CMS-10262 Supporting Statement A--REVISED 011609--submitted to CMS

HIFA Enrollee Survey Supporting Statement

Supporting Statement

For Collection of HIFA Evaluation Project Data

This project represents a recently approved supplemental survey that is part of the Evaluation of the Health Insurance Flexibility and Accountability (HIFA) initiative under a Medicare/Medicaid Research and Demonstration (MRAD) Task Order contract between the University of Minnesota (UMN) and CMS (HHSM-500-2005-00027I, T.O. 2). CMS seeks to evaluate the statistical significance and strength of the relationship between the HIFA initiative and the number and rate of uninsured for health care in states that have implemented HIFA waivers. The Centers for Medicare & Medicaid Services (CMS) requests a one-year clearance for the HIFA data collection from the Office of Management and Budget.

BACKGROUND

Purpose of the collection: The purpose of the Task Order is to evaluate the statistical significance and strength of the relationship between the HIFA initiative and the number and rate of uninsured for health care in states that implement HIFA waivers. The Base Contract for this evaluation relies on analysis of secondary data, using data from the Current Population Survey (CPS) and Behavioral Risk Factor Surveillance System (BRFSS). However, several key policy questions cannot be answered without interviewing those actually enrolled in a HIFA program. While the CPS and BRFSS data allow us reasonably to identify people who fit the eligibility criteria for HIFA initiatives, those data do not identify actual enrollment in a HIFA program. Hence, the need to identify such people through state enrollment files and then to gather information described below specifically on these enrollees.

Research questions to be answered: This custom survey of HIFA enrollees is designed to answer five research questions:

1. What type of health insurance do HIFA enrollees self report in surveys?

2. What are the demographic characteristics of these enrollees?

3. What type of health insurance coverage, if any, did HIFA enrollees have just prior to enrollment in the HIFA program?

4. Among those with prior coverage, what prompted participation in the HIFA program?

5. What type of health insurance, if any, would HIFA enrollees have in the absence of HIFA?

Questions 1 and 2 are the focus of analyses based on secondary data conducted under the primary task order. The enrollee survey will permit additional answers to these questions. For example, learning how HIFA respondents characterize their coverage can inform analyses that assess the distribution of health insurance coverage types (or lack of coverage) over time in HIFA versus non-HIFA states. In addition, we can explore whether demographic characteristics of HIFA enrollees match the target population of those HIFA eligibles identified in the national CPS and BRFSS data sets – thus providing a test of methods used to identify the target populations in CPS and BRFSS.

However, the most important contributions of the HIFA enrollee survey are in answering Questions 3 through 5. There are no data available to answer Question 3 concerning the prior insurance status (insured or not) and insurance type (enrollment in private or public coverage) of those participating in HIFA programs. This is crucial to policy makers’ understandings of premium assistance and crowd-out in the insurance market.

• Premium Assistance. HIFA enrollees’ failure to take up employer offers of insurance before HIFA and willingness to do so with premium subsidies suggests behavioral responses in line with the premium-assistance emphasis of HIFA.

• Crowd-out. If HIFA enrollees were previously uninsured, no crowd out has occurred. HIFA enrollee’s prior enrollment in another public health insurance program suggests substitution of one form of public coverage for another. This is not the intent of the HIFA program and would not lead to a reduction of the rate of uninsurance. However, it is possible that this substitution would be more cost effective for the federal government if done through premium assistance, when enrollees and their employers contribute to the cost of coverage.
  
  Substitution of HIFA coverage for pre-existing private coverage is of greater concern from a policy perspective. Understanding enrollees’ motivations for this substitution in terms of cost, enhanced access to services, and potential health gains cannot be assessed from the secondary data used in the Base Contract.

Question 4 (asking respondents to recall what they might have done had the program not been in existence at the time of enrollment) allows for exploration of circumstances that led to enrollment in HIFA. Specifically, would the respondents have continued coverage through: (a) their own (or a family member’s) employer, (b) a self-purchased plan, or (c) continued enrollment in a public program? Or would the respondents have lost coverage due to a change in employment or some other change?

Question 5 above permits less direct measures of crowd-out and the effects of premium assistance. Although responses to hypothetical questions about past and future states must be used with caution, this line of questioning is a reasonable extension of Questions 3 and 4.

States and target populations to be surveyed: Using state administrative records, known adult HIFA enrollees will be sampled in New Mexico and Oregon.

The criteria for selecting these states are: (1) the existence of HIFA program enrollment that is large enough to draw meaningful samples (i.e., large enough to allow for 400 enrollee surveys to be completed in each state), (2) the presence of a substantial premium assistance component in the HIFA initiative, (3) access to contact information for HIFA enrollees and their length of enrollment, and (4) consistency of the selected HIFA programs with CMS’ most important policy priorities and information needs.

