• 沒有找到結果。

Chapter 6 Discussion

6.5 Limitations

While this study benefits from the large sample size and longitudinal time trend decomposition analysis and Bayesian dynamic linear model, there are some limitations that are noteworthy. First, the database was obtained from the electronic hospital registration and infection control system. The initial electronic database lack individual antibiotics dosing documentation and disease severity index for non-intensive care unit patients, which might affect the exact relative risk estimates.

118

Secondly, this study was a metropolitan teaching hospital-based study, the generalization of specific results such as relative risk or intervention efficacy to other levels of healthcare institutions may be limited. Further study was needed for the external validation. Third, the HAI definition was modified in 2008 during the study period of twenty years, and the inclusion criteria was changed accordingly by the hospital central infection committee. The absolute incidence and the absolute interventions efficacy might be affected, especially in urinary tract infection and bacteremia. Their monthly representation was still useful for evaluating trends. Fourth, we did not include covariates such as patient comorbidities into the model. The most related covariates was departments of admission and infection sites, which represented part of major comorbidities of the patients in this study. In the time series study, the autoregressive order was the proxy for the history information unavailable.

Fifth, the follow-up time for the intervention evaluation may be short, especially for Bundle care intervention. Finally, the model fitting diagnostics were not done, further studies may be needed.

In conclusion, my thesis here developed a novel Bayesian generalized linear mixed ARMA model to monitor and evaluate the long-term time series on monthly

119

frequencies of HAIs in association with the impact of a set of interventions, the effects of time trend, seasonal variation, autoregressive order and innovation (moving average), and personal characteristics (age and gender) taking in account hierarchical correlated data property. This approach can be easily applied to forecasting the outcome of long-term time-series data and can be used for evaluation of the efficacy of intervention programs in the absence of randomized controlled trial design.

120 TABLES

Table 5.1. 1 Incidence rate by calendar year Year patient-days

121

Table 5.1. 2 The incidence rate of HAI by different factors and microbes Variable Classification Incidence rate (95% CI)

Age 0-9 1.27(1.14, 1.41)

Gender Female 4.22(4.13, 4.30)

Male 4.35(4.26, 4.43)

Department CV 3.61(3.38, 3.84)

Chest 7.52(7.21, 7.83)

Season Spring 4.46(4.33, 4.58)

Summer 4.57(4.44, 4.70)

Autumn 4.22(4.10, 4.34)

Winter 3.91(3.79, 4.02)

Infection site Bacteremia 1.12(1.09, 1.15)

RTI 0.72(0.69, 0.74)

SSI: surgical site infection; UTI: urinary tract infection; GI: gastrointestinal system;

SST: skin and soft tissue; EENT: eye, ear, nose, throat, or mouth infection

122

Table 5.1. 3 Univariate and multi-variable analysis for HAIs incidence

Variable Classification Univariate analysis Multi-variable analysis RR(95%CI) p-value aRR(95%CI) p-value

Admission type Emergency/ OPD 2.17(2.11, 2.23) <0.0001 1.60(1.55, 1.65) <0.0001

Infection site Bacteremia / SSI 2.40(2.28, 2.52) <0.0001 2.40(2.28, 2.52) <0.0001 PNEU / SSI 1.57(1.49, 1.66) 1.57(1.49, 1.66) CV: cardiovascular, GI: Gastrointestinal unit, PNEU: pneumonia, SSI: surgical site

123

infection, UTI: urinary tract infection, SST: skin and soft tissue, EENT: eye, ear, nose, throat, or mouth

124

Table 5.2. 1 The seasonal effect on HAIs incidence (univariate analysis)

Season RR 95% CI p value

Spring /Winter 1.13 1.09 1.18 <.0001

Summer /Winter 1.16 1.11 1.21

Autumn /Winter 1.07 1.03 1.12

125

Table 5.2. 2 Time trend analysis with de-seasonalized HAIs

Method Model Trend Estimate SD R2 p value

Original Model a Linear 2.790E-03 1.660E-03 0.0362 0.0941 Quadratic -2.341E-05 1.239E-05 0.0602

