Background

When the World Health Organization on March 11, 2020 officially declared the coronavirus disease 2019 (COVID-19) a global pandemic [1], COVID-19 hospitalisations in the Netherlands were already increasing rapidly. The high influx of patients impacted the critical care capacity, work processes, and availability and use of protective equipment in hospitals [2,3,4,5]. To handle the pressure and high demand of care during this crisis, hospitals reorganised their resources and delivery of care [6]. For example, elective surgeries were postponed or cancelled, intensive care unit (ICU) bed capacity was scaled up, the ratio of healthcare workers allocated to patients was reduced, external staff was hired, and changes to daily care routines, such as the frequency of patient washing, was reduced [7,8,9].

During this pandemic situation, attention to infection prevention and control (IPC) measures may have been deprived given the high work pressure, or redirected towards the prevention of SARS-CoV-2 transmission [10]. In addition, patients hospitalised with COVID-19 are known for having comorbidities, long hospital stays and complex care with multiple invasive devices, putting them at higher risk for healthcare-associated infections (HAIs) [11]. Hence, an increase of HAIs could be expected and is also reported by previous studies [12, 13]. On the other hand, hospitals applied strict, aggressive IPC measures to prevent within-hospital transmission of SARS-CoV-2. As a result, a positive (indirect) effect on HAI occurrence can be expected as well and has been reported by others [14,15,16].

Given these contrasting findings, there is need for adequate HAI reporting not limited to COVID-19 cohorts only, with sufficient historical data to allow pre-pandemic comparisons. The aim of this study was to quantify trends in the number of HAIs in Dutch hospitals during the COVID-19 pandemic, using national surveillance that continued during the pandemic. Second, HAI types were compared between COVID-19 patients versus non-COVID-19 patients.

Methods

Study design and data sources

In this retrospective cohort study, data were derived from the Dutch national nosocomial surveillance network (PREZIES). In short, acute care hospitals voluntarily participate in one or more of the three surveillance modules targeting different HAIs: (1) bi-annually Point Prevalence Surveys (PPS) performed in March and October in which the prevalence of all type of HAIs are measured in all admitted patients (excluding patients admitted to psychiatry and day-care units), (2) Surgical site infection (SSI) incidence surveillance on targeted procedures (see Additional file 1: Table S1 for an overview of the procedures), and (3) hospital-wide catheter-related bloodstream infection (CRBSI) incidence surveillance in patients with a central venous catheter (CVC) in place for ≥ 48 h. For each module, infection control practitioners in each hospital manually review medical records retrospectively according to the national surveillance protocols and annotate which patients meet infection case definitions. The surveillance protocols and case definitions are based on the (European) Centres for Disease Control and Prevention and are described elsewhere [17,18,19]. Only in the PPS and CRBSI modules information was collected about whether the patient was admitted to the hospital due to COVID-19 (positive test at admission). Hospitals that reported their surveillance data yearly to PREZIES over the years 2016–2020 were included in this study and used to evaluate the infection rates during the pre-pandemic and pandemic period.

Definition pre-pandemic and pandemic period

Based on COVID-19 hospitalisation rates in the Netherlands, the PPS surveys of 2016–2019 were defined as pre-pandemic and the surveys of March and October 2020 were defined as the pandemic period. Data from the SSI and CRBSI modules were divided in pre-pandemic (January 2016 to February 2020) and pandemic (from 1st of March 2020 to December 2020).

Statistical analyses

Per module, patient-, surgery-, or CVC- related characteristics were reported and compared between the pre-pandemic and pandemic period, using a chi-square test for categorical variables and Mann–Whitney U test for continuous variables. Thereafter, we quantified the number of HAIs during the pandemic. For PPS data, the difference in observed HAI rates between the pre-pandemic and pandemic period was tested using chi-square 2-tailed test with Yates’ correction.

