Burden of hospital-acquired SARS-CoV-2 infections in Germany: occurrence and outcomes of different variants.

BACKGROUND: Avoiding in-hospital transmissions has been crucial in the COVID-19 pandemic. Little is known on the extent to which hospital-acquired SARS-CoV-2 variants have caused infections in Germany. AIM: To analyse the occurrence and the outcomes of HAI with regard to different SARS-CoV-2 variants.
METHODS: Patients with SARS-CoV-2 infections hospitalized between March 1st, 2020 and May 17th, 2022 in 79 hospitals of the Helios Group were included. Information on patients' characteristics and outcomes were retrieved from claims data. In accordance with the Robert Koch Institute, infections were classified as hospital-acquired when tested positive >6 days after admission and if no information hinted at a different source.
FINDINGS: In all, 62,875 SARS-CoV-2 patients were analysed, of whom 10.6% had HAI. HAIs represented 14.7% of SARS-CoV-2 inpatients during the Wildtype period, 3.5% during Alpha (odds ratio: 0.21; 95% confidence interval: 0.19-0.24), 8.8% during Delta (2.70; 2.35-3.09) and 10.1% during Omicron (1.10; 1.03-1.19). When age and comorbidities were accounted for, HAI had lower odds for death than community-acquired infections (0.802; 0.740-0.866). Compared to the Wildtype period, HAIs during Omicron were associated with lower odds for ICU (0.78; 0.69-0.88), ventilation (0.47; 0.39-0.56), and death (0.33; 0.28-0.40).
CONCLUSION: Hospital-acquired SARS-CoV-2 infections occurred throughout the pandemic, affecting highly vulnerable patients. Although transmissibility increased with newer variants, the proportion of HAIs decreased, indicating improved infection prevention and/or the effect of immunization. Furthermore, the Omicron period was associated with improved outcomes. However, the burden of hospital-acquired SARS-CoV-2 infections remains high.

Introduction

When the novel severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) started spreading throughout the world, its high transmissibility soon became apparent. Under unfavourable conditions, a single person was able to infect many others, causing so-called super-spreader events and outbreaks in different settings. This posed major challenges for hospitals to prevent nosocomial spread. Initially, knowledge about the disease and its transmission routes was scarce; this led to a wide variety of prevention measures being employed, including some that now seem superfluous, such as the use of coveralls or thermographic cameras or even setting up separate hospitals for the treatment of COVID-19 patients. At the beginning of the pandemic, there was broad consensus among the scientific community that the transmission occured mainly via droplets, like in other respiratory viruses and in rare cases via other body fluids or indirectly via surfaces [1]. The focus of prevention measures rested therefore on separation and isolation, the use of surgical masks, face shields, protective gowns, frequent hand washing [2] and enhanced environment disinfection protocols [3]. Increasing evidence of super-spreader events with transmissions among persons without direct contact and over significant distances made the airborne route appear increasingly plausible [4, 5]. This shifted the emphasis on measures such as masks with higher filtration capability (FFP2 or KN 95-masks) and towards improving indoor air using frequent ventilation and filtration.

Despite an increasing understanding of the infection, its transmission mechanisms and the corresponding tailoring of precautions, outbreaks have continued to occur throughout the pandemic in care facilities and hospitals. Little is yet known on the total burden of these infections in Germany during the course of the pandemic. This study analyses the occurrence and the outcomes of hospital-acquired SARS-CoV-2 infections within a large hospital group, as well as the influence of virus variants.

Methods

This study is a retrospective analysis based on claims and surveillance data. We included all inpatients hospitalized with the International Classification of Diseases (ICD-10) code U07.1 (= polymerase chain reaction-(PCR)-confirmed SARS-CoV-2 infection) as main or secondary diagnosis in 79 Hospitals of the Helios Group between March 1st, 2020 and May 17th, 2022. Helios is a privately owned company with hospitals spread throughout Germany. Its proportion of basic to tertiary care is comparable to the overall distribution of hospitals in the country, treatment in Helios hospitals being covered by all German health insurances. Patients from the claims data retrieval were linked to patients documented in the iNOK database, this being a Helios' own, group-wide, intranet-based surveillance program, in which trained infection control nurses document daily new patients with a SARS-CoV-2 infection. In accordance with the criteria of the Robert-Koch institute [6], infections were classified as being hospital-acquired as folows:

Positive SARS-CoV-2 PCR after day 6 of the inpatient stay and no other obvious source of infection, e.g. previously infected household members or visitors;

Positive SARS-CoV-2 PCR on day 3-6 of hospital stay and strong suspicion of transmission in hospital;

Readmission of a patient with a positive SARS-CoV-2 PCR ≤ 6 days after discharge and strong suspicion of transmission in hospital.

