TL;DR
This paper validates the integrated Alzheimer's Disease Rating Scale (iADRS) as a robust measure of cognitive and functional decline in early symptomatic Alzheimer's disease (AD). Using classical and Rasch psychometric analyses on data from three clinical trials, the iADRS demonstrates excellent reliability, validity, and sensitivity to change, outperforming its individual components.
What Problem Does This Paper Address?
Alzheimer's disease (AD) is characterized by a progressive decline in both cognition and daily functioning. Accurately measuring this decline is crucial for clinical trials evaluating new therapies, especially in the early symptomatic stages of AD. Existing measures often focus on either cognition or function separately, or are less sensitive to subtle changes. The iADRS was developed to address this gap by integrating cognitive and functional assessments into a single, more sensitive endpoint. However, rigorous psychometric validation of this integrated scale, particularly across multiple clinical trials and using advanced analytical methods, was needed to confirm its utility.
Methods
The study employed two primary psychometric methodologies:
- Classical Test Theory (CTT): This standard approach was used to assess the iADRS's internal consistency reliability (using Cronbach's alpha), construct validity (through correlations with established measures like the CDR-SB and MMSE), and sensitivity to change (by examining effect sizes and correlations with changes in other measures).
- Rasch Measurement Theory (RMT): This more advanced method was used to evaluate the iADRS's performance as a measurement continuum. RMT analyses assessed item targeting (how well the items match the patient population's ability range), response functioning (whether item response options are ordered logically), item fit (how well individual items conform to the Rasch model), and reliability (using the person-separation index, PSI). RMT was also used to create "crosswalks" – conversion tables that allow for direct comparison of scores between the iADRS, CDR-SB, and MMSE.
Datasets
The study utilized data from three double-blind, placebo-controlled clinical trials:
- TRAILBLAZER-ALZ: A Phase 2 trial of donanemab.
- TRAILBLAZER-ALZ 2: A Phase 3 trial of donanemab.
- EXPEDITION-3: A Phase 3 trial of solanezumab.
These trials included participants with early symptomatic AD who had confirmed amyloid pathology and met specific cognitive criteria (MMSE scores between 20-28 or 20-26, depending on the trial). The combined dataset comprised over 30,000 participant-visit records.
Key Insights and Ideas
- Integration is Key: Combining cognitive and functional items into the iADRS not only preserves the psychometric properties of the individual components but also enhances overall reliability and measurement precision.
- RMT Reveals Deeper Structure: RMT analysis provided a granular view of how cognitive and functional items are interwoven along the disease severity continuum, demonstrating that the iADRS effectively captures this integrated decline.
- Crosswalks Bridge the Gap: The development of crosswalk tables is a significant contribution, enabling researchers and clinicians to translate scores between the iADRS, CDR-SB, and MMSE, facilitating interpretation of clinical trial results and real-world patient assessments.
- Improved Reliability: The iADRS achieved a higher person-separation index (PSI = 0.92) compared to its constituent scales (ADAS-Cog: 0.88, ADCS-iADL: 0.85), indicating superior reliability in measuring the underlying construct.
Key Results
- Reliability: The iADRS demonstrated excellent internal consistency reliability across trials (Cronbach's alpha 0.79-0.84) and high reliability in RMT analyses (PSI = 0.92).
- Validity:
- Construct Validity: Moderate to strong correlations were observed between the iADRS and established measures like the CDR-SB (r = -0.64 to -0.70) and MMSE (r = 0.53 to 0.65), supporting its role as a measure of global disease severity.
- Sensitivity to Change: The iADRS showed large effect sizes (|ES|>0.80) for participants who worsened on CDR-SB and MMSE, indicating its ability to detect meaningful clinical decline.
- Targeting: RMT analysis showed that the iADRS items were well-targeted to the participant population, covering the full range of observed disease severity without significant floor or ceiling effects.
- Crosswalks: The study successfully created crosswalk tables that allow for direct score conversions between iADRS, CDR-SB, and MMSE, with adequate reliability (ICC ≥ 0.70) for most comparisons.
Limitations and Future Work
- Generalizability: The authors acknowledge that findings may not be fully generalizable to real-world clinical practice due to potential limitations in participant diversity (e.g., comorbidities, socioeconomic status, ethnicity) in clinical trials.
- Visit Schedule: The RCT design limited the ability to assess iADRS test-retest reliability due to the time intervals between visits.
- Response Frequencies: Low response frequencies for some item options, while expected, may have impacted item calibration and precision in RMT analyses.
- Disease Severity Range: The current characterization is primarily for early symptomatic AD. Future work should incorporate longer observation periods and a wider range of baseline disease severity to enrich the understanding of scale performance in later stages of AD.
- Clinical Practice Application: While crosswalks are provided, their application in clinical practice should be approached with the aforementioned caveats regarding generalizability.
Summary updated