Detecting Intraocular Pressure Harmonic Frequencies in Glaucoma and the Sampling Targets Needed to Identify Them With...

The purpose of this study was to determine the detectability and occurrence of circadian and ultradian rhythms in patient-acquired intraocular pressure (IOP) measurements with sampling targets for reliable harmonic detection.

A multi-harmonic non-uniform discrete Fourier transform (NDFT), with bootstrap amplitude confidence intervals and independent Lomb-Scargle periodogram (LSP) confirmation, was applied to 286 home-tonometry monitoring blocks from 137 eyes of 69 patients with glaucoma. Integer harmonic frequencies (F) from 1 to 10 cycles per day were tested. Characterized frequency was defined when noise-threshold significance, amplitude lower 95% confidence bound >0, and phase 95% confidence interval met prespecified criteria. Detection probability was modeled by a logistic generalized additive model (GAM) accounting for total observation count, clock-hour coverage, and the longest sparse-data gap.

The 24-hour rhythm (F1) was characterized in 35.7% of blocks; at least one of F1, F2, F3, or F4 was characterized in 44.8% (128/286). Of these 128 blocks, 34% (43) had 2 or more harmonics simultaneously characterized and 7% (9) had 3 or more. Detected periodicities were almost exclusively integer harmonics. Non-integer LSP peaks arose in only 3.1% of blocks (9/286). Where 15 or more clock-hours contained at least 2 observations and no more than 2 consecutive clock-hours contained one or none, approximately 100 to 130 observations over 4 weeks gave a model-predicted detection probability of 75% to 85%, rising to 90% at approximately 150 observations.

One or more characterizable 24-hour integer IOP harmonic rhythms could be detected with an appropriate self-tonometry protocol in nearly half of treated patients with glaucoma.