Infection Control & Hospital Epidemiology
311
Table 4. Posterior Summary of the Incidence Rate Ratios of Surveillance Years, Relative to the First Year, for Healthcare-Associated Bloodstream Infections and the Most Common Subtypes for the Cohort and for Hospitals that Participated in BACTOT for <10 Years
Infection Hospitals
Surveillance 2 3 4 5 6 7 8 9 10 HABSI
All Hospitalsa 1.06
(0.93–1.21) 1.02
(0.9–1.17) 1.04
(0.92–1.21) 1.05
(0.92–1.22) 1.05
(0.92–1.22) 1.06
(0.93–1.24) 1.03
(0.91–1.2) 1.05
(0.92–1.23) 1.06
(0.93–1.25)
<10 yb 1.04
(0.89–1.22) 1.02
(0.88–1.21) 1.03
(0.87–1.23) 1.03
(0.87–1.23) 1.06
(0.89–1.32) 1.04
(0.88–1.27) 1.04
(0.88–1.3) 1.04
(0.88–1.28)
Posterior Mean of the Incidence Rate Ratio (95% Posterior Credible Interval) CA-BSI
NCA-BSI
All Hospitalsa 0.98
(0.81–1.19) 0.97
(0.79–1.16) 0.99
(0.82–1.21) 0.97
(0.78–1.16) 0.93
(0.75–1.12) 0.97
(0.78–1.15) 0.92
(0.71–1.09) 0.94
(0.75–1.13) 0.95
(0.76–1.13)
infection secondary to urinary tract infection. aModels fitted to data from the entire cohort. bModels fitted separately to data from hospitals that have participated in BACTOT for <10 years.
starting rates in the tenth year of surveillance. The increase was absent in the subgroup analysis. Although its absence may have been due to reduced power, it may also suggest that the starting 40 hospitals differed from the rest of the cohort in their response to surveillance or that changes independent of surveillance have been occurring. An increase in NCA-BSI rates in these 40 hospitals between years 2014 and 2017 was also highlighted when calendar time trends were investigated elsewhere.24 HABSI rates were higher in the August and September periods compared to April rates. The literature indicates that BSIs caused by gram-negative bacteria, especially Escherichia coli, tend to increase in the summer months.30,31 Although the reasons for such seasonality remain unclear,32 our results suggest that the change in September may be driven in part by CA-BSI, as it also increases. The influence of calendar time, and, in turn, time-dependent confounders, is difficult to eliminate completely when investi- gating trends over surveillance time. This is especially true when a large number of hospitals share the same entry year, particu- larly during the years when only these hospitals remain in the cohort. Adjusting for this by directly including a calendar time variable in the model would create problems due its collinearity with surveillance time. The alternative we opted for in this study was to adjust for the calendar year of entry as a time-invariant effect captured as a random intercept for the hospital effect. To our knowledge, this method has not been used before in the published surveillance literature. Including calendar year of entry in the model reduces the deviance information criterion (DIC), a hierarchical modeling generalization of the Akaike information criterion (AIC), indicating that it captures enough variation in the data to consider it parsimonious to keep it in the model.33
The large case number and patient days covered by our study
was made possible by using a cohort of hospitals with different lengths of BACTOT participation. Without this novel method, a choice between investigation of long-term trends post surveillance and representativeness of post-surveillance trends would have to have been made. Limiting the cohort to hospitals with longer par- ticipation periods would exclude hospitals that may have not begun participation for reasons related to their surveillance capa- bilities or HAI incidence. Results from such a cohort would not be considered representative of hospitals eligible to participate in BACTOT. If a representative cohort was instead chosen, analyses would be limited to the first 3 years of surveillance time, preventing a long-term understanding of HABSI post surveillance. The flex- ibility of Bayesian model writing allowed the fitting of a multilevel model to the available data for each hospital while borrowing strength across hospitals. Notably, any changes or lack thereof reported here cannot be
