TABLE 3.

Statistical Significance of the Influence of the Social Variables on Adherence, Controlling by Confounding Factor, for Different Event Detection Settings

Social variable, controlled confounder	Average difference	N	P
Fit to night data
W1M
Worker	0.191	259	<.001
Hospital room	0.209	337	<.001
No control	0.199	47,703
SRSSI
Worker	3.740	260	<.001
Hospital room	5.700	337	<.001
No control	−1.279	47,703
Fit to day and night data
W1M
Worker	0.191	256	<.001
Hospital room	0.217	337	<.001
No control	0.191	47,694
SRSSI
Worker	3.618	257	<.001
Hospital room	6.082	337	<.001
No control	−1.519	47,694
Strict identification of hand hygiene opportunities and dispenser activationa
W1M
Worker	0.251	227	<.001
Hospital room	0.455	294	<.001
No control	0.305	21,206
SRSSI
Worker	5.125	232	<.001
Hospital room	11.718	297	<.001
No control	1.927	21,206

NOTE. Opportunities and dispenser activations are detected differently under different settings. Each row reports the average increase of the social variable upon adherence (average difference column), the number of pairs after matching by confounder (N), and the statistical significance of the increase assessed using Wilcoxon matched-pairs signed-rank test (P value). For reference, the noncontrolled average differences have been added as well; in such cases, the total number of opportunities, rather than the number of pairs, is given. The P value was not reported if insignificant (2-tailed t test, P > .1). SRSSI, sum of received signal strength to peers at the moment of the opportunity; W1M, the number of encounters within 1 minute of the opportunity.

^aSee Monsalve et al.²³