Cubic -5.777E-07 2.183E-07 0.0087

Model b Linear -1.210E-03 6.868E-04 0.0091 0.0785 Quadratic -1.475E-05 1.212E-05 0.2250

Model c Linear -1.070E-03 6.767E-04 0.0069 0.1165 Excluding Model a Linear 2.610E-03 1.690E-03 0.0365 0.1241

Outliers* Quadratic -2.287E-05 1.259E-05 0.0707

Cubic -5.660E-07 2.221E-07 0.0115

Model b Linear -1.320E-03 7.007E-04 0.0111 0.0605 Quadratic -1.466E-05 1.233E-05 0.2357

Model c Linear -1.180E-03 6.916E-04 0.0091 0.0884 Model a: time trend analysis using residual for linear, quadratic, and cubic

Model b: time trend analysis using residual for linear, and quadratic Model c: time trend analysis using residual for linear

*excluding outliers : May-95, Jun-96, Feb-97, Dec-99, Nov-00

126

Table 5.2. 3 De-seasonalization and de-trend time series of HAI incidence Autoregressive Coefficient SD p value

AR(1) -0.0982 0.0688 0.1532

AR(2) 0.0775 0.0683 0.2565

AR(3) 0.1068 0.0682 0.1175

AR(4) 0.0307 0.0678 0.6506

AR(1) -0.0846 0.0686 0.2174

AR(2) 0.0819 0.0686 0.2321

AR(3) 0.1254 0.0678 0.0644

AR(1) -0.0735 0.0689 0.2860

AR(2) 0.0833 0.0681 0.2214

AR(1) -0.0794 0.0679 0.2426

127

Table 5.2. 4 The seasonal effect on bacteremia HAIs incidence (univariate analysis)

Season RR 95% CI p value

Spring /Winter 1.16 1.07 1.25 <.0001

Summer /Winter 1.19 1.10 1.29

Autumn /Winter 1.08 1.00 1.17

128

Table 5.2. 5 Time trend analysis with de-seasonalized residual for bacteremia HAIs

Method Model Trend Estimate SD R2 p value

Original Model a Linear 3.920E-03 2.160E-03 0.0397 0.0710 Quadratic -3.485E-05 1.618E-05 0.0324

Cubic -7.838E-07 2.850E-07 0.0065

Model b Linear -1.500E-03 8.977E-04 0.0101 0.0953 Quadratic -2.310E-05 1.584E-05 0.1464 Model c Linear -1.270E-03 8.859E-04 0.0049 0.1521 Excluding Model a Linear 2.430E-03 2.180E-03 0.0341 0.2650 Outliers* Quadratic -1.448E-05 1.617E-05 0.3713

Cubic 2.134E-08 2.878E-07 0.9410

Model b Linear 2.580E-03 8.851E-04 0.0388 0.0040 Quadratic -1.476E-05 1.568E-05 0.3476 Model c Linear 2.700E-03 8.749E-04 0.0393 0.0023 Model a: time trend analysis using residual for linear, quadratic, and cubic

Model b: time trend analysis using residual for linear, and quadratic Model c: time trend analysis using residual for linear

*excluding outliers : May-95, Jun-96, Feb-97, Dec-99, Nov-00

Table 5.2. 6 Bacteremia HAI incidence time series after seasonalization and

de-129 trend

Autoregressive Coefficient SD p value

AR(1) 0.0582 0.0693 0.4012

AR(2) 0.0478 0.0688 0.4870

AR(3) 0.1122 0.0687 0.1023

AR(4) -0.0736 0.0688 0.2850

AR(1) 0.0461 0.0687 0.5025

AR(2) 0.0465 0.0687 0.4987

AR(3) 0.1057 0.0685 0.1229

AR(1) 0.0482 0.0690 0.4851

AR(2) 0.0582 0.0689 0.3977

AR(1) 0.0470 0.0688 0.4941

130

Table 5.2. 7 The seasonal effect on pneumonia HAIs incidence (univariate analysis)

