Marketplace Health Plan Options for People with HIV Under the ACA: An Approach to More Comprehensive Cost Assessment
Enrollee Health Profiles
- In the “low utilizer” scenario, the enrollee is HIV positive with well managed disease and an undetectable viral load. This enrollee is on a once daily combination antiretroviral treatment and has biannual checkups with a specialist, along with routine laboratory monitoring. Beyond this no additional health care usage is anticipated.
- In the “high utilizer” scenario, the enrollee is HIV positive but has a more complex disease state along with multiple comorbidities common among people with HIV (diabetes, high cholesterol, hyperlipidemia, and depression) and is being treated for each condition. More frequent provider visits (four times annually) and routine laboratory monitoring are also included in the scenario.
These two enrollee health profiles provide a contrast in health spending related to just two possible utilization patterns for analytic purposes. Trends and averages found are not representative of what can be expected from all HIV positive enrollees. Actual plan costs for HIV positive enrollees will vary based on care and treatment and other individual factors.
In both scenarios, for purpose of identifying premium costs, which vary by age, the individual is assumed to be 35 years old and, in order to assess income as a percent of the FPL, have a family size of 1. The enrollee in all scenarios was assumed to be a non-tobacco user, which also has an impact on premiums in some states. Care and treatment regimens in both scenarios were constructed in consultation with HIV providers and based on the U.S. Department of Health and Human Services National HIV Treatment Guidelines.1 (More details available in Appendix B).
Profile: Lower-need enrollee
Frequency of specialist visits: 2x annually, spaced six months apart
Frequency of labs: 2x annually, spaced six months apart
Labs: Basic Chemistry/Comprehensive Metabolic Profile; CBC w/ Differential; Fasting Lipid Profile; Urinalysis; Viral Load (HIV-1 RNA); CD4 (CD4:CD8 profile); HbAlc (Glycosylated Hemoglobin).
Profile: Higher-need enrollee
Treatment: Sertraline (Zoloft), 50mg
Treatment: Atorvastatin (Lipitor), 40mg
Treatment: Hydrochlorothiazide (Microzide), 25mg and Lisinopril (Prinivil, Zestril) , 25mg
Treatment: Metformin (Glucophage), 1000,g
Frequency of specialist visits: 4x annually, spaced three months apart
Frequency of labs: 4x annually, spaced three months apart
Labs: Basic Chemistry/Comprehensive Metabolic Profile; CBC w/ Differential; Fasting Lipid Profile; Urinalysis; Viral Load (HIV-1 RNA); CD4 (CD4:CD8 profile); HbAlc (Glycosated Hemoglobin).
Plans in this analysis were drawn from the most populous zip code in the following sites: Los Angeles, California; New York City, New York; Miami, Florida; Atlanta, Georgia; and Dallas, Texas. These sites were chosen for several reasons. Together the states account for about 50% of the HIV national epidemic, with these urban areas alone account for one third (32%) of those with diagnosed HIV in the United States.2 In addition, these states have made different decisions regarding their insurance marketplaces with California and New York running their own state-based marketplace and Florida, Georgia, and Texas using the federally facilitated marketplace.
Potential out-of-pocket costs for each enrollee scenario were examined in five plans in each of the five zip codes identified above: the two silver plans with the lowest premiums and, for comparison, the platinum, gold, and bronze plans with the lowest premiums. In sum, 25 plans were assessed (5 plans each in 5 states). Plans were identified using either the state or federally facilitated marketplace, as appropriate. In order to conduct the analysis, plan documents were collected from the marketplace websites and directly from the issuer website.
As a result of graduated subsidies available to those between 100%-400% FPL, QHP enrollee costs vary both by plan selection and household income. To assess the role of income, and subsidy eligibility on the cost measures explored in this analysis, costs associated with the types are examined at six different annual income levels in each location: $50,000, $40,000, $32,000, $25,000, $20,000, and $16,500. This range of incomes captures an enrollee at each eligibility level for subsidies, considering both reduced premiums and cost-sharing reductions (see Table 1). The $50,000 income level enrollee is eligible for neither subsidy and thus effectively represents costs for enrollees above 400% FPL and below 100%FPL.
|Table 1. Income Scenarios in Analysis|
|Income Levels in Plan Analysis||% FPL*||Eligible for Premium Tax Credit||Eligible for Cost-sharing Reduction|
|$25,000||214%||Yes||Yes (73% AV)|
|$20,000||171%||Yes||Yes (87% AV)|
|$16,500||141%||Yes||Yes (94% AV)|
* Based on 2014 Federal Poverty Guidelines which determined subsidies in 2015 plans
Appendix B: Methodology
In order to assess estimated health costs in the plans analyzed, the cost of obtaining a predetermined set of services and treatments (described in Appendix A) was measured against individual plan benefit designs. The annual premium amount was then added to this cost estimate. In order to identify maximum (in-network) liability, a plan’s out-of-pocket maximum was added to the annual cost of premiums.