The basic elements of the HIFA programs for adults in New Mexico and Oregon are as follows:

NEW MEXICO

Description: The State contracts with managed care organizations to provide an insurance product for employers to offer to their low-income workers. The policy is paid for with a combination of State, Federal, employer, and employee contributions. Individuals are eligible for the plan if unemployed, but must pay both the individual and the employer share of the premium. Individuals must have been uninsured for a minimum of six months to be eligible, to discourage crowd-out of existing private insurance.

Eligibility: Adults below 200 percent FPL who are not eligible for Medicaid, Medicare, or CHAMPUS.

Premium Assistance: The employer premium share was initially projected as $75 per enrollee per month, with the employee share based on a sliding scale as follows: Up to 100% FPL, $0; 101-150% FPL, $20; 151-200% FPL, $35. Employers participating are required to provide coverage to a minimum of 75 percent of their total employees.

Enrollment: Overall HIFA enrollment in New Mexico was 7,444 as of May 2007. Given the design of the New Mexico initiative, most of this enrollment was premium assistance.

OREGON

Description: Oregon provides premium subsidies for the purchase of private health insurance for individuals with incomes up to 185 percent of FPL through Family Health Insurance Assistance Program (FHIAP). When ESI is unavailable, FHIAP offers premium assistance to purchase individual policies, including Oregon Medical Insurance Pool (OMIP) coverage, although there are some limitations to this option. Depending on cost effectiveness, individuals may also be enrolled into the Oregon Health Plan (OHP).

Eligibility: Parents, 100 percent-185 percent FPL; Childless Adults, 100 percent-185 percent of FPL; and Children, 175 percent-185 percent FPL.

Premium Assistance: Eligible individuals must enroll in premium assistance rather than receive direct coverage if they have access to cost-effective employer-sponsored insurance. The premium assistance component of the Oregon program is not limited to HIFA waiver-expansion populations. Employers are not required to contribute, but most do.

Enrollment: Enrollment in premium assistance programs was 5,535 in 2006. Overall HIFA enrollment was 41,057 as of May 2007. (Both enrollment figures include a small number of children.)

Note that, in Oregon, the HIFA waivers cover children 18 and under, as well as adults (parents and childless). The proposed survey will not seek information about these child enrollees, for four reasons: 1) state officials in Oregon indicate that the enrollments of children under HIFA are very small, compared to the much larger programs for adults; 2) there are no compelling technical or policy reasons to collect information on the small group of HIFA children; 3) surveys of children involve special complexities (e.g., survey is actually done through an adult parent or guardian); and 4) the overall survey increases in complexity and cost (e.g., for instrument design, testing, and survey administration) if the survey must collect health insurance information on both adults and children in this one state.

Methods: A telephone survey similar to the BRFSS will be administered. For our research Questions 1 and 2, the BRFSS survey includes a question about health insurance coverage and respondent demographics (which will be supplemented with questions about type of coverage that are adapted from the CPS). Questions have been created to measure self-reports of health insurance coverage prior to enrollment in HIFA, circumstances surrounding enrollment in HIFA, and perceptions of the likelihood of having health insurance in the absence of the HIFA program (our research Questions 3 through 5).

The justification for using questions drawn from BRFSS and CPS is that the main analysis under this Task Order draws on secondary analysis of BRFSS and CPS data, as noted above. Learning how known HIFA enrollees respond to questions similar to the health insurance series in the BRFSS and CPS may allow for a refinement of the models used in the main Task Order analysis comparing rates of coverage in HIFA and non-HIFA states over time.

Thus our goal is to ask the insurance questions in a manner similar to that of these secondary source data sets with a few improvements. Specifically we opted to create the HIFA enrollee survey in a manner that combines aspects of the BRFSS and CPS health insurance question series, but does not fully represent either. It is important to note the strengths and weaknesses of the health insurance question series in the BRFSS (only one question with no specificity about the type of coverage) and CPS (e.g., specifies type of coverage held, but using a reference period that impedes recall that is included at the end of a long survey and therefore suffers from item non-response). Therefore, we begin with the one question BRFSS item that asks the respondent/enrollee about current coverage of any type:

Do you (does NAME) have any kind of health care coverage, including health insurance, prepaid plans such as HMOs, or government plans such as Medicare?

1 Yes – GO TO EMPLOY

2 No – GO TO EMPLOY1

7 Don’t know / Not sure – GO TO EMPLOY

9 Refused – GO TO EMPLOY

Responses to this question then lead the respondent/enrollee into a CPS series which asks about specific types of coverage (employment based, self-purchased, public), asking about current coverage, rather than coverage in the prior calendar year.

