Non-Substantive Change - Facility Three Rounds (2)

Justification of Non-Material Change

This is a request for a Non-Substantive Change, returning the frequency of collection of the facility instrument from two to three rounds per year in the Medicare Current Beneficiary Survey (OMB 0938-0568).

The Medicare Current Beneficiary Survey (MCBS) has been continuously collecting health and health care related data on the Medicare population since September 1991. This Medicare population consists of both individuals residing in the community as well as those that have been institutionalized. Historically the MCBS collects information using computer assisted in-person interviewing with interview timeframes occurring three times a year (January-April) [winter], (May-August) [summer], and (September-December) [fall].

In 2008 the Centers for Medicare and Medicaid Services (CMS) faced a tight budget year which necessitated finding cost savings in the MCBS. It was reasoned that by reducing data collection of the facility component (administered to our institutionalized population) from three rounds to two, CMS could save on cost while not diminishing data quality. Unlike in the community interview the facility interview is conducted with a facility staff proxy and aided by facility records, so there was little concern regarding recall limitations. Unfortunately the MCBS has encountered other challenges to data quality that we did not factor into our initial analysis.

All of the facility cases for the summer round were designated as “holdover cases”, (a field tracking mechanism which allows one round of data collection to be skipped) and were not released to the field. These cases were then released in the fall round. The data collection period for these cases was from the date of their last interview in the winter round to the date of the current interview in the fall round. This effectively doubled the data collection period from 4 months to 8 months on average. The unanticipated consequence of this change in design translates into lost cases.

There has been an increase in the number of crossover cases that are classified as non-response because so much time has elapsed that we cannot continue the timeline for the beneficiary. Crossover cases are cases in which the beneficiaries are in transition and they are moving from one component of facility to the next. This decrease in response rate can be measured by a drop in utilization rates. The estimated mean utilization rates per facility resident for emergency room, medical provider and outpatient visits decreased once the summer data collection round was eliminated. Additionally, the user rates for a majority of the ancillary services also showed a decrease during the same time period.

CMS would like to return the MCBS facility design to its previous three round interview cycle per year. While there will be an increase in burden upon the facility, this increase is modest. The following table illustrates the hours per case before and after changing the methodology.

Fall Round	Hours per Case
Pre-Elimination	3.22
Post-Elimination	3.92
Change	0.70

On average a facility component interview takes an hour to conduct with the Fall interview round taking a little longer due to the introduction of a new panel. Prior to the elimination of the three round survey year our Fall interviews were, on average, taking 3.22 hours to complete an interview. This time includes travel to and from the facility, obtaining agreement for participation by the facility administrator, and administrative tasks following the interview that don’t impact the facility. We were left with the interview taking roughly one hour to conduct.

Once the May through August facility round was removed, our Fall interview increased from 3.22 hours to 3.92 hours or 7 tens of an hour (42 minutes). This means, the reduction in burden on the facility dropped by 18 minutes a year under the two round interview cycle. By returning to the three round interview cycle we would be adding the May through August round and by extension, an hour of data collection time. Over the course of the year this hour will be offset by the reduction of approximately 42 minutes from the following interview round.