For the SSI and CRBSI incidence, we estimated the expected infection rates for the pandemic period based on pre-pandemic data and compared this with the actual observed rates in the pandemic period. To estimate the expected incidence rate for SSI, the National Nosocomial Infections Surveillance System (NNIS) risk index in pre-pandemic data was used to predict the risk of SSI for each NNIS category for the pandemic period (Additional file 1: Figure S1). The NNIS risk index, ranging from 0 to 3, is composed of 1 point for each of the following criteria: wound class classified as contaminated or infected; American Society of Anaesthesiologists (ASA) score of 3, 4, or 5; and an operation duration above the 75th percentile [20]. The predicted infection rate was compared with the observed infection rate using a chi-square test. In addition, two sensitivity analyses for SSI were performed. First, the same analyses were repeated for deep SSI only, with the rationale that superficial SSIs may have been missed during follow-up in the pandemic period: patients avoided contact with healthcare providers afraid of becoming infected with SARS-CoV-2, patients did not want to be a burden on the system, and follow-up appointments were replaced by remote care because of stay-at-home orders [9]. Second, trends in SSI incidence rates were checked per surgical specialty. In case an increasing or decreasing trend was observed pre-pandemic, the expected SSI rate was recalculated based on 2019 data only. To estimate the expected CRBSI incidence per 1000 CVC days in the pandemic period, the mean pre-pandemic incidence per 1000 CVC days for each of the three application-based categories (total parenteral nutrition (TPN); dialysis; and the remaining other applications) was multiplied with the pandemic number of CVCs in each category (Additional file 1: Figure S1). The predicted and observed incidence rates were compared using a mid-P exact test. Last, differences in patient characteristics, medical device use, and HAIs were investigated in COVID-19 patients versus non-COVID-19 patients based on PPS and CRBSI data, by using a chi-square test or Mann–Whitney test. A p value of < 0.05 was considered statistically significant and analyses were performed using SAS version 9.4 software (SAS Institute, Cary, NC).

Results

Table 1 shows the number of hospitals participating in the three different modules, per year. The number of hospitals reporting PPS data during the pandemic year 2020 was less than half compared with previous years. Subsequent analyses were performed for the PPS, SSI, and CRBSI module, using data from 10, 51, and 11 hospitals respectively that reported their yearly surveillance data in 2016–2020 to PREZIES (Table 1). In these hospitals, the absolute annual number of admissions (PPS module) and surgeries (SSI module) was lower in 2020 compared to previous years, while there was a slight increase in the number of inserted CVCs (CRBSI module).

Table 1 Overview of hospitals included in this study
Full size table

Healthcare-associated infections during the first pandemic year

Point prevalence survey results

During the pandemic period, a higher proportion of hospitalised patients was male, patients had slightly higher McCabe scores and more ICU admissions were observed (Table 2). The proportion of patients having a medical device increased during the pandemic period, in particular the use of CVCs. The proportion of patients with antibiotic treatment at the time of the survey was slightly higher during the pandemic (42.6%) versus pre-pandemic (37.7%; p < 0.01). The total HAI prevalence was higher during the pandemic period compared to pre-pandemic period, mainly because of an increase in gastro-intestinal infections and infections of the central nervous system (Tables 3, 4). The proportion of patients with lower respiratory tract infections (LRTIs) in the pandemic period was similar compared to pre-pandemic, however, a larger proportion was associated with mechanical ventilation (ventilator-associated pneumonia (VAP), 22.5% pandemic versus 13.5% pre-pandemic, Table 4).

Table 2 Patient-, surgery-, and central venous catheter characteristics
Full size table
Table 3 Infection rates pre-pandemic, predicted infection rates during pandemic, and observed infection rates during the pandemic
Full size table
Table 4 Distribution of healthcare-associated infections in pre-pandemic and pandemic PPS cohort
Full size table