Season RR 95% CI p value

Spring /Winter 1.03 0.93 1.13 0.0897

Summer /Winter 1.12 1.02 1.24

Autumn /Winter 1.09 0.98 1.20

131

Table 5.2. 8 Time trend analysis with de-seasonalized residual for pneumonia HAIs

Method Model Trend Estimate SD R2 p value

Original Model a Linear 2.590E-03 2.130E-03 0.0369 0.2240 Quadratic -1.413E-05 1.589E-05 0.3747

Cubic 4.453E-09 2.825E-07 0.9874

Model b Linear 2.620E-03 8.654E-04 0.0414 0.0027 Quadratic -1.420E-05 1.538E-05 0.3570 Model c Linear 2.750E-03 8.538E-04 0.0421 0.0015 Excluding Model a Linear 3.070E-03 2.140E-03 0.0421 0.1526 Outliers* Quadratic -1.377E-05 1.590E-05 0.3875

Cubic -3.984E-08 2.829E-07 0.8882

Model b Linear 2.790E-03 8.683E-04 0.0466 0.0015 Quadratic -1.323E-05 1.539E-05 0.3910 Model c Linear 2.910E-03 8.564E-04 0.0478 0.0008 Model a: time trend analysis using residual for linear, quadratic, and cubic

Model b: time trend analysis using residual for linear, and quadratic Model c: time trend analysis using residual for linear

*excluding outliers: Jul-03, Apr-08, Nov-08

132

Table 5.2. 9 Pneumonia HAI incidence time series after seasonalization and de-trend

Autoregressive Coefficient SD p value

AR(1) 0.1174 0.0691 0.0895

AR(2) 0.1671 0.0693 0.0158

AR(3) 0.1099 0.0692 0.1125

AR(4) -0.0701 0.0694 0.3124

AR(1) 0.1108 0.0688 0.1072

AR(2) 0.1561 0.0684 0.0225

AR(3) 0.1025 0.0688 0.1367

AR(1) 0.1284 0.068 0.0591

AR(2) 0.1695 0.0681 0.0127

AR(1) 0.1546 0.068 0.023

133

Table 5.2. 10 The seasonal effect on SSI HAIs incidence (univariate analysis)

Season RR 95% CI p value

Spring /Winter 1.24 1.10 1.40 0.0029

Summer /Winter 1.20 1.06 1.36

Autumn /Winter 1.10 0.97 1.25

134

Table 5.2. 11 Time trend analysis with de-seasonalized residual for SSI HAIs

Method Model Trend Estimate SD R2 p value

Original Model a Linear -2.920E-03 2.280E-03 0.1214 0.2018 Quadratic -6.460E-06 1.698E-05 0.7042

Cubic -3.195E-07 2.997E-07 0.2877

Model b Linear -5.140E-03 9.299E-04 0.1209 <.0001 Quadratic -1.600E-06 1.636E-05 0.9221 Model c Linear -5.120E-03 9.138E-04 0.1250 <.0001 Excluding Model a Linear -3.400E-03 2.320E-03 0.1312 0.1433

Outliers* Quadratic -4.780E-06 1.715E-05 0.7808

Cubic -2.863E-07 3.033E-07 0.3463

Model b Linear -5.400E-03 9.439E-04 0.1317 <.0001 Quadratic -5.668E-07 1.655E-05 0.9727 Model c Linear -5.390E-03 9.292E-04 0.1359 <.0001 SSI: surgical site infection

Model a: time trend analysis using residual for linear, quadratic, and cubic Model b: time trend analysis using residual for linear, and quadratic Model c: time trend analysis using residual for linear

*excluding outliers:

135

Table 5.2. 12 SSI HAIs incidence time series after de-seasonalization and de-trend Autoregressive Coefficient SD p value