Claims and surveillance data were linked by a double-pseudonymised case number.

Information on sex, age, co-morbidities, treatment in an intensive care unit (ICU), mechanical ventilation and death were retrieved from claims data. Mechanical ventilation was defined as ventilation with pressure support via either invasive devices like tracheal tube or tracheostomy, or use of non-invasive devices. Mortality was defined as death during the same hospital stay. Cases with discharge due to hospital transfer were excluded from the calculation of the mortality rate. Claims data on co-morbidity was summarized in the Elixhauser Comorbidity Index [7], a score for categorizing patient co-morbidities based on ICD-10 codes. As surrogate parameter for the number of patients with symptoms among those infected with SARS-CoV-2, we included the documentation of a severe acute respiratory infection (SARI), as defined by the ICD-10-codes J09-J22 according to the method of SARI surveillance [8].

For further analysis, patients were associated with the variant that was prevalent at the time of them becoming infected. Actual results of variant analyses were not available. The time period of predominance of each variant was derived from to the weekly reports of the Robert Koch Institute on variant development in Germany [9] as follows:

Wild type period: February 4th, 2020 – March 7th, 2021

Alpha period: March 8th, 2021 – June 25th, 2021

Delta period: June 26th, 2021 – January 2nd, 2022

Omicron period: January 3rd, 2022 – present.

Statistics

Inferential statistics were based on generalized linear mixed models (GLMM) specifying hospitals as random factor[10]. Effects were estimated with the lme4 package (version 1.1-26) [11] in the R environment for statistical computing (version 4.0.2, 64-bit build) [12]. For all tests we apply a two-tailed 5% error criterion for significance. For the description of the patient characteristics of the cohorts, we employed χ2-tests for binary variables and analysis of variance for numeric variables. We report proportions, means, standard deviations, and p-values.

For the comparison of outcomes, we used logistic GLMMs with logit link function. We report proportions, odds ratios together with confidence intervals and p-values.

Logistic GLMMs were used for multivariable analysis of binary outcomes. The analysis of weekly HAI proportions was done via beta regression. Numerical predictors were centred on their mean and scaled to unit variance.

We report statistics for Elixhauser comorbidity index. For this index, the Agency for Healthcare Research and Quality (AHRQ) algorithm was applied [13].

Ethics

The local ethics Committee (vote: AZ490/20-ek) and Helios Kliniken GmbH data protection authority approved data use for this study.

Results

Occurrence of hospital-acquired SARS-CoV-2 infections

In total, 62,875 patients with a PCR-confirmed SARS-CoV-a2-infection were included. In 6,652 of them (11.8 %), the infection was categorized as hospital-acquired (HAI) (table I ).

Table 1

Characteristics and outcome of patients with community-acquired vs. hospital-acquired SARS-CoV2-infections. CAI, Community-acquired infection; HAI, Hospital-acquired infection; OR, Odds Ratio; 95% CI, 95 % confidence interval; ECI., Elixhauser Comorbidity Index, SARI, Severe Acute Respiratory Infection; ICU, Intensive Care Unit; SD, standard deviation; n.s, not significant.

