Introduction
Canadian acute care hospitals continue to be burdened by healthcare-associated infections (HAIs), including those caused by antimicrobial-resistant organisms (AROs), resulting in increased morbidity, mortality, and excess healthcare costs. 1 The coronavirus disease 2019 (COVID-19) pandemic caused by the SARS-CoV-2 virus caused major operational changes in hospitals, disrupting infection prevention and control (IPC) and antimicrobial stewardship programs. Reference Micheli, Sangiorgi and Catania2 Conversely, reduced population mobility and international travel along with greater IPC awareness may have decreased HAI transmission. Reference Micheli, Sangiorgi and Catania2,Reference Lastinger, Alvarez and Kofman3 International studies reported varying changes in HAI and ARO rates during the pandemic. Reference Lastinger, Alvarez and Kofman3
In Canada, hospital-based surveillance on adult and pediatric HAIs, including AROs, are conducted through the Canadian Nosocomial Infection Surveillance Program (CNISP), a collaboration between the Public Health Agency of Canada, including the National Microbiology Laboratory, the Association of Medical Microbiology and Infectious Disease Canada, and participating sentinel hospitals.
This study assessed the immediate and long-term association between the COVID-19 pandemic and the incidence of healthcare-associated (HA) rates of Clostridioides difficile infection (CDI), methicillin-resistant Staphylococcus aureus (MRSA) bloodstream infection (BSI), vancomycin-resistant Enterococcus (VRE) BSI, carbapenemase-producing Enterobacterales (CPE) infections among hospital inpatients, and central line-associated bloodstream infection (CLABSI) in the adult intensive care unit (ICU) in Canadian acute care hospitals participating in CNISP between 2018 and 2022.
Methods
Data sources
The study surveillance period spanned 60 months (January 1, 2018–December 31, 2022). We selected five priority HAIs of interest with significant hospital burden: HA-CDI, HA-MRSA BSI, HA-VRE BSI, and HA-CPE infections among all hospital inpatients, and CLABSIs in adult ICUs with a mix of patient types (i.e., medical/surgical). Hospitals may select surveillance participation based on resource capacity and relevance, and thus varied by HAI. For each HAI, we restricted our study sample to hospitals that submitted comprehensive data for all five surveillance years.
Data collection, HAI case definitions, and laboratory methods were previously described. 4 Monthly patient days were not available and were estimated by dividing the quarterly totals by the number of days in the quarter, then multiplying by the number of days in the month—assuming consistent patient days throughout each quarter and month. Monthly infection rates were calculated per 10,000 patient days by dividing the number of monthly incident infections by monthly patient days, then multiplying by 10,000. For CLABSIs, we calculated monthly infection rates per 1,000 central line days.
Statistical analysis
We used a generalized linear model with quasi-Poisson distribution, offset by the log number of patient days to perform interrupted time-series modeling to assess changes in monthly infection rates between the pre-COVID-19 pandemic period (January 1, 2018–February 29, 2020; 26 time points) and the COVID-19 pandemic period (March 1, 2020–December 31, 2022; 34 time points). We selected these periods based on the most available and complete hospital infection data to reasonably assess changes in HAI trends.
We adjusted models for seasonality and select hospital-level covariates: geographic region, bed size category, teaching status, and hospital type. We assessed both immediate (step) and long-term (slope) changes using step-and-slope modeling for each eligible HAI. We reported results as incidence rate ratios (IRRs) with 95% confidence intervals (CIs) adjusted for hospital-level clustering. For each infection, a forward addition stepwise procedure was used to select the most parsimonious model. We assessed autocorrelation using plots of the autocorrelation and partial autocorrelation functions. We conducted all analyses using R 4.1.1 at an α ≤ 0.05.
Results
We included a total of 11,267 HAIs in this study from 26–56 Canadian acute care hospitals from 2018 to 2022 (Table 1), with 43% (n = 4,862) and 57% (n = 6,405) reported during the pre-pandemic and pandemic periods, respectively. Most infections were HA-CDI (n = 7,949, 70.5%), followed by HA-MRSA BSI (n = 1,277, 11.3%), HA-VRE BSI (n = 1,062, 9.4%), adult mixed ICU CLABSIs (n = 814, 7.2%), and HA-CPE infections (n = 165, 1.5%). The largest number of participating hospitals were from Central Canada (37–50%), medium-sized (42–54%), and teaching hospitals (82–88%). Supplemental information regarding denominators, patient characteristics, and outcomes are available in Tables S2 and S3.
