Evaluating the Veteran's Health Administration (VHA) system service structure, processes, and outcomes is contingent upon an integrated information system that monitors a Veteran's functional status across the continuum of care (acute, post-acute and community care). Since VA health professionals use a variety of functional instruments, valuable information may be lost as Veterans move across health care settings. In particular, within the VHA, researchers and clinicians face challenges when comparing results from the Functional Independence Measure, which is collected to monitor functional outcomes in VA inpatient rehabilitation facilities (IRF), and results from the Minimum Data Set (MDS) of the Resident Assessment Instrument, which is collected to monitor functional status in VA Community Living Centers (CLC - previously referred to as Nursing Home Care Units). While tremendous resources have been invested in developing and implementing these useful instruments, data collected with the FIM cannot be readily compared to similar data collected with the MDS.
The long-term objective of this proposed research is to provide the VA with a state-of-the-art measurement system that builds upon the well-established VA infrastructure use of the FIM and the MDS. The proposed measurement system will provide practitioners with: 1) measures that translate between the two instruments, 2) flexibility in test administration (e.g., practitioners can administer either the FIM or MDS in their entirety or in abbreviated forms), 3) reduced administration and Veteran burden using short forms and computerized adaptive test (CAT) administration, and 4) maximize measurement precision across patient diagnosis severity. The Specific Aims are to: 1) Create a FIM-MDS item bank that meets item response theory model (IRT) requirements 2) Generate IRT-based short forms and CATs from the item bank 3) Compare the precision of the IRT-based short forms, CATs and item bank to the original FIM and MDS measures 4) Assess the accuracy of the IRT-based short forms, CATs and item bank in classifying Veterans into Function Related Groups.
We will obtain a final linked FIM-MDS data from the Austin Information Technology Center (AITC) for 3000 Veterans. The dataset will be randomly divided into two datasets, one to develop the IRT-based short forms and CATs and one to test the precision and validity of these instruments. The final short forms and CATs will be designed to meet the criteria of unidimensionality, model fit, monotonicity, local independence and differential item functioning. The IRT-based short forms and CATs will be designed to reduce respondent/administrator burden and maximize precision across the breath of the item bank. We will compare the precision of the different short forms, CATs and item bank by examining the information functions for each administration procedure. We will investigate the accuracy of the short forms, CATs and item bank in classifying Veterans into Function Related Groups. The final product of this three-year study will be a state-of-the-art measurement system that can be used to monitor Veterans across IRF and CLC continuum of care.
This project successfully 1) examined dimensionality and item-level psychometric properties of an item bank measuring activities of daily living (FIM_MDS: Functional Independence Measure and Minimum Data Set) across inpatient rehabilitation facilities and community living centers, 2) compared measurement accuracy of 4- ,8- and 13-item short forms generated from the self-care FIM_MDS and the 4- and 8-item short forms from a motor FIM_MDS. This project confirmed that a FIM_MDS is an appropriate item bank for developing efficient test forms such as short forms and computerized adaptive tests. Also, when using the self-care FIM_MDS, maintaining the number of the items is recommended to generate a continuum of care measurement, and for motor FIM_MDS, using 8-item short forms is suggested for the best balance between precision and efficiency.
We demonstrated modern psychometrics can be used to efficiently and precisely link existing instruments that measure the same construct, but may do so utilizing unique items and rating scales. Most efforts to standardize measurement across health care systems have focused on the development of new measures in the hope that entities adopt these new measures. This approach is costly, time consuming and burdensome for providers and administrators that need to learn a new measurement system. Our approach has the advantage of allowing users to continue to measure patients using existing measures, but getting their results in a metric that is shared with the linked measure. Specifically, we created a crosswalk between the FIM daily activities and MDS counterpart that shares the same metric and allows for direct comparison and follow-up. We also demonstrate efficient ways to reduce measurement burden while maintaining precision by utilizing short forms (4 and 8 items). Overall, we provide a method for following Veterans across the continuum of care utilizing a common metric (based on FIM and MDS) to track their functional ability in performing basic tasks.
External Links for this Project
Grant Number: I01HX000670-01
- Li CY, Romero S, Simpson K, Simpson A, Bonilha H, Hong I, Velozo CA. Linking Existing Instruments to Develop A Continuum of Care Measure: Accuracy Comparison using Functional Related Group Classification. Quality of life research : an international journal of quality of life aspects of treatment, care and rehabilitation. 2017 May 24; 1-10. [view]
- Li CY, Romero S, Bonilha HS, Simpson KN, Simpson AN, Hong I, Velozo CA. Linking Existing Instruments to Develop an Activity of Daily Living Item Bank. Evaluation & the health professions. 2018 Mar 1; 41(1):25-43. [view]
- Li CY, Romero S, Simpson K, Simpson A, Bonilha H, Hong I, Velozo CA. Comparing Measurement Accuracy of Short Forms: Individual Point Difference and Functional Related Group Classification. Poster session presented at: American Occupational Therapy Association Annual Conference; 2016 Apr 7; Chicago, IL. [view]
- Li CY, Romero S, Simpson A, Bonilha H, Simpson K, Hong I, Velozo CA. Comparing measurement precisions of short forms: Measuring self-care physical function in Veterans. Paper presented at: American Occupational Therapy Association Annual Conference; 2016 Apr 7; Chicago, IL. [view]
- Lee MJ, Romero S, Pei Q, Cowper Ripley DC, Jia H. Create an Interval ADL Scale with Rasch Analysis. Poster session presented at: University of Florida College of Public Health and Health Professions Annual Research Day; 2016 May 20; Gainesville, FL. [view]
- Sood P, Romero S, Velozo CA, Gruber-Baldini A, Shulman L. Defining Self-Efficacy for the Self-Management of Chronic Conditions and Identifying Priority Sub-domains: Web-based DELPHI Technique. Poster session presented at: American Congress of Rehabilitation Medicine Annual Meeting; 2015 Nov 15; Dallas, TX. [view]
- LI CY, Romero S, Bonilha H, Simpson K, Simpson A, Hong I, Velozo CA. Developing continuum of care assessment across post-acute care in Veterans. Poster session presented at: American Occupational Therapy Association Annual Conference; 2016 Apr 7; Chicago, IL. [view]
- Lee MJ, Romero S, Pei Q, Cowper Ripley DC, Jia H. Using Rasch Analysis to Create an Interval ADL Scale. Poster session presented at: VA North Florida / South Georgia Veterans Health Care System Annual Research Day; 2016 May 20; Gainesville, FL. [view]
- Velozo CA, Romero S. Using Rehabilitation Measures to Generate Medicare G-Codes and Guide Clinical Interventions. Presented at: American Congress of Rehabilitation Medicine Annual Meeting; 2013 Nov 16; Orlando, FL. [view]
Aging, Older Veterans' Health and Care, Health Systems