Abstract：
BACKGROUND: Detection and containment of hospital outbreaks currently depend on variable and personnel-intensive surveillance methods. Whether automated statistical surveillance for outbreaks of health care-associated pathogens allows earlier containment efforts that would reduce the size of outbreaks is unknown.
METHODS: We conducted a cluster-randomized trial in 82 community hospitals within a larger health care system. All hospitals followed an outbreak response protocol when outbreaks were detected by their infection prevention programs. Half of the hospitals additionally used statistical surveillance of microbiology data, which alerted infection prevention programs to outbreaks. Statistical surveillance was also applied to microbiology data from control hospitals without alerting their infection prevention programs. The primary outcome was the number of additional cases occurring after outbreak detection. Analyses assessed differences between the intervention period (July 2019 to January 2022) versus baseline period (February 2017 to January 2019) between randomized groups. A post hoc analysis separately assessed pre-coronavirus disease 2019 (Covid-19) and Covid-19 pandemic intervention periods.
RESULTS: Real-time alerts did not significantly reduce the number of additional outbreak cases (intervention period versus baseline: statistical surveillance relative rate [RR]=1.41, control RR=1.81; difference-in-differences, 0.78; 95% confidence interval [CI], 0.40 to 1.52; P=0.46). Comparing only the prepandemic intervention with baseline periods, the statistical outbreak surveillance group was associated with a 64.1% reduction in additional cases (statistical surveillance RR=0.78, control RR=2.19; difference-in-differences, 0.36; 95% CI, 0.13 to 0.99). There was no similarly observed association between the pandemic versus baseline periods (statistical surveillance RR=1.56, control RR=1.66; difference-in-differences, 0.94; 95% CI, 0.46 to 1.92).
CONCLUSIONS: Automated detection of hospital outbreaks using statistical surveillance did not reduce overall outbreak size in the context of an ongoing pandemic. (Funded by the Centers for Disease Control and Prevention; ClinicalTrials.gov number, NCT04053075. Support for HCA Healthcare\'s participation in the study was provided in kind by HCA.).
METHODS: We conducted a cluster-randomized trial in 82 community hospitals within a larger health care system. All hospitals followed an outbreak response protocol when outbreaks were detected by their infection prevention programs. Half of the hospitals additionally used statistical surveillance of microbiology data, which alerted infection prevention programs to outbreaks. Statistical surveillance was also applied to microbiology data from control hospitals without alerting their infection prevention programs. The primary outcome was the number of additional cases occurring after outbreak detection. Analyses assessed differences between the intervention period (July 2019 to January 2022) versus baseline period (February 2017 to January 2019) between randomized groups. A post hoc analysis separately assessed pre-coronavirus disease 2019 (Covid-19) and Covid-19 pandemic intervention periods.
RESULTS: Real-time alerts did not significantly reduce the number of additional outbreak cases (intervention period versus baseline: statistical surveillance relative rate [RR]=1.41, control RR=1.81; difference-in-differences, 0.78; 95% confidence interval [CI], 0.40 to 1.52; P=0.46). Comparing only the prepandemic intervention with baseline periods, the statistical outbreak surveillance group was associated with a 64.1% reduction in additional cases (statistical surveillance RR=0.78, control RR=2.19; difference-in-differences, 0.36; 95% CI, 0.13 to 0.99). There was no similarly observed association between the pandemic versus baseline periods (statistical surveillance RR=1.56, control RR=1.66; difference-in-differences, 0.94; 95% CI, 0.46 to 1.92).
CONCLUSIONS: Automated detection of hospital outbreaks using statistical surveillance did not reduce overall outbreak size in the context of an ongoing pandemic. (Funded by the Centers for Disease Control and Prevention; ClinicalTrials.gov number, NCT04053075. Support for HCA Healthcare\'s participation in the study was provided in kind by HCA.).
摘要：
背景：目前，对医院暴发的检测和控制取决于可变和人员密集的监测方法。对卫生保健相关病原体暴发的自动统计监测是否允许更早的遏制努力来减少暴发的规模尚不清楚。
方法：我们在一个更大的医疗保健系统内的82家社区医院进行了一项整群随机试验。当感染预防计划发现爆发时，所有医院都遵循爆发响应协议。一半的医院还使用了微生物数据的统计监测，这提醒了感染预防计划的爆发。统计监测也应用于控制医院的微生物学数据，而没有提醒他们的感染预防计划。主要结果是爆发后发生的额外病例数。分析评估了随机组之间干预期（2019年7月至2022年1月）与基线期（2017年2月至2019年1月）之间的差异。事后分析分别评估了2019年冠状病毒病前期(新冠肺炎)和新冠肺炎大流行干预期。
结果：实时警报并未显着减少额外爆发病例的数量（干预期与基线相比：统计监测相对率[RR]=1.41，对照RR=1.81；差异差异，0.78；95%置信区间[CI]，0.40至1.52；P=0.46）。仅将疾病前期干预与基线期进行比较，统计暴发监测组的额外病例减少了64.1%(统计监测RR=0.78,对照RR=2.19;差异差异,0.36；95%CI，0.13至0.99）。大流行期与基线期之间没有类似的相关性(统计监测RR=1.56，对照RR=1.66；差异差异，0.94；95%CI，0.46至1.92）。
结论：在大流行持续的背景下，使用统计监测自动检测医院暴发并没有减少总体暴发规模。（由疾病控制和预防中心资助；ClinicalTrials.gov编号，NCT04053075。HCA为HCAHealthcare参与研究提供了实物支持。).
方法：我们在一个更大的医疗保健系统内的82家社区医院进行了一项整群随机试验。当感染预防计划发现爆发时，所有医院都遵循爆发响应协议。一半的医院还使用了微生物数据的统计监测，这提醒了感染预防计划的爆发。统计监测也应用于控制医院的微生物学数据，而没有提醒他们的感染预防计划。主要结果是爆发后发生的额外病例数。分析评估了随机组之间干预期（2019年7月至2022年1月）与基线期（2017年2月至2019年1月）之间的差异。事后分析分别评估了2019年冠状病毒病前期(新冠肺炎)和新冠肺炎大流行干预期。
结果：实时警报并未显着减少额外爆发病例的数量（干预期与基线相比：统计监测相对率[RR]=1.41，对照RR=1.81；差异差异，0.78；95%置信区间[CI]，0.40至1.52；P=0.46）。仅将疾病前期干预与基线期进行比较，统计暴发监测组的额外病例减少了64.1%(统计监测RR=0.78,对照RR=2.19;差异差异,0.36；95%CI，0.13至0.99）。大流行期与基线期之间没有类似的相关性(统计监测RR=1.56，对照RR=1.66；差异差异，0.94；95%CI，0.46至1.92）。
结论：在大流行持续的背景下，使用统计监测自动检测医院暴发并没有减少总体暴发规模。（由疾病控制和预防中心资助；ClinicalTrials.gov编号，NCT04053075。HCA为HCAHealthcare参与研究提供了实物支持。).
