Access to medicaid providers: availability of mental health services for children and adolescents in child welfare in Louisiana

ABSTRACT The purpose of the study was to determine the level of access that youths in child welfare have to mental health providers in a single state. Mystery shoppers called every provider publicly advertised in Medicaid managed care organization networks. Results showed that 25.4% of the advertised network was able to schedule a new appointment for a child in Department of Children and Family Services guardianship. There were 9.7 accessible providers of any discipline (MD, PhD, or licensed masters-level clinician) per 10,000 Medicaid-enrolled youths in the population. The level of access to MDs was 4.1 times lower than the nationally recommended level.

cannot receive quality services, however, if they do not have access to mental health providers, and there have been few reports documenting such access for children involved with child-welfare systems.
Children in foster care are almost completely dependent on Medicaid for their health insurance needs in the USA, as nearly 100% of them are eligible for Medicaid through state or federal statutory provisions. Children in foster care are also disproportionately heavy users of Medicaid. In an analysis of 2005 Medicaid claims data, youths in foster care represented 3.2% of all children enrolled in Medicaid, but 15% of children using behavioral health services (Pires, Grimes, Allen, Gilmer, & Mahadevan, 2013) . Children who receive child welfare services and require behavioral health services are also more expensive to care for than children in child welfare who do not require behavioral health services. Less than half (44%) of children who receive child welfare services have a behavioral health diagnosis, but they account for 78% of total Medicaid spending for children in child welfare. It is estimated that child-welfare recipients with a behavioral health diagnosis have a per capita expenditure of $11,907 compared with $2,499 spent for their counterparts who did not have a behavioral health diagnosis (MACPAC, 2015).
Whether or not these youths have access to Medicaid providers has been a critical concern because the existing data indicate that the majority of youths with mental disorders in the child welfare systems do not receive services. A survey of 3,191 youths in the child welfare system of Los Angeles county during a 1-year period found that 68% of the total sample qualified for at least one of four mental disorders that were measured (mood disorder, anxiety disorder, attention-deficit/hyperactivity disorder, or behavior disorder) but only 38% of the total sample had received individual therapy (He et al., 2017). There are multiple reasons why youths in child welfare systems may fail to receive appropriate mental health services, including lack of universal screening, limited training for caseworkers, placement changes, barriers of access to providers who accept Medicaid, and lack of clinicians trained to provide effective services to this population.
To understand the barriers to access faced by caregivers seeking outpatient care for their children and adolescents, multiple studies have utilized the mystery shopper methodology. Upon the request of Congress to evaluate the adequacy of access to care for enrollees in Medicaid managed care networks, the Office of Inspector General conducted a survey of all providers participating in the 32 states contracted with managed care organizations (MCO) (Office of Inspector General, 2014). Using a mystery shopper methodology to make phone calls to a stratified random sample of 1,800 primary-care providers and specialists, the survey results showed that slightly more than half of the providers could not offer appointments to the enrollees: 35% could not be contacted at the location listed by the insurance plan, 8% said they were not participating in the plan, and 8% were not accepting new patients. This survey, however, focused on medical care, and the specialists they called did not include mental health providers.
In a study focused on access to outpatient mental health care for children and adolescents, Gallo and colleagues targeted 264 of the 340 outpatient mental health clinics in the state of New York licensed to serve children and adolescents (Gallo et al., 2017). The mystery shoppers were research assistants posing as a mother seeking both medication management and psychotherapy for her 14-year-old daughter with depression. The clinics that were called by the mystery shoppers were selected by stratified random sampling based on geographic region and callers were randomized to call with commercial versus Medicaid insurance. The researchers found that the shoppers were unable to make any appointments at 31% of the clinics, with the two most common reasons being that the clinic was full or the child was too young. The authors provided few details but made no mention of differences in access when the callers had commercial versus Medicaid insurance.
The only other known study that examined access to outpatient mental health care for children and adolescents also included access to primary-care pediatricians for comparison (Cama et al., 2017). Five cities were chosen for study that represented different geographic regions of the USA -Boston, Chapel Hill, Houston, Minneapolis, and Seattle. Providers in these cities were selected with stratified random sampling to equally represent providers who accepted either private Blue Cross Blue Shield preferred provider organization, Medicaid, or self-pay. The mystery shoppers posed as the parent of a 12-year-old child with depression. Their phone calls were answered or returned by 95% of the pediatrician offices versus 81% of the child and adolescent psychiatrists. The type of insurance made no difference in whether calls were answered or returned. Callers were able to make appointments with 40% of the pediatricians compared with only 17% of the child and adolescent psychiatrists.
Findings such as these have raised concerns that the number of providers that insurance organizations have been able to recruit into their networks is inadequate to meet the needs of their members. One consumer-driven concern in particular is that insurance organizations advertise artificially inflated networks that appear larger than the true network of available providers, known as "ghost networks." These "ghost networks" may be inflated by listing providers multiple times at different addresses, providers who no longer accept the insurance, and providers who surreptitiously limit the number of patients that they will accept.
In summary, the true extent of access represents a gap in our understanding. There are no known investigations of access to mental health providers for children in foster care. There has been no investigation of all mental health providers who accept Medicaid in a single state. The study of the state of New York by Gallo and colleagues did not include solo practitioners who accept Medicaid. Furthermore, no previous study has examined the availability of providers by training in regards to psychiatrist, psychologist, or other licensed professional counselor.
To address these gaps, this study conducted mystery shopper calls to every provider listed in Medicaid MCO networks in Louisiana to examine three research questions. (1) What percentage of the advertised Medicaid network of providers for the state of Louisiana was willing and able to actually schedule an intake appointment for a new patient? (2) What is the true number of providers in the Medicaid provider network of Louisiana who could treat youths, and how are they distributed by training, regions, and population size? (3) Does the level of access match the level of need? Because this was a descriptive study, directional hypotheses were not formulated.

