Responses to OMB questions_HomeHealthCAHPS_30June2009

Responses to Questions from the U.S. Office of Management and Budget on the

National Implementation of the Home Health Care CAHPS Survey

Jurisdiction

1. Under what authority is CMS mandating this collection? Has this been through the rule making process?

Response: Initially, the survey is voluntary. When it becomes mandatory, it will be through the rule making process (Home Health payment rule).

Confidentiality

2. Based on our initial reading, we would like to confirm that the Privacy Act will be applicable to the home health CAHPS. If so, please provide a SORN for this collection.

RESPONSE: The Privacy Act will be applicable to the mode experiment but not to the national implementation of the survey. For the national implementation only de-identified data will be submitted by survey vendors. The relevant SOR for the mode experiment data is HPMS SOR published in Federal Register/Vol. 73/No. 9/January 14, 2008.

Sampling Questions

3. For the mode experiment, can CMS please clarify how the sample will be drawn? Will drawing from volunteers produce a representative sample? Is a representative sample necessary for the purposes of conducting the mode experiment?

Response

The sample of home health agencies (HHAs) that will participate in the mode experiment was selected using a frame containing all active HHAs in the United States.  The sample was selected using a stratified systematic sample, where the explicit stratification was on the number of patients served. The HHAs were then recruited for participation. Therefore, the sample is a representative random sample and not a convenience sample. Additionally, RTI monitored the characteristics of the recruited HHAs to ensure that the sample represents the universe of HHAs.  The final sample of HHAs that will participate in the mode experiment has characteristics that are very similar to the universe. Patients within the sampled HHAs will be randomly selected. For larger agencies, systematic sampling will be used. For smaller agencies, a census of patients will be conducted to achieve the target of 300 completed interviews. Note that some HHAs that were not selected for the sample volunteered to participate in the mode experiment and will be allowed to participate. 

 

4. Please provide the final specifications of the stratification for your sampling plan, including the definition of the strata, the universe and sample size for each, and what variables are being used for sorting (implicit strata). The submission offers several proposed methods, but OMB requires the final methods of your choosing.

Response

The explicit stratum is the number of patients served.  Prior to selecting the sample, HHAs were sorted into one of 5 strata. The strata, number of HHAs in the universe for each stratum and the number of HHAs selected for each stratum are shown below.

Stratum	Description	Universe	No. Selected
1	HHAs serving fewer than 315 patients	3,983	18
2	HHAs serving 315 – 624 patients	1,349	16
3	HHAs serving 625 – 1,760 patients	1,459	36
4	HHAs serving 1761 – 10,105 patients	709	25
5	HHAs serving 10,106 and more patients	38	5

 

                

The implicit stratification used serpentine sorting and controlled for the following characteristics:  Patients Served (continuous), the State’s Rank in Utilization of HHA services per Medicare beneficiary (1-51), Type of Facility (Free-Standing/Hospital), Type of Control (For-Profit/Not-for-Profit) and Urbanicity (Urban /Rural)

The sample sizes are not proportionate to the number of HHAs in each stratum. The large sample of responses needed for the estimation of the candidate characteristics in the mode experiment requires oversampling the larger HHAs. The selection of patients within HHAs will be random, with random assignment to the three modes of survey administration.

5. 300 completed surveys are required of each HHA, rather than mandating a set precision criteria per HHA. Why not give all HHAs a set precision criteria, allowing smaller providers to conduct fewer surveys while still achieving the baseline precision? 

Response

There is a precedent for using a fixed target for the sampling in other CAHPS surveys.  The methods of conducting the survey of potentially 9000 HHAs with multiple vendors serving multiple HHAs would make monitoring the methods of computation of individualized sample sizes infeasible. In addition, the setting of individual targets based on individual variances and patient counts will not be possible until there is a track record with each HHA. 

There will also be a size threshold set for HHAs participating in the CAHPS surveys. The very small agencies will not be candidates for the survey.

6. Please provide the OMB number for the AHRQ testing referenced in part A of the submission, as well as the AHRQ report it generated.

Response

OMB No. 0935-0124. The Final Report is attached.

7. Are the results of the mode experiment being published or not? Page A20 is unclear on this point.

Response

Yes, results from the mode experiment will be published in a peer-reviewed journal, such as Health Affairs, the Health Care Financing Review or another appropriate journal.

8. Is hierarchal linear modeling being used to analyze at both the patient level and facility level? If not, why not?

Response

We do not intend at this time to use hierarchical modeling techniques. Sample size permitting, there is an inherent simplicity in using fixed effect models rather than random effects models. The effect of clustering of the data within HHAs on the standard errors can be accounted for.

9. Is CMS expecting the costs to approximate $6 per fielded survey or $6 per completed survey? At what rate of response are the costs to the HHA estimated?

Response

This is the cost per fielded survey. (NOTE TO SUZANNE AND LIZ. THE MENTION OF $6 WAS INCLUDED IN CMS’S RESPONSES TO THE PUBLIC COMMENTS. We assume that this amount was based on the experienced on HCAHPS?

10. A maximum rather than a minimum number of phone contact attempts is specified – was this intentional? Given that the standard is usually to set a minimum number of attempts, why the diversion from the usual policy?

Response

Survey vendors will be required to make a minimum of five call attempts to complete the interview by phone. The decision was made to require 5 attempts originally because requiring more than that will increase the cost of conducting the telephone survey; however, we will specify that a minimum of five call attempts must be made and allow HHAs to work with their survey vendors to determine the maximum number of call attempts. 

3