A. JUSTIFICATION

1 . Need and Legal Basis

CMS approved this supplemental survey to be part of the larger HIFA initiative evaluation under a MRAD Task Order contract between the University of Minnesota and CMS (HHSM 500-2005-00027I, T.O. 2). CMS seeks to evaluate the statistical significance and strength of the relationship between the HIFA initiative and the number and rate of uninsured for health care in states that implement HIFA demonstrations.

Information must be collected directly from known HIFA enrollees in order to understand: 1) enrollees’ health insurance coverage prior to participation in HIFA and 2) what enrollees envision their coverage status would be in the absence of HIFA. The only way to obtain this information is to ask enrollees. This information is crucial to policy makers’ understanding of crowd-out in the insurance market and of the effects of premium assistance in reducing the rate of uninsured. As discussed above, gathering HIFA enrollees’ self-reports of their insurance coverage and demographic characteristics also will improve the analyses of secondary data underway in the Base Contract of the evaluation, which is limited to analyses of national surveys.

2. Information Users

This information will be used by the Phase Two HIFA Evaluation project and CMS to answer the five research questions outlined above. As stated above, no data are available from any other source to provide information about HIFA enrollees’ prior insurance status (insured or not), 2) type of coverage (private or public), and (3) insurance coverage in the absence of the HIFA initiative. This information is crucial to policy makers’ understanding of the potential impact of HIFA-oriented programs for improving access to health insurance coverage, the reduction of uninsurance, the reduction of crowd-out, and the effects of premium assistance. If this survey receives OMB approval, we will immediately finalize sampling and survey arrangements with New Mexico and Oregon. These arrangements will include provisions to share de-identified survey results with these states, to permit these data to be of greater use to state policy makers.

3. Use of Information Technology

The HIFA enrollee survey collects self-reported information through computer assisted telephone interviewing (CATI) techniques. Trained interviewers contact HIFA enrollees by telephone and the CATI program records this information electronically. This methodology reduces respondent and interviewer burden and reduces the potential for inconsistent responses. CATI enables the interviewer to move through the survey efficiently, especially for responses with complex skip patterns. This reduces the time burden to respondents and the need for call backs to correct errors. Verbal consent for participation is sought at the start of the interview; no signature from the respondent is required. Information will only be collected about adults, from adult respondents. As a result, the HIFA enrollee survey avoids additional complexities and burdens that would be created by having to collect information about children through interviews with parents/guardians.

4. Duplication of Efforts

No one has collected systematic data concerning: 1) HIFA enrollees’ prior health insurance coverage at the time of enrollment in HIFA programs, or 2) their sources of coverage in the absence of HIFA. To the best of our knowledge, the information collection we propose does not duplicate any other effort, and the information cannot be obtained from any other source. Specifically, the states to be surveyed (New Mexico and Oregon) have not run any surveys useful for our purposes, nor has any other organization or branch of government.

5. Small Businesses

This collection does not affect small businesses.

6. Less Frequent Collection

With reference to “less frequent” collection, this item is not applicable, as the proposed data collection is a one-time only survey.

If the collection is not conducted, it will reduce CMS’ ability to evaluate the statistical significance and strength of the relationship between the HIFA initiative and the number and rate of uninsured for health care in states that implement HIFA demonstrations. It will also reduce CMS’ ability to analyze issues of premium assistance and crowd-out within the HIFA initiatives of New Mexico and Oregon.

7. Special Circumstances

The most important special circumstance to note is whether valid and reli­able results can be generalized to the universe of study. The universe of interest in this survey is HIFA enrollees. The budget allows for a survey of enrollees in two states with HIFA initiatives, thereby limiting our ability to generalize beyond the two study states. However, the state by state variation in the design of HIFA initiatives, to some extent, justifies limiting the number of study states. And the two study states were selected because they are a window on particularly important issues in the HIFA initiative: specifically, premium assistance and crowd-out.

8. Federal Register/Outside Consultation

A 60-day FR notice was published on March 28, 2008.

We have consulted with CMS to ensure that none of the states collects this information.

9. Payments/Gifts to Respondents

No payments will be made to respondents.

10. Confidentiality

All survey vendor staff directly involved in data collection receive training in the protection of human subjects.

At the start of the interview, respondents will be told how their contact information was obtained, that their participation in the survey is voluntary, that all information will be kept private (except to the extent required by law) and that the data will be publicly reported only in aggregate form rather than individually identifiable form.