AR(1) 0.125 0.0696 0.0728

AR(2) 0.0286 0.0701 0.6837

AR(3) -0.0352 0.0701 0.6160

AR(4) -0.0778 0.0696 0.2637

AR(1) 0.1287 0.0696 0.0646

AR(2) 0.0263 0.0702 0.7080

AR(3) -0.0452 0.0696 0.5164

AR(1) 0.1275 0.0695 0.0666

AR(2) 0.0205 0.0695 0.7678

AR(1) 0.1302 0.0688 0.0583

136

Table 5.2. 13 The seasonal effect on urinary tract HAIs incidence (univariate analysis)

Season RR 95% CI p value

Spring /Winter 1.11 1.04 1.18 <0.0001

Summer /Winter 1.16 1.09 1.24

Autumn /Winter 1.04 0.97 1.11

137

Table 5.2. 14 Time trend analysis with de-seasonalized residual for UTI HAIs

Method Model Trend Estimate SD R2 p value

Original Model a Linear 6.330E-03 1.850E-03 0.0517 0.0007 Quadratic -2.600E-05 1.384E-05 0.0617

Cubic -9.053E-07 2.438E-07 0.0003

Model b Linear 6.705E-05 7.787E-04 -0.0052 0.9315 Quadratic -1.242E-05 1.374E-05 0.3672 Model c Linear 1.913E-04 7.662E-04 -0.0044 0.8031

Excluding Model a Linear 6.260E-03 1.880E-03 0.0502 0.001

Outliers* Quadratic -2.530E-05 1.403E-05 0.0727

Cubic -9.036E-07 2.473E-07 0.0003

Model b Linear -1.418E-05 7.933E-04 -0.0059 0.9858 Quadratic -1.201E-05 1.394E-05 0.3901 Model c Linear 9.803E-05 7.820E-04 -0.0047 0.9004 Model a: time trend analysis using residual for linear, quadratic, and cubic

Model b: time trend analysis using residual for linear, and quadratic Model c: time trend analysis using residual for linear

*excluding outliers : May-95, Jun-96, Feb-97, Dec-99

138

Table 5.2. 15 UTI HAI incidence time series after de-seasonalization and de-trend Autoregressive Coefficient SD p value

AR(1) 0.0748 0.0686 0.2755

AR(2) 0.1033 0.0683 0.1304

AR(3) 0.0416 0.0683 0.5424

AR(4) -0.1408 0.0682 0.0389

AR(1) 0.0652 0.0688 0.3431

AR(2) 0.0922 0.0686 0.1793

AR(3) 0.0347 0.0687 0.6138

AR(1) 0.066 0.0688 0.3370

AR(2) 0.0903 0.0687 0.1886

AR(1) 0.0735 0.0687 0.2842

139

Table 5.2. 16 The seasonal effect on E.coli HAIs incidence (univariate analysis)

Season RR 95% CI p value

Spring /Winter 1.10 0.85 A1.43 0.7962 Summer /Winter 1.08 0.83 1.39

Autumn /Winter 1.14 0.88 1.47

140

Table 5.2. 17 Time trend analysis with de-seasonalized residual for E.coli HAIs

Method Model Trend Estimate SD R2 p value

Original Model a Linear 1.099E+00 8.453E-01 0.0760 0.9958 Quadratic 1.078E+00 8.335E-01 <.0001

Cubic 1.136E+00 8.796E-01 0.6073 Model b Linear -1.610E-03 1.530E-03 0.0799 0.2943 Quadratic -1.088E-04 2.606E-05 <.0001 Model c Linear 8.710E-05 1.540E-03 -0.0056 0.9550

Excluding Model a Linear -4.585E-05 3.590E-03 0.0746 0.9898 Outliers* Quadratic -1.139E-04 2.835E-05 <.0001

Cubic -2.418E-07 4.936E-07 0.6249

Model b Linear -1.630E-03 1.560E-03 0.0787 0.2973 Quadratic -1.091E-04 2.655E-05 <.0001 Model c Linear 1.339E-05 1.570E-03 -0.0058 0.9932 Model a: time trend analysis using residual for linear, quadratic, and cubic

Model b: time trend analysis using residual for linear, and quadratic Model c: time trend analysis using residual for linear