All plan data were collected from federal and state marketplaces, as appropriate, in early 2015 (for the 2015 plan year). For each of the 5 states, the plan with the lowest premium in each metal tier was selected. In addition, the second lowest cost silver plan was also included in the sample. If the first and second lowest premium silver plans had premiums of the same amount, the plan that was listed first on the marketplace was selected and then the plan that had the next lowest, but different, premium was also selected. Premium and imbedded subsidy amounts were generated by the marketplaces and were based on select enrollee characteristics detailed in Appendix A (i.e. age, family size, smoking status and location).
Plan design and associated cost-sharing were identified based on plan documents found on the marketplaces and issuer websites (e.g. summaries of benefits and coverage, plan brochures and more detailed plan documents). If necessary, calls to the issuer consumer numbers were made to gain greater detail or clarity regarding benefit benefits and coverage. When plan information was gathered through calls to the issuer, repeat calls were made to ensure that identical information about plan design was relayed during the second call as a way to improve reliability of the information. In some cases more than two calls were made to confirm benefit details. Most calls to the issuer were made to determine when deductibles applied (e.g. could an enrollee access medications before meeting the deductible) rather than to clarify costs associated with a particular service.
While calculation of total liability was fairly straightforward, calculation of expected health costs was somewhat more complicated. Drug and treatment needs for high and low utilizer enrollee profiles were developed in consultation with HIV specialists and in accordance with the U.S. Department of Health and Human Services National HIV Treatment Guidelines.3 The profiles include the frequency and type of provider visits, labs, and medications necessary to maintain good health. Costs to access this care were than mapped onto each plan scenario for the two types of enrollees.
While costs associated with care were straightforward to calculate when a copayment applied and when there was no deductible (or the deductible had been met), a baseline price for services had to be established in order to calculate costs when a coinsurance applied or before a deductible was reached. Identifying the actual cost to access medical care and prescription drugs in each plan would be near impossible given the proprietary nature of pricing for these services. While actual costs vary by issuer, plan, and provider, this analysis necessitated identification of proxy costs for services that could be used across the plan scenarios. Pricing data from publically available websites aimed at price transparency for cash paying consumers was used to identify these proxy costs. While this is an imperfect solution, a range of stakeholders in the HIV medical and prescription drug fields corroborated that this methodology provided reasonable cost estimates.
Medical costs (provider visits and lab fees) were obtained from the website Healthcare Bluebook (https://healthcarebluebook.com/). The CPT code for a Level 4 established patient visit was used for a provider visits based on research that found that at the median, HIV providers spend 30 minutes with established patients.4 For this analysis a Level 4 provider visit was priced at $223. The plan costs associated with specialist visits were used in this analysis (over primary care) given the specialty nature of many HIV providers. Labs used in the profiles of this analysis and the associated costs are below (see Table 2).
|Table 2. Labs and Associated Costs Used in Analysis|
|Laboratory Test||Bluebook Cost
(as of Feb. 2015)
|Basic Chemistry (Comprehensive Metabolic Profile w/ Creatine Kinase)*||$27||Test performed every 6 months||Test performed every 3 months|
|CBC w/ Differential||$20|
|Fasting Lipid Profile||$35|
|Viral Load (HIV-1 RNA)||$221|
|CD4 (CD4:CD8 profile)||$122|
|HbAlc (Glycosated Hemoglobin)||$25 (only high utilizer)|
|TOTAL (low/high utilizer)||$431/$456|
* Incl.: Serum Na, K, HCO3, Cl, BUN, creatinine, fasting glucose, phosphorus, CrCl, ALT, AST, T. Bilirubin, Glucose (does not include HC03 but Al says that is probably fine)
As with, medical costs, the cost to access prescription drugs before reaching a deductible and to establish coinsurance, had to be identified. Proxy costs were established using the website GoodRx (GoodRx.com), a consumer website that displays cash prices for drugs at pharmacies by zip code. GoodRx also acts as a mediator of drug prices in certain cases, negotiating deals with pharmacies to provide drugs at lower than cash prices, offering consumers a coupon. Assuming that if GoodRx is able to negotiate a lower price, issuers are as well, the mean coupon price at 5 nationally known pharmacies was taken from the same zip code used for plan selection.5 Again stakeholders, including those in the prescription drug field, were consulted about this methodology and while it is not an exact assessment of potential out-of-pocket costs, is thought to be a reasonable proxy. The drugs identified in the profiles and the proxy costs used for this analysis are below (see Table 3). The generic version of the drug was selected in all cases for treatment of non-HIV comorbidities. The HIV drug selected was Atripla, the most commonly prescribed antiretroviral, and as a generic equivalent is not available, the brand pricing was used.6 In the case of Atripla, the cost identified using the above methodology with GoodRx pricing is comparable to the Wholesale Acquisition Cost (WAC) which is frequently used in similar types of analysis of brand drugs. Drugs were assumed to have been obtained through a brick and mortar pharmacy rather than online, which in some cases would offer additional discounts to the enrollee.