Method
The Louisiana Child Welfare Trauma Project was a service implementation demonstration project funded by the national Children's Bureau. The project was primarily focused on training the child welfare workforce of Louisiana on a new psychiatric screen. A concurrent task was the current project to assess the adequacy of the provider network for this population.

Participants
The project was conducted in Louisiana, which is located in the southern region of the USA. It is ranked 25th in population and 31st in size among the 50 states. According to the latest available US Census Bureau statistics from 2010, it is 73.2% urban (United States Census Bureau, 2010). It includes 64 counties, 11 cities that meet the US Census Bureau definition of an Urbanized Area (50,000 or more people) and 64 cities that meet the definition of an Urban Cluster (at least 2,500 and less than 50,000 people). The racial composition is 62.6% White, 32% Black, 1.5% Asian, 0.7% American Indian, 1.5% other, and 1.6% two or more races.
The study population consisted of all providers listed in the online provider directories of the Medicaid MCOs. The project took place from 2014 through 2017. In March of 2012, Louisiana state government launched the Louisiana Behavioral Health Plan, a plan to transition to a privately contracted MCO to "improve the quality and accessibility to these services" (Magellan Health Services, 2012) (prior to 2012, a state agency managed these services).
At the start of the project, there was only one MCO authorized to contract with providers for Medicaid services. Magellan Health Services was awarded this contract to manage behavioral health care for Medicaid recipients beginning in March 2012. In December 2015, Magellan was replaced with five MCOs -Aetna Better Health, Amerigroup (a subsidiary of Anthem), Amerihealth, Louisiana Healthcare Connections (a subsidiary of Centene), and UnitedHealthcare Community Plan. After December 2015, consumers had the choice to enroll in one of the five MCOs. Providers had the option to contract with as many of the five MCOs as they wished.
For the child welfare system of Louisiana, administered by the Department of Children and Family Services (DCFS), the state is divided into nine administrative regions. Table 1 displays the percentage of each region defined as urban by the US Census Bureau. Calls for the first five regions (Regions 2, 3, 5, 7, and 9) began when Magellan was the single MCO for the state, and calls for the last four regions began when five MCOs operated in the state.