*excluding outliers: May-95, Feb-97, Dec-99, Apr-00

141

Table 5.2. 18 E. coli bacteremia HAI incidence time series after de-seasonalization and de-trend

Autoregressive Coefficient SD p value

AR(1) -0.0266 0.0755 0.7242

AR(2) -0.1758 0.0753 0.0196

AR(3) 0.0791 0.0755 0.2948

AR(4) -0.0536 0.0723 0.4582

AR(1) -0.0385 0.076 0.6126

AR(2) -0.1713 0.0751 0.0225

AR(3) 0.0376 0.0729 0.6065

AR(1) -0.0431 0.075 0.5660

AR(2) -0.1547 0.072 0.0315

AR(1) -0.0302 0.0727 0.6778

142

Table 5.2. 19 The seasonal effect on Pseudomonas aeruginosa HAIs incidence

Season RR 95% CI p value

Spring /Winter 0.94 0.67 1.31 0.9171

Summer /Winter 1.05 0.76 1.47

Autumn /Winter 1.01 0.73 1.40

143

Table 5.2. 20 Time trend analysis with de-seasonalized Pseudomonas aeruginosa bacteremia HAIs

Method Model Trend Estimate SD R2 p value

Original Model a Linear -1.650E-03 4.210E-03 -0.0033 0.696 Quadratic -2.140E-05 3.358E-05 0.5249

Cubic -1.793E-07 5.796E-07 0.7575

Model b Linear -2.820E-03 1.830E-03 0.0025 0.1255 Quadratic -1.815E-05 3.180E-05 0.5690 Model c Linear -2.540E-03 1.760E-03 0.0069 0.1509

Excluding Model a Linear -1.770E-03 4.310E-03 -0.0021 0.6811 Outliers* Quadratic -1.877E-05 3.410E-05 0.5828

Cubic -1.873E-07 5.911E-07 0.7517

Model b Linear -3.000E-03 1.870E-03 0.0038 0.11 Quadratic -1.553E-05 3.243E-05 0.6328 Model c Linear -2.770E-03 1.800E-03 0.0089 0.1256

Model a: time trend analysis using residual for linear, quadratic, and cubic Model b: time trend analysis using residual for linear, and quadratic Model c: time trend analysis using residual for linear

*excluding outliers: May-95, Sep-96, Dec-99, Jan-00

144

Table 5.2. 21 P. aeruginosa HAI incidence time series after de-seasonalization and de-trend

Autoregressive Coefficient SD p value

AR(1) 0.1270 0.0685 0.0635

AR(2) -0.0036 0.0684 0.9578

AR(3) -0.1402 0.0686 0.0409

AR(4) 0.1186 0.0687 0.0842

AR(1) 0.1192 0.0687 0.0827

AR(2) -0.0050 0.0692 0.9428

AR(3) -0.1201 0.0688 0.0809

AR(1) 0.1202 0.0691 0.0819

AR(2) -0.0162 0.069 0.8144

AR(1) 0.1179 0.0683 0.0846

145

Table 5.2. 22 ARMA model for the HAI infection sites

Outcome Model Effect Estimate S.E. p-value

146

Table 5.2. 23 ARMA model for the bacteremia species

Outcome Model Effect Estimate S.E. p-value

E. coli ARMA(1,1) Mean 4.24E-08 5.38E-05 0.9994

MA(1) -0.5227 0.5101 0.3066

AR(1) -0.4318 0.5393 0.4243

P. aeruginosa ARMA(3,2) Mean 3.81E-06 3.88E-05 0.9219

MA(1) -1.2203 0.0094 <.0001

MA(2) -0.9983 0.0102 <.0001

AR(1) -1.0610 0.0705 <.0001

AR(2) -0.8077 0.0855 <.0001

AR(3) 0.0442 0.0711 0.5347

147

Table 5.3. 1 The adjusted time series model for HAIs incidence (normal distribution)

