Date Published: April 3, 2018
Publisher: BioMed Central
Author(s): Catherine Reed, Mark Belger, Grazia Dell’Agnello, Kristin Kahle-Wrobleski, Gopalan Sethuraman, Ann Hake, Joel Raskin, David Henley.
Comparison of disease progression between placebo-group patients from randomised controlled trials (RCTs) and real-world patients can aid in assessing the generalisability of RCT outcomes. This analysis compared outcomes between community-dwelling patients with mild Alzheimer’s disease (AD) dementia from two RCTs (pooled European (EU) data from EXPEDITION and EXPEDITION 2) and similar patients from the EU GERAS observational study.
Data from placebo-group patients with mild AD dementia from the RCTs (EU countries only) were compared with data from GERAS patients with mild AD dementia. Between-group differences for changes over 18 months were analysed for cognition, functioning, neuropsychiatric symptoms, health-related quality of life (HRQoL) and caregiver time using propensity score-adjusted models. A sensitivity analysis compared EU/North American (EU/NA) EXPEDITION patients with GERAS patients.
EU EXPEDITION patients (n = 168) were younger than GERAS patients (n = 566) (mean (standard deviation, SD) age 71.9 (7.4) versus 77.3 (6.9) years; p < 0.001) and were more likely to use AD treatment (95% versus 84%; p < 0.001). Cognitive performance was similar at baseline in both populations, although GERAS patients showed greater functional impairment (p = 0.005) and lower HRQoL (p < 0.05). At 18 months, no statistically significant differences between EXPEDITION (n = 133) and GERAS (n = 417) patients were observed for changes in cognitive, functional, neuropsychiatric and HRQoL outcomes. Least squares mean (95% confidence interval) change in caregiver time (hours/month) spent on instrumental activities of daily living (iADL; 29.22 (19.16, 39.27) versus 3.20 (−11.89, 18.28), p = 0.001) and supervision (66.59 (47.49, 85.69) versus 3.04 (−25.39, 31.48), p < 0.001) showed greater increases in GERAS than EXPEDITION. In the sensitivity analysis, changes in neuropsychiatric and HRQoL scores and caregiver time spent on basic ADL were also significantly greater in GERAS than in EU/NA EXPEDITION patients. Patients with mild AD dementia participating in the EU EXPEDITION RCTs and the GERAS observational study showed a similar decline in cognitive, functional and neuropsychiatric symptoms over 18 months, whereas changes in caregiver time measures were significantly greater in GERAS. Results indicate the importance of using similar regions when comparing real-world and RCT data. ClinicalTrials.gov NCT00905372 EXPEDITION. Registered 18 May 2009.
Randomised controlled trials (RCTs) are the gold standard for assessing treatment efficacy  and are designed with internal validity as a priority. Although internal validity is important in a trial setting, RCTs are criticised for their lack of external validity or generalisability . The key issue with respect to this is that trial populations are recruited using extensive exclusion criteria, which aim to select a homogeneous population with few comorbidities; this does not always represent the more heterogeneous patient populations (potentially with many comorbidities) in clinical practice. There may also be differences between RCT protocols and routine practice (e.g. diagnostic methods, treatments used and timing of treatment) and the setting of the trial (e.g. country, healthcare system or primary versus secondary care), all of which can influence external validity .
Our analysis showed similar results between RCT and real-world data for changes in cognitive, functional and neuropsychiatric symptoms over 18 months in patients with mild AD dementia, after controlling for baseline differences.
Using a propensity score matching approach, a similar decline in cognition and functioning was observed between results from RCT and observational study patients in the EU population. Our findings support that RCTs and observational studies can provide complementary data to assess longitudinal disease progression in patients with mild AD dementia. Confounding factors inherent within the different study designs and inclusion criteria mean that some findings require additional explanation; for example, the increase in caregiver time over 18 months in the GERAS study but not during the RCTs may have resulted from differences in the type of care received or baseline differences not accounted for in propensity matching. Use of similar geographic regions is important when comparing real-world and RCT data to ensure that differences between studies are not simply due to country-specific variations. Further opportunities to assess the comparability of real-world and RCT data will arise when data from additional countries in the GERAS study (USA and Japan) are available. These will help inform which parameters and outcomes are culturally/geographically dependent.