Characteristics of hospitals participating in the surveillance of healthcare-associated infections, Canadian Nosocomial Infection Surveillance Program, 2018–2022
* Mixed hospitals include both adult and pediatric patient populations.
** Pediatric hospitals do not have adult mixed ICU wards.
Abbreviations: BSI, bloodstream infection; CPE, carbapenemase-producing Enterobacterales; CLABSI, central line-associated bloodstream infections; CDI, Clostridioides difficile infection; HA, healthcare-associated; ICU, intensive care unit; MRSA, methicillin-resistant Staphylococcus aureus; VRE, vancomycin-resistant Enterococcus.
Among the HAIs collected across all hospital inpatients, the overall crude incidence rates per 10,000 patient days from 2018 to 2022 were highest for HA-CDI (3.59), followed by HA-MRSA BSI (0.45), HA-VRE BSI (0.29), and HA-CPE (0.06). For adult mixed ICU CLABSIs, the overall crude incidence rate was 1.40 per 1,000 line-days.
Compared to the pre-pandemic period, crude pandemic period rates per 10,000 patient days increased overall for HA-VRE BSI (0.26 to 0.30, p = 0.037) and HA-CPE infections (0.04 to 0.07, p = 0.002), remained stable for HA-CDI (3.56 to 3.61, p = 0.536), and decreased for HA-MRSA BSI from 0.50 to 0.41 infections per 10,000 patient days (p = 0.0009). Additionally, adult mixed ICU CLABSI rates increased from 1.25 to 1.51 infections per 1,000 line-days (p = 0.009) during the pandemic period.
Multivariable step-and-slope modeling results (Figure 1, Table S4) indicated that after adjusting for seasonality, clustering, and hospital covariates, the COVID-19 pandemic was significantly associated with a 21% step-change decrease in MRSA BSI rates (aIRR: 0.790 (0.639–0.977), p = 0.030) while a significant slope change increase was observed for HA-CDI (aIRR: 1.011 (1.004–1.017), p = 0.0007) compared to the pre-pandemic period. During the COVID-19 pandemic, modeled monthly HA-CDI rates remained elevated and stable compared to the declining pre-pandemic trend. We found no significant step or slope changes in monthly rate trends for all other HAIs.
Modeled rates of healthcare-associated infection using interrupted time-series analysis before and during the COVID-19 pandemic, Canadian Nosocomial Infection Surveillance Program, 2018–2022.
Discussion
We modeled the association between the COVID-19 pandemic and rates of HAIs including those caused by AROs in a large, national network of Canadian acute care hospitals. We found that the pandemic was associated with a significant immediate step decrease in HA-MRSA BSI rates and a long-term slope increase in HA-CDI rate trends compared to the pre-pandemic period, while all other HAI rate trends did not significantly change.
Our previous surveillance data showed that secular, pre-pandemic rate trends were generally consistent with those observed during the pandemic for all HAIs except HA-CDI. 4 The pandemic increase in HA-CDI rate trends compared to previously declining rate trends can likely be attributed to competing pandemic-related factors including changes to inpatient acuity, routine care, hospital resources, hospital staffing, and IPC practices. Reference Gottlieb and Fridkin5 Monthly pandemic increases in HA-CDI rate trends may be attributed to surges in elderly and vulnerable patient admissions, particularly from Canadian long-term care homes which experienced significant COVID-19 outbreaks. Reference Choi, Du and Silva6 Additionally, the empiric use of broad-spectrum antibiotics at the start of the pandemic may have disrupted the fecal microbiota in this disproportionately affected elderly population and increased HA-CDI risk. Reference Lewandowski, Rosołowski and Kaniewska7 Pandemic-era HA-CDI rates varied by country, with increases noted in select single-center studies in Europe Reference Lewandowski, Rosołowski and Kaniewska7,Reference Karampatakis, Tsergouli, Kandilioti, Nikopoulou, Katsifa and Kachrimanidou8 while decreases were observed in high-income jurisdictions including the US. Reference Lastinger, Alvarez and Kofman3
Although trends in HA-MRSA BSI rates did not significantly change, step-and-slope modeling resulted in an immediate decrease at pandemic onset. Similar immediate MRSA BSIs decreases were reported in sub-populations in China and South Korea Reference Li, Cao, Ding, Yang and Duan9,Reference Lee, Kim, Kim, Park and Lee10 while other studies in France and Brazil observed immediate increases. Reference Micheli, Sangiorgi and Catania2 The US reported a 12% increase in lab-identified MRSA bacteremia standardized infection ratios at pandemic onset, though attributed to the decline in 2020 Q2 patient days. Increased antimicrobial use including empiric vancomycin at pandemic onset and enhanced IPC practices including hand hygiene, personal protective equipment usage, and environmental cleaning may have offset pandemic-related HAI risk factors and contributed to reductions observed in a Japanese hospital network. Reference Micheli, Sangiorgi and Catania2,Reference Ponce-Alonso, Sáez De La Fuente and Rincón-Carlavilla11 System-wide changes in bed assignment policies, visitor restrictions, increased patient and staff vigilance, outpatient care diversion, reduced elective care, and patient-directed healthcare avoidance may have further contributed to these reductions. 12
Although no significant immediate or long-term changes were observed for HA-VRE BSI, HA-CPE infections, and adult mixed ICU CLABSIs, pre-pandemic increasing trends persisted through the pandemic. Many risk factors such as increased patient acuity, staffing challenges, and device use may have already contributed to rising trends prior to the pandemic. 13 These risks partially offset by enhanced IPC practices may have resulted in sustained, rather than amplified, increases in these HAIs during the pandemic.