Surgical site infections

Within the SSI module, 217,212 surgeries were included in the pre-pandemic period versus 35,793 surgeries during the pandemic. Compared to the pre-pandemic period, patients operated during the pandemic period were more often of the male gender, had slightly higher ASA- and NNIS scores and had shorter hospital stays (Table 2). The observed SSI incidence for all type of surgeries combined in the pandemic period was significantly lower than predicted (1.8% versus 2.1%, respectively) (Fig. 1 and Table 3). When stratified by surgery type, only the SSI incidence after colon surgery was significantly lower during the pandemic (p < 0.01; Table 3). During 2016–2019, already a decreasing trend in SSI incidence after colorectal surgeries was observed (7.2%; 7.2%; 6.3%; 5.0%, respectively), while the proportion of closed procedures increased (p < 0.01, Additional file 1: Figure S2). When calculating the expected SSI incidence after colorectal surgery based on 2019 data only, the SSI rate in de pandemic was as predicted (predicted SSI rate: 5.1%; 95%-CI 4.5–5.8, observed SSI rate: 4.4%; 95%-CI: 3.9–5.0, p = 0.1). Sensitivity analysis comparing observed and expected incidence of deep SSI only showed similar results (Additional file 1: Table S2).

Fig. 1
figure 1

Infection rates pre-pandemic, and predicted and observed infection rates during the pandemic period. HAIs healthcare-associated infections, PPS point prevalence survey, SSIs surgical site infections, RTIs respiratory tract infections, BSIs bloodstream infections, UTIs urinary tract infections, CRBSIs catheter-related bloodstream infections

Full size image

Catheter-related bloodstream infections

During the pandemic period, patients with a CVC were slightly younger, more often of the male gender and more often admitted to the ICU compared to the pre-pandemic period. During the pandemic period, the number of inserted CVCs per patient was slightly higher and the CVC duration was longer. CVCs were more frequently used for antibiotics and hemodynamic monitoring and less often for TPN (Table 2). The observed CRBSI incidence of 4.0/1000 CVC days (95%-CI 3.2–4.9/1000) in the pandemic period was significantly higher than the predicted rate of 1.4/1000 CVC days (95%-CI 1.0–2.1/1000; p < 0.01) (Fig. 1).

Healthcare-associated infections within COVID-19 patients

Within the PPS module, COVID-19 status was only registered during the survey in October 2020: 50 (6.6%) patients were SARS-CoV-2 positive during admission and were compared with 713 (93.4%) non-COVID-19 patients. COVID-19 patients were more often admitted to the ICU, had more often medical devices and antibiotic use (Table 5). A significantly higher HAI prevalence was observed in this patient group as compared to non-COVID-19 patients (12% versus 0.4% respectively, p < 0.01), with bloodstream infections (BSI) as the most predominant manifestation (Additional file 1: Table S3).

Table 5 Differences in COVID-19 patients versus non-COVID patients admitted to the hospital, March 2020–December 2020
Full size table

A total of 9 out of 11 hospitals participating in the CRBSI module reported whether the patient was admitted to the hospital due to COVID-19. These COVID-19 patients were more often male, were slightly younger in age, and had significant longer ICU length of stay compared to non-COVID-19 patients with a CVC during the pandemic period. In addition, COVID-19 patients had more CVCs inserted and with a longer duration (Table 5). The CVC was more often used for antibiotics and less for TPN compared to non-COVID-19 patients. The CRBSI incidence was 8.1/1000 CVC days (95%-CI 5.9–10.8) in COVID-19 patients compared to 3.4/1000 (95%-CI 2.2–5.0) in patients without COVID-19 (p < 0.01). When stratifying the COVID-19 patients to ICU and non-ICU, CRBSI rates were 7.8/1000 CVC days (95%-CI 5.6–10.7) and 11.1 (95%-CI 5.0–24.7) respectively. When stratifying the non-COVID-19 patients to ICU and non-ICU, CRBSI rates were 4.8/1000 CVC days (95%-CI 3.0–7.6) and 1.7 (95%-CI 0.7–4.0) respectively. The CRBSI incidence for non-COVID-19 patients in the ICU (4.8/1000) was significantly higher compared to pre-pandemic years (0.7/1000; 95%-CI 0.5–1.1) as well.

Discussion

During the first pandemic year CRBSIs, VAPs, gastro-intestinal- and central nervous system infections occurred more frequently among hospitalised patients, while SSIs and catheter-associated urinary tract infection (CAUTI) rates remained stable. HAIs occurred more often in COVID-19 patients, however, in non-COVID-19 patients admitted to the ICU an sevenfold increase of CRBSI was observed during the pandemic compared to pre-pandemic as well.