|Table 3. Prescription Drugs and Associated Monthly Costs Used in this Analysis|
|Condition: Drug||Monthly Rx Cost|
|Depression: Sertraline, 50mg||$11|
|Hyperlipidemia: Atorvastatin, 40mg||$22|
|Hypertension: Hydrochlorothiazide, 25mg and Lisinopril, 25mg||$10|
|Diabetes: Metformin, 1000mg||$8|
With baseline costs for services and treatments established, these costs were mapped onto plan benefit designs according to each of the two enrollee profiles across the various plan scenarios. Whether a deductible applied to each service or treatment was taken into account. This process was repeated twice and when the resulting costs identified differed, was repeated a third time, to improve reliability. Once costs were established for each of 300 enrollee scenarios, additional analysis was performed.
There are several limitations to this analysis. This analysis identifies costs for enrollees that fit two different medical profiles. While health status will not impact liability within a plan, the degree of individual health need could have a significant impact on expected OOP costs, especially if an enrollee had multiple comorbidities requiring expensive brand name treatments. The enrollees profiled here with more complex medical needs were able to access relatively inexpensive treatment for non-HIV needs as the drugs used in this analysis were available in generic form. Actual costs an individual would face might be much higher depending on medical need, cost of treatment, and cost of prescription drugs, especially if generic treatments are not available. More broadly, costs an individual might face in a health plan will vary significantly based on a number of individual factors such as age, family size, smoking status and income. Therefore these findings should not be considered to be the actual costs any specific individual would face in marketplace plans. Rather they represent the costs of the hypothetical individuals profiled.
Another limitation is that only those plans with the lowest premiums in each metal tier were examined in this analysis so findings are limited to these products, which include the plans that have seen the highest level of enrollment. It is certainly possible that lower costs or lower liability could be found in plans with higher premiums which were excluded from this sample. Additionally, as mentioned above, standardized pricing was used to compare benefits across plans. This approach potentially masked the actual cost differences an enrollee would find within these products. In addition, as discussed in the report, this analysis assumed that all services were received in-network and that prescription drugs were received from in-network brick and mortar pharmacies. Yet, it has been shown that health plan enrollees often inadvertently access out-of-network care and thus associated costs maybe higher and may not count towards an out-of-pocket maximum.7 Generally, staying in-network maybe especially challenging for enrollees in narrow network plans which have become increasingly common in the ACA era. Also, this analysis looked at plan costs in urban areas but people in more rural regions could have fewer plan choices which may drive higher premiums, and therefore these findings are not necessarily applicable to those living in rural areas.
It should be noted that expected costs in this analysis were examined in the aggregate across a year so do not show where a deducible made the first part of a year very costly (before plan benefits kicked in) but the later part of year more affordable (after the deductible had been met). A plan identified here as the option with the lowest potential costs could conceivably have had high initial upfront costs, especially if a deductible applied to HIV treatment costs. This would likely be significant barrier to staying engaged in care and treatment for an individual paying those costs out-of-pocket but might be something that was manageable for a third party payer covering cost-sharing, especially if it meant longer term savings.
Finally, it is worth noting that a limitation for individuals, researchers, third party payers, or enrollee entities conducting similar assessments going forward is the difficulty that exists in obtaining these data. While efforts were made to confirm details about plan design, especially when that information was gained verbally from plan representatives, plan design information used to conduct this analysis is only as good as what was provided by the issuer and was sometimes very difficult to find. In addition, the process of establishing costs of drugs and services was challenging and inexact as a result of the lack of cost transparency (resulting from the proprietary nature of the data) and there was a definite lack of clarity as to how to some plan benefit designs worked, particularly with respect to when a deductible applied. This latter challenge required intensive searching for additional plan documents and lengthy and multiple calls and wait times to issuers. Working through these challenges is a barrier to future research and to making informed enrollment decisions.