Measures
To determine the level of access, phone calls were made to each and every individual provider as listed by a Medicaid MCO for youths under the age of 18 throughout the nine regions of Louisiana. The purpose of the calls was to determine whether they were willing and able to schedule an appointment for a new patient. The situation described to providers over the phone was created so that providers would have to meet these criteria: (1) treat children, (2) accept Medicaid, (3) accept children in custody of DCFS, and (4) currently have availability to accept new clients.

Procedure
This project was reviewed by the Tulane University Committee on the Use of Human Subjects and was deemed to meet the criteria for exempt human  2014), informed consent of the clinicians was not required. Researchers believed that if clinicians knew the purpose of the project, a portion of clinicians who selectively limit the number of Medicaid clients that they accept would misrepresent themselves as being available. There were seven investigators who posed as mystery shoppers. Five investigators posed as a mystery shopper for a single region, and two investigators each completed two regions. Calls began in two regions (Regions 2 and 3) in 2014, three regions in 2015 (Regions 5, 7, and 9), two regions in 2016 (Regions 1 and 4), and two regions in 2017 (Regions 6 and 8).
The first step was to compile the total listings for each region by searching through the online provider directories. The method of searching varied depending on the user interface and the availability of search filters for each MCO. Using the filter options that were available in each interface, researchers included only licensed clinicians (MD, PhD, PsyD, LCSW, and LPC) and behavioral health agencies. Individuals with degrees who were not eligible to bill Medicaid as independent providers for individual assessment or psychotherapy, such as MSW and PLPC, were excluded. Researchers also included only those clinicians who were able to see youths 18 years or younger.
The second step was to eliminate duplicate listings. For example, if one individual was listed at three different physical locations, this was counted as one unique listing and two duplicate listings. If an individual was listed at a physical address and an agency was also listed at the identical physical address, the agency was counted as a duplicate listing. If an agency was listed that had no individuals listed at that physical address, the agency was counted as a unique listing.
The third step was to call every unique listing. If an individual was listed at more than one physical address, the mystery shopper called the individual at only one of those locations. Because the location called was picked at random, it seemed unlikely this method would be biased to pick locations that had gone out of business. If a facility listed multiple providers, researchers called the facility for only one of those providers. Researchers could not call the facility for multiple providers with the same mystery shopper scenario because the facility receptionist was likely to catch on that researchers were posing as a mystery shopper. If a live receptionist answered the phone, it was often possible to obtain information about which providers accepted Medicaid and which did not. If that level of detail was not available from a receptionist, then all of the providers at the facility were counted as meeting criteria if the provider being asked about met criteria. Conversely, all of the providers were counted as not meeting criteria if the provider being asked about did not meet criteria. The justification for this was that agencies usually credential all of their providers in the same networks as conditions of employment.
If a clinician was unable to schedule at the time due to a long waiting list, this was counted as meeting our criteria even though they were not willing to schedule an appointment. The justification for this is that the timeline of the study did not allow an open-ended waiting period for calling back clinicians, and the researchers opted to err on the side of counting these clinicians as accessible.
The caller identified herself as an aunt who had custody of a 10-year-old girl with anxiety about attending school. The caller read from a script, making clear that she was the aunt, the child was in DCFS guardianship, she needed a provider who accepted her Medicaid, and that she was looking for counseling.
If the provider or their assistant answered the phone, the caller asked if an appointment could be scheduled. If they responded affirmatively, the caller said she needed to confirm with her husband but did not make an actual appointment. If the provider did not answer the phone, the caller left a message with the full details of the script and a call-back number. If the provider did not call back within two weeks, they were considered to not meet criteria.

