The difference between in bed and out of bed activity as a behavioral marker of cancer patients: A comparative...

The current study was conducted to provide normative data on actigraphic dichotomy index (I < O) (the percentage of in bed activity counts that are less than the median of out of bed counts) in healthy population and to assess whether the I < O could be an effective index in discriminating the circadian motor activity of cancer patients from healthy controls. In this retrospective study, we recovered 408 actigraphic records from two databases: healthy controls (n = 182; 79 males; mean age 38.7 ± 12.6) and patients with metastatic colorectal cancer (n = 226; 149 males; mean age 58.4 ± 11.4). Beside the usual actigraphic sleep parameters (time in bed, sleep onset latency, total sleep time, wake after sleep onset, sleep efficiency, number of awakenings, and mean motor activity), we also computed the dichotomy index and number of actigraphic wake parameters, namely, diurnal motor activity, diurnal total sleep time, number of sleep episodes, and the mean duration of the longest diurnal sleep episode. Using the Youden index, we calculated the cut off value that performed the best for I < O and actigraphic wake parameters. Finally, we created Receiver Operator Characteristic curves to test the efficacy of each actigraphic parameter to discriminate cancer patient from healthy controls. Mean I < O was 99.5% (SD, 0.48%) in the healthy group, as compared to 96.6% (SD, 3.6%) in the cancer group (p < 0.0001). Important age-related effects appeared unlikely after performing both the main analysis with age as a covariate, and a subset analysis in 104 subjects matched for age and sex. In the main analysis, all actigraphic parameters, except total sleep time, significantly differentiated the two groups of participants. However, the I < O was the one that clearly performed best. Here, we provide the first large dataset on I < O in healthy subjects, we confirm the relevance of this circadian index for discriminating advanced stage colorectal cancer patients from healthy subjects, and we lay the grounds for further investigations of this circadian index in patients with other chronic diseases.