Supporting Statement Part B -- Adopting Best Practices for Medicaid Readmissions 11-26-2012

SUPPORTING STATEMENT

Part B

Adapting Best Practices to Reduce Avoidable Hospital Readmissions for Medicaid Patients

Version: November 26, 2012

Agency of Healthcare Research and Quality (AHRQ)

Table of Contents

B. Collections of Information Employing Statistical Methods 1

1. Respondent Universe and Sampling Methods 1

2. Information Collection Procedures 2

3. Methods to Maximize Response Rates 3

4. Tests of Procedures 3

5. Statistical Consultants 3

B. Collections of Information Employing Statistical Methods

In that the focus of this project is on quality improvement, a generalizable sample is not proposed and statistical analysis methods are not employed. The participating hospitals are responsible for conducting and analyzing the primary data. The data collected will be shared only at an aggregate level with other hospitals and states involved in the study, as well as with AHRQ and the JSI project team.

1. Respondent Universe and Sampling Methods

Three safety net hospitals are involved in this project: one in Pennsylvania, one in Massachusetts, and one in Texas. All three serve large populations of Medicaid-insured patients and all three have a strong interest in better managing avoidable readmissions in this population group.

For each wave of data collection each hospital will be asked to administer the Diagnostic Interview Tool the next 10 Medicaid-insured patients who were readmitted within the last 30 days in a particular category, with the category reflecting each hospital’s specific circumstances and readmissions priorities. For example, one hospital may be working on readmissions for its Medicaid patients with heart disease. This hospital would select the next 10 Medicaid-insured patients who were readmitted within 30 days to its cardiology unit. Another hospital may focuse on readmissions for non-English speaking patients. This hospital would select the next 10 Medicaid-insured patients who were readmitted within 30 days and who do not speak English as their primary language regardless of medical diagnosis. Thus, the process of patient selection is purposive and tailored to each hospital’s specific readmissions focus. The patient sampling procedure supports the quality improvement process of rapid cycle testing that is the intent of this project.

JSI will work with each hospital to determine the most efficient source for identifying its candidate patients, which may vary from hospital to hospital. For example, a common process used in the STAAR project (from which this data collection process is adapted) is to have hospitals review admission logs at the end of each day when primary data collection is underway to determine readmission status. Readmissions occurring within 30 days of the index admission are flagged. Hospital staff are instructed to attempt interviews the next day on all flagged cases until 10 cases have been assessed or until hospital staff are seeing a convergence of patterns of response.

Hospital staff will use the tools developed by the project to collect information on Medicaid patients who have recently been readmitted to the hospital. Hospital staff will comply with HIPAA and hospital specific privacy regulations in obtaining patient or family/caregiver consent to participate in the interview. The hospital staff will aggregate the information and provide a summary report to the project team. No identifiable patient data will be included in the report. Neither the contractor (JSI) nor AHRQ staff will be involved directly in collecting the patient/family/caregiver or provider assessments or aggregating the results. Hospital staff will be instructed to store the data in a secure environment that can only be accessed by hospital quality improvement staff involved in the project, and relying on procedures that are in compliance with HIPAA.

2. Information Collection Procedures

Both the Diagnostic Interview Tool – Driver Assessment and the Diagnostic Interview Tool – Strategy Assessment has three components: 1) medical records review, 2) patient/family/caregiver interviews, and 3) provider interviews. A nurse (or professional equivalent) from each hospital’s quality improvement team is assigned to collect the data through the Diagnostic Interview Tool.

The purpose of the medical records review component of the Diagnostic Interview Tool is to gather background information about the index admission and the readmission to be able to engage in an informed discussion with the patient, family/caregivers, and providers involved. The staff member collecting the data will review the electronic medical record of the selected patient case and record the results on a paper questionnaire.

After completion of the medical records review, component two and three of the Diagnostic Interview Tool will be completed. Component two consists of interviews with the patient and a family member or caregiver. The quality improvement staff member conducting the interviews is instructed to attempt to interview both the patient (if possible given health status and if the patient is of majority age) and a family member or caregiver who has permission to discuss the patient’s case. If needed, interpreter services will be secured through the hospital’s usual processes. Responses will be recorded on a paper questionnaire as the interview is underway. Hospitals will be instructed to try to conduct the interview while the patient is in the hospital. If this is not possible, the data collector will attempt at least two times to reach the patient and/or caregiver by telephone within 48 hours of discharge.

The third component of the data collection process will be to conduct interviews with providers involved in the initial hospitalization and the subsequent readmission. This may include the discharging hospitalist in the index admission, the admitting hospitalist on the readmission, the emergency room physician on the readmissions, the primary care physician, a home care nurse, the nurse in the skilled nursing facility who referred the patient to the emergency department, or others, as specific to the case. The quality improvement staff member conducting the interviews is instructed to attempt to reach at least two members of the provider team per case while the patient in in hospital or within 48 hours of discharge. Responses will be recorded on paper questionnaire as the interview is underway.

The coaching for the project, including the administration of the Diagnostic Interview Tool, relies on general principles of quality improvement science and specifically the processes developed by the IHI, which have been adopted widely by the Center for Medicare and Medicaid Services. The coaching also draws upon the 140-page how-to guide that Dr. Boutwell developed for the IHI’s STAAR program. Coaching and instruction will be provided by the JSI team to hospital staff who will be conducting these interviews over the telephone. Additionally, JSI staff will be available by phone should questions arise during this process.

3. Methods to Maximize Response Rates

In that the diagnostic interview process is part of a quality improvement process rather than a research protocol, there is no set number of interviews required. Participating hospitals are instructed that it is a process that should continue until patterns of response converge and little new information is being learned, with 10 cases as the maximum during any one wave.

It is anticipated that it will be more difficult to find patients after discharge; thus, the JSI team will encourage that interviews be conducted on patients during their hospital stay whenever possible. In-hospital interviews may not be feasible in all cases; thus, participating hospitals are instructed to conduct the patient and/or caregiver and provider interviews within 48 hours of discharge for those not reached during their hospital stay.

4. Tests of Procedures

The diagnostic interview process being used in this study is a similar process to that used by the STAAR initiative. The Co-Principal Investigator of JSI’s project team, Dr. Amy Boutwell, is the founder and original co-principal investigator of the STAAR initiative and has worked with dozens of hospitals to reduce readmissions. The primary data collection and interview process derives from this work; and, thus the process and tool is informed by the hospitals involved with the STAAR initiative, with adjustments made to address Medicaid issues.

The tools were revised to incorporate Medicaid specific issues, such as socio-demographic considerations more prevalent in a Medicaid population (e.g., literacy, language, housing stability), diagnoses more prevalent in Medicaid population (e.g., mental health and substance abuse), and resources and transitional care settings more applicable to Medicaid-insured populations (e.g., temporary or unstable housing, social service connections, safety net ambulatory settings). These items were added based on discussions with AHRQ, the project’s advisory board, and the hospital teams as well as the secondary data analysis conducted by the hospitals to determine their drivers of Medicaid readmissions.

5. Statistical Consultants

The JSI project team will coach each of the participating hospitals in the use and qualitative analysis of the Diagnostic Interview Tool. JSI’s Co-Principal Investigator, based on her experience with the STAAR initiative, will train the participating hospitals and provide any consultation needed around the primary data collection and analysis. The contact information for the Co-PI is provided below:

Amy Boutwell, MD, MPP, Co-PI

Collaborative Healthcare Strategies

(617) 710-5785

Page | 2