Parameter Level Estimate S.E. Chisq p-value

Season Spring/Winter 0.0101 0.0168 1.16 0.7619

Summer/Winter 0.0174 0.0168

Autumn/Winter 0.0131 0.0170

Time trend Linear 2.6649E-05 0.0002 0.01 0.9104 Quadratic -5.1747E-06 1.7619E-06 8.62 0.0033 Cubic -1.1630E-07 3.1100E-08 13.93 0.0002 Autocorrelations AR (1) -0.0347 0.0055 39.40 <.0001

Age 0-9 / >=80 -0.9433 0.0257 2639.75 <.0001 10-19 / >=80 -0.9245 0.0257

20-29 / >=80 -0.9346 0.0257 30-39 / >=80 -0.8989 0.0256 40-49 / >=80 -0.7309 0.0255 50-59 / >=80 -0.6117 0.0254 60-69 / >=80 -0.4370 0.0253 70-79 / >=80 -0.2422 0.0252

Gender Male / Female 0.0250 0.0119 4.43 0.0354

148

Table 5.3. 2 The adjusted Poisson model for HAIs incidence

Parameter Level Estimate S.E. Chisq p-value Season Spring/Winter 0.0381 0.0212 11.4 0.0097

Summer/Winter 0.0710 0.0213 Autumn/Winter 0.0477 0.0218

Time trend Linear 7.4737E-04 0.000296757 6.34 0.0118 Quadratic -1.6703E-05 2.3646E-06 50.61 <.0001 Cubic -3.4250E-07 4.05E-08 71.48 <.0001 Autocorrelations AR (1) 8.7558E-04 0.0004 6.05 0.0139 Age 0-9 / >=80 -2.0160 0.0591 6508.2 <.0001

10-19 / >=80 -1.7863 0.0783 20-29 / >=80 -2.1405 0.0531 30-39 / >=80 -1.8330 0.0420 40-49 / >=80 -1.1082 0.0309 50-59 / >=80 -0.8066 0.0262 60-69 / >=80 -0.5360 0.0229 70-79 / >=80 -0.2868 0.0214

Gender Male / Female 0.0301 0.0149 4.1 0.043

149

Table 5.3. 3 The adjusted negative binominal model for HAIs incidence

Parameter Level Estimate S.E. Chisq p-value

Season Spring/Winter 0.0446 0.0319 4.23 0.2379 Summer/Winter 0.0628 0.0318

Autumn/Winter 0.0480 0.0324

Time trend Linear 6.3281E-04 4.4455E-04 2.02 0.1548 Quadratic -1.5758E-05 3.5092E-06 20.01 <.0001 Cubic -3.5340E-07 6.07E-08 33.68 <.0001 Autocorrelations AR (1) 7.5456E-04 0.0005 1.95 0.1629 Age 0-9 / >=80 -2.0345 0.0658 2481.32 <.0001

10-19 / >=80 -1.8159 0.0842 20-29 / >=80 -2.1219 0.061 30-39 / >=80 -1.8335 0.0510 40-49 / >=80 -1.1325 0.0421 50-59 / >=80 -0.8336 0.0387 60-69 / >=80 -0.5421 0.0364 70-79 / >=80 -0.2803 0.0355

Gender Male / Female 0.1063 0.0225 22.68 <.0001

150

Table 5.3. 4 The adjusted time series model for HAIs bacteremia incidence (normal distribution)

Parameter Level Estimate S.E. Chisq p-value

Season Spring/Winter 0.0989 0.0675 4.91 0.1785 Summer/Winter 0.1175 0.067

Autumn/Winter 0.0084 0.068

Time trend Linear -1.30E-03 0.0004 11.18 0.0008

Autocorrelations AR (1) -1.28E-02 0.0203 0.4 0.5263

Age 0-9 / >=80 -1.7992 0.1054 546.83 <.0001 10-19 / >=80 -1.9445 0.1059

20-29 / >=80 -1.8778 0.1067 30-39 / >=80 -1.7759 0.1057 40-49 / >=80 -1.3476 0.1031 50-59 / >=80 -1.0596 0.1017 60-69 / >=80 -0.7646 0.101 70-79 / >=80 -0.4505 0.1007