Regarding SSI, less surgeries were performed in 2020 and the patients that have been operated had slightly higher ASA and NNIS scores compared to previous years, possibly explained by prioritising urgent procedures during the pandemic period. Although this patient population may be more likely to develop SSIs, no increase in incidence was observed. Remarkable is the relative high number of laparoscopic colon surgeries during the pandemic, which may be induced by policies to relieve ICU capacity and the shift to minimally invasive surgery to protect operating room personnel from SARS-CoV-2 aerosol transmission [21]. Future data will show whether open surgery had been replaced during the pandemic by closed surgery, or whether the open surgeries were postponed.

The findings of this study are in line with previous research: several studies reported increases during the pandemic in among others CRBSIs, BSIs, and VAPs [12, 13, 22,23,24,25]. The PPS data showed that the prevalence of LRTIs did not change, however the proportion of LRTIs associated with ventilation increased, likely due to the increased use of mechanical ventilation [26]. Importantly, the work pressure, burden and influx of COVID-19 patients was not constant throughout 2020: COVID-19 surges varied during the year, by region and by hospital [27]. Especially for the PPS, the timing of the surveys (March and October) may not have paralleled the COVID-19 surges and circumstances and therefore may have underestimated potential effects: we did not find any increase in CRBSIs or CAUTIs in the PPS data while this was reported by others [23, 24]. Within the CRBSI module, the number of CRBSI events was too low to perform sub-analyses to evaluate stronger effects on incidence rates during COVID-19 surges.

Most studies published so far are of variable quality as they are limited to retrospective cohort studies. Moreover, they focus solely on COVID-19 patients, and lack standardized case definitions without differentiating between settings or specialties [28]. The current surveillance-based study has a retrospective design as well and hospitals performed the surveillance themselves, however by using standardized case definitions and large sample sizes from a fixed number of hospitals for several years, the results of our study may be more robust. Still, with our study design, we cannot fully explain (causal) reasons for the change in HAIs observed during the pandemic. Several hypotheses are possible, probably all contributing to some degree. In part, the increase in HAIs can be explained by the fact that hospitalisations were dominated by COVID-19 patients who may have been more vulnerable for HAIs and other co-infections due to immune dysregulation [29,30,31,32]. This is also reflected by the high antibiotic use observed in these patients, which will increase risk of antibiotic resistance. In Germany, there was no ICU overcrowding due to COVID-19 patients because of their high ICU bed capacity as compared with the Netherlands, and no increase in device-associated infections was observed in this country [33]. In addition, COVID-19 patients in general are more exposed to known risk factors for HAIs such as longer durations of mechanical ventilation, higher number of CVCs inserted, corticosteroid treatment, prone positioning, and longer lengths of stay [24]. Although not observed within this study, the composition of characteristics of remaining non-COVID-19 hospitalised patients is likely to be different than pre-pandemic, due to numerous elective procedures that were cancelled and postponed. Unfortunately, within the surveillance modules we only have limited patient- and clinical information, restricting the adjustment for casemix. Although we used data of a fixed set of hospitals and used the NNIS score and CVC applications to calculate the expected infection rates, we may not have completely addressed the shift in characteristics of the patient population during the pandemic. The increased CRBSI incidence in non-COVID-19 ICU patients may indicate that both a change in patient mix or the reorganization of care, such as changed IPC practices, modified use of personal protective equipment, and additional (unskilled ICU) temporary staff, may have contributed to the increased infection risk [5, 16, 34, 35]. To fully explain HAI dynamics in pandemic circumstances indicators describing the local healthcare context at institutional level are needed, such as patient characteristics, disruption of IPC practices, prescribing- and (microbiological) order practices, and antimicrobial resistance patterns [36].

Conclusions

Summarized, we observed an increase in rates of CRBSI, VAP, gastro-intestinal- and central nervous system infections among hospitalised patients during the first pandemic year. Furthermore, CRBSI incidence was also increased in the non-COVID-19 ICU population during the pandemic. The full scope and driving factors of this change in HAIs need to be studied in more detail to be able to anticipate—from an infection prevention perspective—more adequately on future epidemics of COVID-19 or other severe acute respiratory infections.