Comparison of activity levels derived from two accelerometers in dogs with osteoarthritis: Implications for clinical trials S. Mejia, F.M. Duerr*, M. Salman Department of Clinical Sciences, James L. Voss Veterinary Teaching Hospital, Colorado State University, Fort Collins, CO, 80525, USA A R T I C L E I N F O Keywords: Accelerometer Activity Dog Gait Osteoarthritis A B S T R A C T Accelerometer measurements are frequently reported as total weekly activity counts (AC). Methods of utilizing activity parameters to allow differentiation of activity intensities (i.e., manually derived cut- points) have been described. While this information may provide valuable data for researchers, only few investigators have utilized these methods. This may, in part, be due to the challenge associated with data processing. Some devices, however, generate activity intensity data automatically. This study was conducted to evaluate a novel accelerometer that allows for remote download of data via Wi-Fi (Heyrex), to compare automatically generated parameters quantifying activity levels to previously established cut- points (Actical) and to describe the variability of accelerometer data in dogs with osteoarthritis. Twelve client-owned dogs with osteoarthritis were fitted with a collar with two accelerometers (Heyrex and Actical). Accelerometer data were recorded for 28 days. Pearson bivariate correlations and coefficient of variation (CV%) were calculated for accelerometer data. There was a strong correlation between the AC reported by Heyrex and Actical devices. Several automatically generated parameters showed strong correlations to previously validated cut-points and displayed lower CV% than weekly AC. The activity intensity derived from the Heyrex was associated with the lowest CV% of all parameters from both accelerometers. Automatically generated activity intensity parameters should be considered as potential outcome measures in clinical trials for dogs with osteoarthritis. This novel technology may allow for convenient acquisition of activity intensity data in companion animals. However, technical difficulties (e.g., lack of Wi-Fi connectivity) may pose challenges when using this novel device. Published by Elsevier Ltd. Introduction Researchers currently use various subjective and objective outcome measures to determine the treatment efficacy for canine osteoarthritis. To date, no single outcome measure that can reliably detect changes in functional disability in these patients has been identified (Belshaw et al., 2016). A frequently used objective outcome measure in small animal osteoarthritis research is the quantification of patient activity in total weekly activity counts (AC) measured via accelerometers (Brown et al., 2010a; Rialland et al., 2012; Belshaw et al., 2016). Accelerometers have been used in veterinary medicine to objectively track activity rhythms (Siwak et al., 2003; Gerencsér et al., 2013; Zanghi et al., 2013), evaluate correlations between patient characteristics and physical activity (Brown et al., 2010b; Morrison et al., 2014), quantify stress levels (Jones et al., 2014), and even monitor levels of pruritus (Schwab-Richards et al., 2014). While previous studies have validated the use of different accelerometers in monitoring physical activity in dogs for clinical research (Yam et al., 2011; Preston et al., 2012; Walton et al., 2013), only a limited amount of currently available devices have been investigated scientifically (Yashari et al., 2015; den Uijl et al., 2017; Belda et al., 2018). The Actical is a commercially available, omnidirectional accelerometer designed for research purposes, which has been validated by several investigators to monitor at-home physical activity in in companion animals (Hansen et al., 2007; Lascelles et al., 2008). Methods of utilizing data provided by this device to allow differentiation of varying physical activity intensities (e.g., sedentary, walking, trotting) have been established (Michel and Brown, 2011). The Actical device also generates several activity intensity output parameters automatically without the need to perform further calculations. While activity intensity information appears to be of clinical value for researchers, to the authors’ knowledge, only two research groups have utilized this or similar methodologies to calculate activity intensity data (Michel and * Corresponding author. E-mail address: felix.duerr@colostate.edu (F.M. Duerr). http://dx.doi.org/10.1016/j.tvjl.2019.105355 1090-0233/Published by Elsevier Ltd. The Veterinary Journal 252 (2019) 105355 Contents lists available at ScienceDirect The Veterinary Journal journal homepage: www.else vie r.com/locate /t vjl