This large, multicenter study included five years of standardized hospital surveillance data for five priority HAIs from all ten Canadian provinces. However, there are limitations to this study. While we captured the first six major COVID-19 pandemic waves, we did not assess the wave-based variation in public health measures, screening practices, antibiotic usage, or dominant SARS-CoV-2 virus strain in our model. Modeled associations were assessed nationally and may differ by hospital due to hospital-specific IPC measures, hospital or community outbreaks, and regional public health measures before and during the pandemic. Future studies should aim to incorporate site-specific COVID-19 hospitalization and ICU admission data to assess the impact of various waves on the burden of HAIs in Canadian acute care settings. Estimating monthly patient and central line days may have not fully captured the monthly variation in patient volume, potentially biasing modeled point estimates and estimates of variance. However, hospital administrative data used as a proxy showed monthly inpatient occupancy variation from March 2020 to June 2021 was under 20%. 12
Conclusion
While the COVID-19 pandemic placed a significant burden on the Canadian acute care hospitals, HA-MRSA BSI significantly decreased at its onset, while only HA-CDI rate trends increased compared to the pre-pandemic period. Understanding these pandemic epidemiological effects in the context of changing patient populations and clinical and IPC practices is essential for continued awareness, preparedness, and management of HAIs including those caused by AROs in Canadian acute care settings.
Supplementary material
To view supplementary material for this article, please visit https://doi.org/10.1017/ice.2025.10247
Acknowledgements
We gratefully acknowledge the contribution of the physicians, epidemiologists, infection control practitioners, and laboratory staff at each participating hospital and those participating in the following CNISP working groups: viral respiratory infections, C. difficile, central line-associated bloodstream infections, carbapenemase-producing organisms, vancomycin-resistant Enterococcus, and methicillin-resistant S. aureus. Thank you to the staff at Public Health Agency of Canada in the Centre for Communicable Diseases and Infection Control (Diane Lee, Cassandra Lybeck, Cecilia McClellan, Erin McGill, Andrew Neitzel, Linda Pelude, Zainab Suleman, and Olivia Varsaneux).
Author contributions
AS, JJB, JC, KBC, and RM significantly contributed to the conceptualization, methodology, analysis, interpretation, preparation, and review of the manuscript. JLC, CF, SSH, JJ, KCK, SWS, JAS, KNS, and NT significantly contributed to the methodology, interpretation, and review of the manuscript.
Financial support
This work was supported by the Public Health Agency of Canada.
Competing interests
KNS holds a leadership position as Physician Director on the Infection Prevention and Control Canada Board of Directors. JJB has received grants from the WHO and support via a fellowship program for attending in-person meeting for the Pandemic EVIDENCE Collaboration hosted by Kellogg College, Oxford University. SSH has received an honorarium for a personal consultation meeting from Ferring Pharmaceuticals and holds a leadership position as President-Elect (previously Director) for the Association of Medical Microbiology and Infectious Diseases Canada. The authors [KNS, JJB, SSH] and the funding provided by the Public Health Agency of Canada have not had influence on the findings of this manuscript. No other authors have any declared conflicts of interest.
Patient consent statement
The design of the work conforms to the Public Health Agency of Canada (PHAC) Research Ethics Board (REB) standard.
Due to the routine nature of the surveillance activities conducted by the Canadian Nosocomial Infection Surveillance Program (CNISP) and the collection of de-identified patient information as per the signed PHAC data sharing agreements by each participating hospital, the PHAC REB has deemed the activities and analysis conducted by CNISP as monitoring and quality improvement and therefore does not require the express written consent of the patient.
Open access