attributed solely to surveillance. Unavailable HABSI data from hospitals prior to surveillance means an absence of a counterfactual that would allowus to estimate a causal effect. In our study, we used the data from the first year of surveillance as a baseline to which we compared following years. Although the first year of surveillance is not necessarily representative of pre-surveillance rates, we believe that full effects of surveillance require time. After 1 year of partici- pation in BACTOT, hospitals receive an annual report with their rates, compared to the rest of the province, stratified by hospital status.16 First-time reception of this report by a hospital could drive local initiatives to improve HABSI rates for 2 reasons. First, it can be a revelation to key stakeholders and decision makers about the breadth of preventable HABSIs in their hospital, particularly in settings with previously limited surveillance programs or poor
<10 yb 0.95
(0.73–1.15) 0.95
(0.73–1.16) 0.95
(0.72–1.15) 0.99
(0.79–1.22) 0.94
(0.7–1.15) 0.94
(0.69–1.15) 0.96
(0.73–1.16) 0.95
(0.70–1.16)
All Hospitalsa 1.05
(0.86–1.33) 1.07
(0.87–1.37) 1.06
(0.86–1.37) 1.07
(0.87–1.39) 1.06
(0.85–1.36) 1.14
(0.92–1.53) 1.25
(0.99–1.78) 1.17
(0.94–1.59) 1.29
(1.01–1.89)
<10 yb 1.03
(0.86–1.29) 1.03
(0.85–1.28) 1.05
(0.86–1.35) 1.03
(0.86–1.33) 1.05
(0.86–1.35) 1.03
(0.85–1.32) 1.06
(0.87–1.4) 1.02
(0.83–1.33) BSI-UTI
All Hospitalsa 1.03
(0.90–1.19) 0.98
(0.86–1.12) 1.00
(0.86–1.16) 1.02
(0.89–1.2) 1.04
(0.91–1.22) 1.03
(0.89–1.2) 0.99
(0.85–1.15) 0.97
(0.82–1.12) 0.96
(0.81–1.11) Note. HABSI, healthcare-associated bloodstream infection; CS-BSI, catheter-associated bloodstream infection; NCA-BSI, non–catheter-associated bloodstream infection; BSI-UTI, bloodstream
<10 yb 1.00
(0.84–1.18) 0.99
(0.84–1.17) 1.00
(0.84–1.19) 1.00
(0.84–1.19) 1.04
(0.88–1.29) 1.02
(0.86–1.24) 1.01
(0.84–1.21) 1.00
(0.83–1.24)
Page 1 |
Page 2 |
Page 3 |
Page 4 |
Page 5 |
Page 6 |
Page 7 |
Page 8 |
Page 9 |
Page 10 |
Page 11 |
Page 12 |
Page 13 |
Page 14 |
Page 15 |
Page 16 |
Page 17 |
Page 18 |
Page 19 |
Page 20 |
Page 21 |
Page 22 |
Page 23 |
Page 24 |
Page 25 |
Page 26 |
Page 27 |
Page 28 |
Page 29 |
Page 30 |
Page 31 |
Page 32 |
Page 33 |
Page 34 |
Page 35 |
Page 36 |
Page 37 |
Page 38 |
Page 39 |
Page 40 |
Page 41 |
Page 42 |
Page 43 |
Page 44 |
Page 45 |
Page 46 |
Page 47 |
Page 48 |
Page 49 |
Page 50 |
Page 51 |
Page 52 |
Page 53 |
Page 54 |
Page 55 |
Page 56 |
Page 57 |
Page 58 |
Page 59 |
Page 60 |
Page 61 |
Page 62 |
Page 63 |
Page 64 |
Page 65 |
Page 66 |
Page 67 |
Page 68 |
Page 69 |
Page 70 |
Page 71 |
Page 72 |
Page 73 |
Page 74 |
Page 75 |
Page 76 |
Page 77 |
Page 78 |
Page 79 |
Page 80 |
Page 81 |
Page 82 |
Page 83 |
Page 84 |
Page 85 |
Page 86 |
Page 87 |
Page 88 |
Page 89 |
Page 90 |
Page 91 |
Page 92 |
Page 93 |
Page 94 |
Page 95 |
Page 96 |
Page 97 |
Page 98 |
Page 99 |
Page 100 |
Page 101 |
Page 102 |
Page 103 |
Page 104 |
Page 105 |
Page 106 |
Page 107 |
Page 108 |
Page 109 |
Page 110 |
Page 111 |
Page 112 |
Page 113 |
Page 114 |
Page 115 |
Page 116 |
Page 117 |
Page 118 |
Page 119 |
Page 120 |
Page 121 |
Page 122 |
Page 123 |
Page 124 |
Page 125 |
Page 126 |
Page 127 |
Page 128 |
Page 129 |
Page 130 |
Page 131 |
Page 132
