescents with acute mental health or behavioral problems. Part I: common clinical challenges of patients with mental health and/or behavioral emergencies
.
Pediatrics
.
2016
;
138
(
3
):
e20161570
30
Kolko
DJ
,
Campo
J
,
Kilbourne
AM
,
Hart
J
,
Sakolsky
D
,
Wisniewski
S
.
Collaborative care outcomes for pediatric behavioral health problems: a cluster randomized trial
.
Pediatrics
.
2014
;
133
(
4
).
31
Dowd
B
,
Li
C-H
,
Swenson
T
,
Coulam
R
,
Levy
J
.
Medicare’s Physician Quality Reporting System (PQRS): quality measurement and beneficiary attribution [published online ahead of print June 25, 2014]
.
Medicare Medicaid Res Rev
. doi:
32
Forrest
CB
.
A typology of specialists’ clinical roles
.
Arch Intern Med
.
2009
;
169
(
11
):
1062
1068
33
Kolko
DJ
,
Perrin
E
.
The integration of behavioral health interventions in children’s health care: services, science, and suggestions
.
J Clin Child Adolesc Psychol
.
2014
;
43
(
2
):
216
228
34
National Committee on Quality Assurance
.
Metabolic monitoring for children and adolescents on antipsychotics (APM). Available at: https://www.ncqa.org/hedis/measures/metabolic-monitoring-for-children-and-adolescents-on-antipsychotics/. Accessed August 23, 2020

Competing Interests

POTENTIAL CONFLICT OF INTEREST: The authors have indicated they have no potential conflicts of interest to disclose.

FINANCIAL DISCLOSURE: The authors have no financial relationships relevant to this article to disclose.