Once the survey is complete the only link to the identity of the enrollee will be a survey-assigned ID number, which is separate from the sample file that holds the respondents’ identifying information. Therefore, the information provided by the respondent in the survey is never linked back to specific individual identifying information. Further, the data will not be accessible to anyone outside the research team.

The contractor will not collect or receive respondent’s Social Security Numbers (SSN). When the sample information is obtained from the states, UMN’s HIFA evaluation team will instruct states to exclude SSN in the file provided for the sample frame. The UMN team will only request information needed to contact enrollees to invite them to complete the survey. Specifically, the UMN team will request name, contact information (address, telephone), and date of birth for tracking purposes in the event the state contact information is outdated.

Consistent with CMS policies, publicly reported data that CMS makes available will be aggregated and will not identify HIFA enrollees. The survey data file will be available only to the following: (1) UMN’s HIFA evaluation team, (2) CMS, and (3) in de-identified form, state policy makers in New Mexico and Oregon, after the UMN team reaches the necessary agreements with these states. Future non-team members would be required to apply to CMS to use the data, sign a Data Use Agreement with CMS and meet CMS's data policies and procedures to protect privacy and confidentiality that include, but are not limited to, submitting a research protocol and study purpose for approval.

11. Sensitive Questions

Not applicable, as none of the questions to be asked are of a sensitive nature.

12. Burden Estimates (Hours & Wages)

The sampling strategy is designed to reduce burden on survey respondents. Regardless of HIFA enrollment in a given state, a random sample of approximately 2,000 adult enrollees (de-duplicated within households) will be drawn from state enrollment data. Some sample elements will have incomplete or outdated contact information (address information is typically more accurate than telephone information). Based on its past experience, the UMN survey team expects that a 5:1 ratio of sample elements (2,000) to completed surveys (400 per state) will be adequate. Telephone and reverse look-up directories will be used to complete the sample frame. Data collection will be halted when we have completed surveys for 400 enrollees per state. We anticipate a response rate of approximately 40-50 percent given past experience with public program enrollee samples. This is consistent with the median BRFSS response rates of 51 percent (range is 33 percent to 66 percent based on the data quality report for the 2006 BRFSS survey).¹

The HIFA enrollee survey is a one-time cross sectional survey. Four-hundred enrollee surveys will be completed in New Mexico and Oregon, for a total of 800 interviews. Based on average BRFSS completion time, we anticipate the interview will take a maximum of 15 minutes to complete, resulting in 400 hours total burden to respondents in both states. Respondents will not be required or asked to research or document answers to any questions. The HIFA enrollee survey will be pre-tested to ensure this time burden per respondent is not exceeded.

13. Capital Costs

There are no capital costs associated with the survey.

14. Cost to Federal Government

The supplemental survey of HIFA enrollees represents a one-time cost to the Federal government through a Task Order contract between CMS and the University of Minnesota and its collaborators that totals $177,426.

15. Changes to Burden

This is a new collection.

16. Publication/Tabulation Dates

The planning estimate for the survey is as follows:

Activities	HIFA Enrollee Survey Schedule (Months following OMB Approval)
	1	2	3	4	5	6	7	8	9	10	11
1. Obtain OMB clearance	X
2. State participation commitment; permission for release of sample data; IRB approval		X	X
2. Draw sample			X
3. Finalize and pretest survey			X	X
4. Data collection					X	X
5. Data cleaning						X	X
6. Data analysis							X	X	X
8. Report writing, dissemination									X	X

With respect to complex analytical techniques, we should note the following. This analysis of survey data will generally involve conventional multivariate techniques. However, parts of the analysis will use more complex methods. For example, we will use certain complex analytic techniques to compare the respondent population from the survey (actual enrollees) to the characteristics of the target population identified through the CPS. From the analysis of CPS data that is also part of this project, we will estimate a series of binomial and multinomial logistic regression models, to predict health insurance status using the covariates also being collected in our HIFA surveys. We will only use as predictors those variables from the CPS that we will also be collecting on the HIFA survey respondents. We will then insert the values of the covariates for the HIFA survey respondents and generate probabilities of the various insurance “states” for each respondent in the absence of the HIFA program.

17. Expiration Date

This project represents a one-time only survey. CMS is not opposed to displaying the expiration date.

18. Certification Statement

There are no exceptions to this certification statement.

1 Centers for Disease Control. The 2006 Behavioral Risk Factor Surveillance System Summary Data Quality Report (May 3, 2007). Available at: http://ftp.cdc.gov/pub/Data/Brfss/2006SummaryDataQualityReport.pdf (accessed 8/17/07).

MRAD/TOC HHSM-500-2005-00027I, T.O. 2 Page 9