Gender Male / Female 0.249 0.0474 27.43 <.0001

151

Table 5.3. 5 The adjusted Poisson model for bacteremia HAIs incidence Parameter Level Estimate S.E. Chisq p-value Season Spring/Winter 0.0561 0.0415 3.9 0.2726

Summer/Winter 0.0744 0.0417 Autumn/Winter 0.0226 0.0428

Time trend Linear -1.1876E-03 0.0002 23.22 <.0001 Autocorrelations AR (1) 9.2969E-03 0.0019 22.73 <.0001 Age 0-9 / >=80 -1.5271 0.0994 1346.41 <.0001

10-19 / >=80 -1.7659 0.1619 20-29 / >=80 -2.0728 0.1099 30-39 / >=80 -1.5756 0.0801 40-49 / >=80 -0.8296 0.059 50-59 / >=80 -0.5143 0.0506 60-69 / >=80 -0.3509 0.0461 70-79 / >=80 -0.1959 0.0442

Gender Male / Female 0.1724 0.0292 35.04 <.0001

152

Table 5.3. 6 The adjusted negative binominal model for bacteremia HAIs incidence Parameter Level Estimate S.E. Chisq p-value

Season Spring/Winter 0.0681 0.0499 2.79 0.4255 Summer/Winter 0.0745 0.0499

Autumn/Winter 0.0368 0.0511

Time trend Linear -1.3145E-03 0.0003 19.36 <.0001 Autocorrelations AR (1) 9.3852E-03 0.0024 15.57 <.0001 Age 0-9 / >=80 -1.5427 0.1048 976.92 <.0001

10-19 / >=80 -1.7821 0.1658 20-29 / >=80 -2.0857 0.1153 30-39 / >=80 -1.5854 0.0866 40-49 / >=80 -0.8448 0.0675 50-59 / >=80 -0.5323 0.0603 60-69 / >=80 -0.3626 0.0565 70-79 / >=80 -0.2019 0.055

Gender Male / Female 0.1940 0.0351 30.76 <.0001

153

Table 5.3. 7 The adjusted time series model for HAIs pneumonia incidence (normal distribution)

154

Table 5.3. 8 The adjusted Poisson model for pneumonia HAIs incidence Parameter Level Estimate S.E. Chisq p-value Season Spring/Winter -0.0156 0.0515 10.19 0.0170

Summer/Winter 0.1110 0.0511 Autumn/Winter 0.1022 0.0508

Time trend Linear 1.4807E-03 0.0003 6.34 0.0118 Autocorrelations AR (1) 2.2691E-02 0.0031 53.11 <.0001

AR (2) 0.0003 0.0029 0.01 0.9296

Age 0-9 / >=80 -3.0450 0.2269 1316.81 <.0001 10-19 / >=80 -1.6101 0.1713

20-29 / >=80 -2.3704 0.1395 30-39 / >=80 -1.9751 0.1102 40-49 / >=80 -1.1961 0.0757 50-59 / >=80 -1.0075 0.0634 60-69 / >=80 -0.6025 0.0546 70-79 / >=80 -0.3327 0.0511

Gender Male / Female 0.4891 0.0374 177.03 <.0001

155

Table 5.3. 9 The adjusted negative binominal model for pneumonia HAIs incidence Parameter Level Estimate S.E. Chisq p-value

Season Spring/Winter -0.0441 0.0692 1.04 0.7909 Summer/Winter 0.0101 0.0683

Autumn/Winter 0.0210 0.0679

Time trend Linear 5.8614E-04 0.0004 2.07 0.1499 Autocorrelations AR (1) 1.9548E-02 0.0041 22.21 <.0001

AR (2) 0.0112 0.0042 7.07 0.0078

Age 0-9 / >=80 -3.0691 0.2332 890.61 <.0001 10-19 / >=80 -1.6525 0.1804

20-29 / >=80 -2.4019 0.1562 30-39 / >=80 -2.0246 0.122 40-49 / >=80 -1.2525 0.0924 50-59 / >=80 -0.9384 0.0826 60-69 / >=80 -0.5987 0.0756 70-79 / >=80 -0.2694 0.0727

