imputation procedures that rely on sample group or population data ( Kang et al., 2009 ; Kang et al., 2005 ; Zhuang et al., 2013 ). Alhassan et al.’s ( 2008 ) Within-Minute Average The current study included Alhassan et al.’s ( 2008 ) accelerometer data cleaning protocol. This method is not an imputation
Hotaka Maeda, Chris C. Cho, Young Cho and Scott J. Strath
Kelly R. Evenson, Amy H. Herring and Fang Wen
Few studies measure physical activity objectively or at multiple time points during postpartum. We describe physical activity at 3- and 12-months postpartum among a cohort of women using both self-reported and objective measures.
In total, 181 women completed the 3-month postpartum measures, and 204 women completed the 12-month postpartum measures. Participants wore an ActiGraph accelerometer for 1 week and completed in-home interviews that included questions on physical activity. A cohort of 80 women participated at both time points. Poisson regression models were used to determine whether physical activity differed over time for the cohort.
For the cohort, average counts/minute were 364 at 3-months post-partum and 394 at 12-months postpartum. At both time periods for the cohort, vigorous activity averaged 1 to 3 minutes/day, and moderate activity averaged 16 minutes/day. Sedentary time averaged 9.3 hours at 3-months postpartum and 8.8 hours at 12-months postpartum, out of a 19-hour day. Average counts/minute increased and sedentary behavior declined from 3- to 12-months postpartum.
Interventions are needed to help women integrate more moderate to vigorous physical activity and to capitalize on the improvements in sedentary behavior that occur during postpartum.
Emily Borgundvaag, Michael McIsaac, Michael M. Borghese and Ian Janssen
participants reduces study power and would compromise the validity of the findings if the participants who are excluded because of excessive nonwear time are systematically different from the participants who are included. Imputation has been used, albeit on a limited basis, as an analytical approach to
Nathan Parker, Darran Atrooshi, Lucie Lévesque, Edtna Jauregui, Simón Barquera, Juan Lopez y Taylor and Rebecca E. Lee
Obesity is a critical problem among Mexican youth, but few studies have investigated associations among physical activity (PA) modes and anthropometrics in this population. This study examined associations among active commuting to school (ACS), sports or other organized PA, outdoor play, and body mass index (BMI) percentile and waist circumference (WC) among Mexican youth.
Parents of school children (N = 1996, ages 6 to 14 years, 53.1% female) in 3 Mexican cities reported PA participation using the (modified) fourth grade School Physical Activity and Nutrition Survey. Trained assessors measured BMI percentile and WC in person.
Parents reported that 52.3% of children engaged in ACS, 57.3% participated in sports or organized PA, and a median of 2 days in the previous week with at least 30 minutes of outdoor play. In complete case analyses (n = 857), ACS was negatively associated with BMI percentile, and outdoor play was negatively associated with WC after adjusting for school, age, sex, and income. In analyses incorporating data from multiple imputation (N = 1996), outdoor play was negatively associated with WC (all Ps < . 05).
ACS and outdoor play are favorably associated with anthropometrics and may help prevent childhood obesity in Mexico. ACS and outdoor play should be priorities for increasing youth PA in Mexico.
PA levels discarded 26.0% to 28.6% of the data. We suggest practitioners to use the within-minute average method for adults over 18 years of age, and the day-level imputation method for children and teenagers. To note, the day-level imputation method may be unstable for sample sizes of less than 50
Anders Raustorp and Andreas Fröberg
with self-reported values of <1000 or >30,000 steps were excluded from further analysis as previously recommended. 25 During the 5 times of measurement, several participants had incomplete data. If data from time 4 and time 5 were missing, imputation was conducted 26 in 2 ways. First, the last
Kent A. Lorenz, Hans van der Mars, Pamela Hodges Kulinna, Barbara E. Ainsworth and Melbourne F. Hovell
imputation procedure in SAS 9.3, with 20 imputation cycles performed for each area. The imputed estimates were averaged to produce plausible replacement values that were merged with the observed data. This allowed us to analyze a complete dataset with a combination of observed data and plausible imputation
David B. Creel, Leslie M. Schuh, Robert L. Newton Jr, Joseph J. Stote and Brenda M. Cacucci
variables to be entered into the analyses were estimated using multiple imputation. The dataset included complete data for 89.5% of participants and 98.4% of data points before imputation. To examine categorical relationships between exercise capacity, weight, and age, patients were divided into quartiles
Anass Arrogi, Astrid Schotte, An Bogaerts, Filip Boen and Jan Seghers
analyses. Missing data exceeded the level at which results can be biased (>5% missing). 27 Therefore, imputations were performed to minimize bias and preserve power. 28 Multiple imputation by chained equations was used to replace missing data ( m = 20 imputations). The procedure of chained equations is
Koren L. Fisher, Elizabeth L. Harrison, Brenda G. Bruner, Joshua A. Lawson, Bruce A. Reeder, Nigel L. Ashworth, M. Suzanne Sheppard and Karen E. Chad
participants, resulting in a mean of four participants per neighborhood cluster. Finally, a sensitivity analysis was conducted to determine whether imputation for participants with differing amounts of missing data would have an effect on the results. The multivariate analysis was repeated using imputed values