Data analysis
Descriptive statistics were used to calculate the percentage of listings that met accessibility criteria. To assess the adequacy of the number of unique providers, the proportions of providers per 10,000 youths 0-18 years who were enrolled in Medicaid in each region were calculated. The numbers of Medicaid-enrolled 0-18-year-old youths was available most recently for April 2018 (Louisiana Department of Health, 2018). The percentage of each region that was considered urban versus rural was calculated from countylevel data provided by the USA Census Bureau.

What percentage of the advertised network is the true network?
For the first research question, the numbers of advertised provider listings for each of the nine regions are shown in Table 1, and the total number of advertised listings for the entire state was 2,643 listings. Of these, 1,033 (39.1%) were duplicate listings and 1,610 (60.9%) were unique listings. Within the 1,610 unique listings, 476 (29.6%) could not be contacted, which was due to either the phone being disconnected, a wrong number, or failure to return our call. In addition, 459 (28.5%) were contacted but admitted that they did not meet our criteria. This was due to either they did not see youths or they no longer accepted Medicaid. That left a remainder of 675 unique listings who met criteria for seeing new patients. This represents 41.9% of unique listings and only 25.5% of the original number of 2,643 listings in the publicly advertised Medicaid directories.

Number of providers distributed by training, region, and population
For the second research question, Table 2 shows the numbers and proportions of providers with each type of specialty credentials. Of the 675 unique listings that met criteria for seeing new patients, the vast majority, 56.9%, were licensed clinical social workers or other masters-level licensed professional counselors. Medical doctors who are licensed to prescribe medications made up 11.7%. It is conceivable that nurse practitioners could also be prescribing medications, but they comprised only 2.1% of the network. Psychologists, who are licensed to conduct psychoeducational testing, made up 5.9% of the network.
The region with lowest access to providers based on population was Region 5, with only 12 providers available for a population of 105,475 youths, for a rate of 1.1 providers per 10,000 population of Medicaid-enrolled youths. The region with the highest rate of access was Region 3 with 164 providers available for a population of 79,338 youths, for a rate of 20.6 providers per 10,000. Region 5 had the lowest rate of access but the fourth highest percentage (69.2%) of urban residents among the nine regions, whereas Region 3 had the highest rate of access but the seventh highest percentage of urban residents (63.2%), suggesting that level of access could not simply be determined by urban-rural classification. Overall, for the entire state, the rate of access was 9.7 providers per 10,000 population of Medicaid-enrolled youths. Does the level of access match the level of need?
The third research question examined whether the level of access matched the level of need. The level of access was calculated per convention as the number of providers per 10,000 population of children and adolescents. The level of need was the number of providers that should be available per 10,000 population. The only specialty for which there is a known benchmark estimate of the level of need is in regard to child and adolescent psychiatrists (i.e. MD's), which has been estimated as 4.7 child psychiatrists per 10,000 youths (American Academy of Child and Adolescent Psychiatry, accessed June 3, 2018; Brown, Zhang, & Schupper, 2016). When only psychiatrists are considered in our data, there were 1.1 child psychiatrists per 10,000 Medicaid-enrolled youths in Louisiana, which is 4.1 times below the recommended level.

Discussion
These findings document for the first time the level of access that children in the child welfare population have to mental health specialists within a state Medicaid insurance network. The main finding that the level of access advertised in the provider directories of the Medicaid networks was vastly larger than the actual, active network that was able to schedule appointments raises at least two concerns. The first concern is that trying to find providers within the network appears to be enormously challenging for parents and legal guardians. With 74.5% of the network being a "ghost network," a caregiver would have to make, on average, four phone calls to reach one provider who was active and able to schedule an appointment. Researchers saw this frustration first-hand during the course of the project. When the researchers met with DCFS caseworkers during the implementation of this project, the most common complaint from caseworkers was that there were no providers available to whom they could refer youths. Considering this in light of research that has shown that 54% of those with serious mental illness received no treatment at all in the past 12 months (Kessler et al., 2001), this may be in part due to frustration with trying to find providers.
The second concern is that the level of access appears markedly mismatched with the level of need. Based on the AACAP estimate that 4.7 child psychiatrists are needed for 10,000 youths, the level of access in Louisiana is 4.1 times below the recommended level. There are no known comparable level-of-access recommendations for child psychologists, or for psychotherapists such as licensed clinical social workers or licensed professional counselors.

