Skip to main content
. Author manuscript; available in PMC: 2015 Mar 1.
Published in final edited form as: Infect Control Hosp Epidemiol. 2014 Jan 24;35(3):243–250. doi: 10.1086/675292

Table 4. Time Series Models for Staphylococcus aureus Acquisition Rates in the Intervention Intensive Care Unit (ICU) (2002-2007) and Control ICU (2005-2007).

Intervention (Surgical) ICUa Control (Medical) ICU
Crude rates Rate (per 1,000 pt-days at riskb) Rate (per 1,000 pt-days at riskb)
Pre-intervention 19.73 19.21
Post-intervention 15.63 15.33
Percent change −20.78% −20.18%

Time Series Modelb Beta [95% CI] P Beta [95% CI] p
Intervention −4.10[−7.25, −0.95] 0.012 −6.75[−20.75, 7.26] 0.334
Adjusted S. aureus
   colonization pressure
49.13 [26.21, 72.04] 0.000 25.96 [−15.38, 67.31] 0.210
AR(1) 0.24 [−0.13, 0.61] 0.191
MA(4) −0.38 [−0.74, −0.02] 0.039
Constant 2.40 [−5.99, 10.79] 0.569 12.50 [−9.77, 34.76] 0.261

Model parameters
Adjusted R2 0.28 0.10
Durbin-Watson statistic 1.90 1.99
Q(4) p-value 0.83 0.81
Q(8) p-value 0.87 0.87
Q(12) p-value 0.87 0.88

Note: AR(1)= auto-regressive variable, 1; MA(4)= moving-average variable, 4.

The inclusion of AR and MA terms in the model addressed serial correlation seen, as evidenced by the Durbin-Watson statistic and the Box-Ljung Q(k) tests of the unadjusted model.

a

Models are based on all available data for the surgical ICU (61months, 2002-2007) and medical ICU (24 months, 2005-2007).

b

See study methods for definition of patient-days at risk.

c

In the multivariate time series model, ‘beta’ indicates the magnitude and the direction of the variable in the model, while the p value and 95% CI indicate the precision and significance of the variable within the model.