VA Memo

Veterans Affairs

Department of

Memorandum

Date: July 17, 2007

From: National Director, Hospice and Palliative Care

Subj: Bereaved Family Member Survey OMB Approval

To: Mary Stout, OMB Liaison



This memo is provided in follow up to a conference call with the Office of Management and Budget (OMB) on July 13, 2007. OMB has requested further clarification of the VA’s request for generic “ad hoc” approval of the Bereaved Family Member Survey as submitted in a memo February 7^th, 2007.

Background information and justifications for this Survey’s generic approval were submitted in February of 2007 and provided an overview of the need, development and use of this survey to improve the care of veterans at end of life. Each of the participating VA facilities understands and accepts the requirements outlined in VHA Directive 2003-007 for maintaining privacy, record keeping and reporting of findings.

The three areas requiring clarification prior to further OMB consideration of generic “ad hoc” approval of the Bereaved Family Member Survey are outlined below:

1. Survey burden on veterans’ families. From a broad perspective, the burden on veterans’ families will be substantially reduced once an appropriate sampling methodology is determined as described in #3. From an individual survey perspective, efforts to reduce burden were directed by an oversight committee consisting of researchers, clinicians and administrators who began with 56 survey items and reduced these to the current 16 that provide information most relevant to quality measurement. This much abbreviated version was submitted to OMB and includes a number of questions unique to veterans which are not captured in other surveys of this nature.

Phone administration of this Survey is the preferred method given its brevity (16 questions). Given the sensitive nature of the Survey content, phone interviews engage families in quality assessment and permit immediate referral for supportive counseling if needed. In contrast to this, mail out administration does not permit direct human interaction, places the burden of seeking counseling on the family member and may reduce response rates.

Lastly, as part of the phone administration method, an introductory letter is sent out two weeks prior to the interview to prepare the respondent for the call and to provide an “800” number to “opt out” of the survey if desired. In this manner, potential respondents can avoid all attempts at phone interviews and interviewers can focus their efforts on respondents more likely to complete the survey. Experience with use of the “opt out” number has shown a very small percentage of potential respondents use the “800” number with some using it to express their desire to complete the survey.

2. Potential non-response bias. Given the approximate 50 % response rate for this survey during the research validation phase and for similar surveys, an initial census approach was selected for the pilot phase of Survey implementation to permit analysis for potential non-response bias. By comparing a “low intensity” and “high intensity” approach to survey administration during the census phase of the pilot, non-responder bias can be examined and help direct future sampling methodologies. Sample areas of non-response bias to be examined include ethnicity or relationship to patient. It is worthy to note that during the validation work on the Bereaved Family Member Survey as part of an HSR&D grant non-responder bias was not present for ethnicity, relationship to patient, time from death to survey administration and other variables.

Another potential source of non-responder bias could occur when varying methods are used to identify the bereaved family member respondent. This type of non-response bias is very unlikely in this VA Survey, as contact information will be obtained from the VA electronic medical record database. The VA has received numerous awards for the development and use of its electronic medical record. Each medical record includes the listing of a designated “next of kin” which is updated on each admission and at least yearly for long-term nursing home residents. For all Survey interviews, the next of kin listed in the medical record will uniformly be the first attempted contact by all participating facilities. Should this next of kin direct the interviewer to a more appropriate respondent, attempts will be made to contact this alternative respondent. Should the next of kin or alternative not be available, the most recent clinical notes in the medical record will be used to identify an appropriate respondent. During previous research in validating the Bereaved Family Member Survey, less than 5% of initial next of kin contacts identified others as more appropriate responders for Survey completion.

In anticipation of OMB approval of the Bereaved Family Member Survey, an implementation work group and toolkit have been developed to promote uniform Survey administration. This toolkit outlines the procedures for Survey administration including use of the “next of kin” information listed in the medical record as the first contact for Survey attempts with identification of alternate responders as described above.

 

3. Census approach as a prelude to sampling. While we initially plan on using a census approach during the pilot phase to permit analysis of potential non-responder bias as mentioned above, after determining the most effective approach to limit non-responders, a sampling method (anticipated to be the “high intensity” approach) will be used to effectively and more efficiently yield sufficient completed surveys to permit program comparisons. To limit the cost of survey administration, future sampling methods will include thresholds to limit “over sampling” of large facilities beyond that needed for comparison. For smaller facilities, minimum sampling numbers will be determined to prevent “under sampling”. For example, a large facility that has 500 deaths per year may require only 100 completed surveys to permit comparative analyses whereas a smaller facility (with less than 100 deaths per year) would be unable to achieve 100 completed surveys and may be permitted a lower threshold of completed surveys to permit analysis.

The following items are provided in accordance with VHA Directive 2003-007:

4. Contact: Scott T. Shreve, DO. National Director, Hospice and Palliative Care, Lebanon VA Medical Center, Lebanon, PA. Office phone 717-228-5946.

5. Distribution of date of data collection: While the pilot survey results will only be reported anonymously (with each region or facility only knowing their own site code) these results will be distributed quarterly within their respective regions to promote quality improvement efforts. Assuming OMB approval of the survey occurs by July 31st, 2007, the first date of distribution would be January 31, 2008.

6. Objectives: This project will examine the effectiveness of two recruitment strategies in validating a bereaved family member survey instrument [state them simply here?]. Additionally, feasibility information (such as logs of calls attempted, letters sent and staff time) will be analyzed to identify opportunities for improved efficiency with ongoing feedback from participating facilities to determine the utility of centralized reporting of results. The validated survey results will serve as quality measures of the level of satisfaction of family members of veterans who have died in VA facilities. These results and the associated actions have the potential to be a driving force for ongoing improvements in the quality of end-of-life care provided in VA.

7. Number of respondents and burden: From the results of previous phone satisfaction surveys of family members of veterans that have died in VA facilities, an expected successful response rate of 50% is anticipated. For participating regions, a total of 3,600 people will be asked to respond to the survey over a one year period with an expectation that 1,800 respondents will complete the survey. Each person will complete the survey only once and the total burden time is calculated as follows: 1,800 respondents x 15 minutes per survey= 27,000 minutes/60= 450 hours.

Thank you for the opportunity to provide these clarifications. Please contact me should you have any questions.

CC: Dr. Tom Edes, Chief of Home and Community Based Care

Dr. David Casarett, Researcher and Palliative Care Physician, Philadelphia VAMC

Automated VA FORM 2105

Microsoft Word Version 6.0