Limitations
The number of Medicaid providers who were willing and able to make appointments for youths might have been larger if the researchers had been more persistent and made more follow-up phone calls after the first voicemail message went unreturned. However, if researchers could not reach a provider with one phone call, it is unlikely that parents or DCFS caseworkers would have much more success because they would move on to call other providers to find help for their children. On the other hand, the number of Medicaid providers who were considered willing to make appointments may actually be lower than our estimate because many providers work within facilities that have multiple providers. Once researchers were able to determine that one provider within a facility met the criteria, it was assumed all the listed providers for that facility also met the criteria. Researchers could not call the facility for multiple providers with the same mystery shopper scenario because they were likely to catch on that researchers were posing as a mystery shopper. Despite these limitations, our data are in-line with prior mystery shopper projects which found that ghost networks constituted the majority of the networks (Cama et al., 2017;Office of Inspector General, 2014).
Another limitation is that recommended levels of access could not be found for psychologists and licensed masters level clinical therapists. This limited the ability to determine whether the level of access was appropriate to the level of need.

Future directions
Medicaid insurance organizations are required by law to provide a certain level of access to their members. Federal regulations require states to have a written strategy, including standards for access to care (Office of Inspector General, 2014). There are, however, few concrete guidelines as to what constitutes adequate access. There is no uniform guidance on how frequently or how well insurance providers are required to monitor and update their provider directories. In Louisiana, MCOs are contractually required to provide access to a specialist within one month and that specialist must be within 30 miles of their home for rural residents and within 20 miles for urban residents (Louisiana Department of Health, 2016). The requirements make no mention of how many providers per 10,000 population are needed. It is also the responsibility of the State to enforce adherence to these standards by MCOs. Thus, States must delineate a strategy to assess and improve the quality of care provided by MCOs.
After this project was completed, the Louisiana Department of Health released an internal review of their oversight of their Medicaid network of behavioral health providers (Louisiana Department of Health, 2017). A review by the Office of Behavioral Health conducted from October through December 2016, found that 1,539 mental health providers reported by the managed care organizations were not properly licensed. These violations resulted in $44,000 in fines being levied, presumably on the MCOs. The report noted the difficulty of monitoring MCO networks because of staffing reductions that left only four individuals to monitor the behavioral health networks. There appears to be a need for greater transparency about when provider profiles are updated and more adequate funding for staffing in state agencies to perform the oversight and compliance monitoring (Louisiana Department of Health, 2017) .
Timely care represents another barrier to mental health treatment. In 2016, mystery shopping was used to assess access to care for 150 psychiatrists within a 20-30 mile radius of the Washington D.C. area. Researchers not only investigated the accuracy of three major insurance provider directories and whether psychiatrists were able to accept new patients but also the average wait times for new outpatient appointments. The study found only 7% of the sampled psychiatrists were able to make accommodations for a new patient within two weeks, and the mean average wait time for the next available appointment was 19 days (Blech et al., 2017).
Access and timeliness are only the beginning of determining the adequacy of a provider network. This study made no attempt in this project to determine the level of quality of services. Providers may differ greatly in their expertise to treat effectively different age groups or different disorders, and not all providers are trained equally well to deal with all problems. Further, providers may have varying degrees of comfort and experience serving children involved in the child-welfare system. Just because a provider is available to schedule an appointment does not necessarily mean they are qualified to treat that individual with the quality of care that patient needs.

Funding
This project was funded by the Children's Bureau, Administration for Children and Families, Department of Health and Human Services, award number 90CO1105. Devi Miron, PhD is an Associate Professor in the Department of Psychiatry and Behavioral Sciences at Tulane University School of Medicine. Her research interests include evidencebased psychotherapy and the impact of and recovery from trauma experienced by maltreated children.