Table 2. Comparison of inferred trend from individual surveillance data and from model based on GOPC ILI rate, GP ILI rate and school absenteeism rate with changes in influenza activity^*.

	correlation† with % change in influenza activity between the subsequent and preceding week
surveillance data	pre-pandemic period (Jan 2004–May 2009)	pandemic period (mid-Jun–Dec 2009)	whole period (Jan 2004–Dec 2009)
1. GOPC ILI‡	0.42	-	0.42
2. GP ILI	0.30	0.11	0.24
3. School absenteeism§	0.45	−0.13	0.24
Inferred influenza trend from model based on 1+2+3	0.42	0.28	0.38
Inferred influenza trend from 1+2+3+DFC fever counts	-	0.29	0.38

DFC, designated fever clinic; GOPC, general outpatient clinic; GP, general practitioner; ILI, influenza-like-illness.

^*Influenza activity measured by GP ILI consultation rate×laboratory influenza isolation rate.

^†Correlations between surveillance data and the log ratios were calculated by fitting a univariate dynamic linear model to each data stream, and an overall multivariate model to all data streams. DFC was excluded from the analysis due to insufficient data for estimation of the inferred trend.

^‡GOPC data were interrupted during the pandemic period due to the opening of designated flu clinics.

^§School absenteeism data were occasionally interrupted by school holidays or school closures. Data during the summer holidays were excluded.