Gender Male / Female 0.5422 0.0491 121.82 <.0001

156

Table 5.3. 10 The adjusted time series model for SSI HAIs incidence (normal distribution)

157

Table 5.3. 11 The adjusted Poisson model for SSI HAIs incidence

Parameter Level Estimate S.E. Chisq p-value Season Spring/Winter 0.0904 0.0652 2.11 0.5508

Summer/Winter 0.0321 0.0662 Autumn/Winter 0.0332 0.0669

Time trend Linear -3.3693E-03 0.0004 65.21 <.0001 Autocorrelations AR (1) 6.6174E-03 0.0048 1.92 0.1659 Age 0-9 / >=80 -0.5651 0.1625 172.00 <.0001

10-19 / >=80 -0.1660 0.1893 20-29 / >=80 -0.1050 0.1275 30-39 / >=80 0.0023 0.1165 40-49 / >=80 0.5188 0.1031 50-59 / >=80 0.4849 0.1006 60-69 / >=80 0.5461 0.0968 70-79 / >=80 0.4720 0.0975

Gender Male / Female 0.3457 0.046 57.48 <.0001

158

Table 5.3. 12 The adjusted negative binominal model for SSI HAIs incidence Parameter Level Estimate S.E. Chisq p-value Season Spring/Winter 0.1171 0.0947 1.91 0.5913

Summer/Winter 0.0163 0.0945 Autumn/Winter 0.0439 0.0956

Time trend Linear -4.1034E-03 0.0006 46.35 <.0001 Autocorrelations AR (1) 4.0409E-03 0.0072 0.31 0.5752 Age 0-9 / >=80 -0.6135 0.1861 104.71 <.0001

10-19 / >=80 -0.2015 0.2125 20-29 / >=80 -0.0956 0.156 30-39 / >=80 -0.0522 0.1467 40-49 / >=80 0.4787 0.1357 50-59 / >=80 0.4467 0.1336 60-69 / >=80 0.5307 0.1306 70-79 / >=80 0.5239 0.1309

Gender Male / Female 0.3664 0.0656 31.1 <.0001

159

Table 5.3. 13 The adjusted time series model for UTI HAIs incidence (normal distribution)

160

Table 5.3. 14 The adjusted Poisson model for UTI HAIs incidence

Parameter Level Estimate S.E. Chisq p-value Season Spring/Winter 0.0459 0.035 6.73 0.0810

Summer/Winter 0.0800 0.0347 Autumn/Winter 0.0114 0.0353

Time trend Linear 3.0828E-03 0.0005 33.8 <.0001 Quadratic -9.1979E-06 4E-06 5.6 0.0182 Cubic -5.4700E-07 7E-08 53.1 <.0001 Autocorrelations AR (1) 1.5313E-03 0.0012 1.59 0.2068

AR (2) 6.8500E-03 0.0013 27.54 <.0001

161

Table 5.3. 15 The adjusted negative binominal model for UTI HAIs incidence Parameter Level Estimate S.E. Chisq p-value Season Spring/Winter 0.0431 0.0448 4.89 0.1801

Summer/Winter 0.0891 0.0441 Autumn/Winter 0.0154 0.0447

Time trend Linear 2.5617E-03 0.0007 14.4 0.0001 Quadratic -7.4606E-06 5E-06 2.3 0.1321 Cubic -4.7140E-07 9E-08 24.4 <.0001 Autocorrelations AR (1) 2.2398E-03 0.0015 2.13 0.1447

AR (2) 6.5026E-03 0.0017 14.91 0.0001

162

Table 5.3. 16 The adjusted time series model for E. coli bacteremia incidence (normal distribution) Quadratic -2.95E-07 1.822E-07 2.63 0.1052 Autocorrelations AR (1) -0.0037 0.0057 0.42 0.5161