Variable	Unit	Community-acquired infection (CAI)	Hospital-acquired infection (HAI)	OR (95% CI) HAI vs. CAI	P-value
All	n (proportion HAI)	56,223	6,652 (10.6 %)	n/a	n/a
Age (y)	mean (± SD)	62.5 ± 22.8	72.5 ± 17.0	n/a	<.01
Sex (male)	n (proportion)	28,625 (50.9 %)	3,415 (51.3 %)	n/a	n.s.
ECI	mean (± SD)	8.8 ± 10.7	15.8 ± 13.7	n/a	<.01
SARI	n (proportion)	31,918 (56.8 %)	2,794 (42.0 %)	0.57 (0.54-0.60)	< .001
ICU	n (proportion)	11,581 (20.6 %)	2,305 (34.7 %)	1.99 (1.88-2.11)	< .001
Mechanical ventilation	n (proportion)	7,727 (13.7 %)	892 (13.4 %)	0.92 (0.85-0.99)	.032
Mortality	n (proportion)	7,228 (14.5 %)	1,234 (20.9 %)	1.53 (1.43-1.64)	<.001
Wildtype	n (proportion HAI)	18.299	3.143 (14.7 %)	n/a	n/a
Age (y)	mean (± SD)	67.9 ± 19.3	73.7 ± 15.3	n/a	<.01
Sex (male)	n (proportion)	9,396 (51.3%)	1,560 (49.6%)	n/a	n.s.
ECI	mean (± SD)	10.4 ± 11.1	16.8 ± 13.4	n/a	<.01
SARI	n (proportion)	13,173 (72.0 %)	1,721 (54.8 %)	0.48 (0.44-0.52)	<.001
ICU	n (proportion)	4,450 (24.3 %)	1150 (36.6 %)	1.74 (1.60-1.89)	< .001
Mechanical ventilation	n (proportion)	3,046 (16.6%)	511 (16.3 %)	0.92 (0.81-1.02)	n.s.
Mortality	n (proportion)	3,310 (20.2 %)	760 (27.9 %)	1.49 (1.35-1.63)	<.001
Alpha	n (proportion HAI)	7.567	274 (3.5 %)	n/a	n/a
Age (y)	mean (± SD)	61.0 ± 19.5	71.1 ± 19.0	n/a	<.01
Sex (male)	n (proportion)	4,091 (54.1 %)	134 (48.9 %)	n/a	n.s.
ECI	mean (± SD)	8.7 ± 10.5	15.2 ± 13.3	n/a	<.01
SARI	n (proportion)	5,662 (74.8 %)	122 (44.5 %)	0.30 (0.23-0.40)	<.001
ICU	n (proportion)	2,005 (26.5 %)	90 (32.8 %)	1.53 (1.17-2.01)	.002
Mechanical ventilation	n (proportion)	1,610 (21.3 %)	44 (16.1 %)	0.76 (0.54-1.06)	n.s.
Mortality	n (proportion)	957 (14.1 %)	50 (20.4 %)	1.65 (1.18-2.30)	.003
Delta	n (proportion HAI)	11.285	1.085 (8.8 %)	n/a	n/a
Age (y)	mean (± SD)	60.9 ± 22.5	69.4 ± 19.3	n/a	<.01
Sex (male)	n (proportion)	5,921 (52.5 %)	586 (54.0 %)	n/a	n.s.
ECI	mean (± SD)	8.7 ± 10.5	15.7 ± 14.8	n/a	<.01
SARI	n (proportion)	7,340 (65.0 %)	406 (37.4 %)	0.33 (0.28-0.37)	<.001
ICU	n (proportion)	2,616 (23.2 %)	375 (34.6 %)	1.72 (1.49-1.98)	<.001
Mechanical ventilation	n (proportion)	2,068 (18.3 %)	151 (13.9 %)	0.69 (0.57-0.82)	<.001
Mortality	n (proportion)	1,678 (16.6 %)	201 (20.2 %)	1.24 (1.05-1.47)	.012
Omicron	n (proportion HAI)	19.072	2,150 (10.1 %)	n/a	<.01
Age (y)	mean (± SD)	58.9 ± 26.2	72.5 ± 17.7	n/a	<.01
Sex (male)	n (proportion)	9,219 (48.4 %)	1,135 (52.8 %)	n/a	< .01
ECI	mean (± SD)	7.5 ± 10.3	14.3 ± 13.5	n/a	< .01
SARI	n (proportion)	5,743 (30.1 %)	545 (25.3 %)	0.77 (0.69-0.85)	< .001
ICU	n (proportion)	2,510 (13.2 %)	690 (32.1 %)	2.93 (2.64-3.35)	< .001
Mechanical ventilation	n (proportion)	1,003 (5.3 %)	186 (8.7 %)	1.58 (1.33-1.86)	< .001
Mortality	n (proportion)	1,283 (7.7 %)	223 (11.4 %)	1.51 (1.30-1.76)	< .001

The number of hospital acquired infections varied during the pandemic. It peaked during the Wildtype period (208 HAI/week) and Omicron period (229 HAI/week) (Figure 1 , graph A). Generally, the number of HAI increased with the number of community acquired infections (CAI) (odds ratio OR = 2.871, 95% confidence interval 2.680 - 3.071, p < .001). However, the proportion of HAI among all inpatients did not remain constant, and rather varied between 0 % and 31.2 % (Figure 1, graph B). There was no association between the proportion of HAI and the number of CAI, i.e. periods with a high number of inpatients with community-acquired infections were neither associated with more nor with fewer transmissions (OR = 0.968, 95% CI= 0.830-1.123, p = 0.667). The fluctuations in the proportions were rather determined by the variants. During the Wildtype period, the proportion of HAI was highest, in average 14.7%; it decreased to 3.5 % during Alpha (OR = 0.21, 95% CI = 0.19 – 0.24, p< .001) and increased again to8.8 % during Delta (OR =2.70, 95% CI = 2.35 – 3.09, p < .001) and then to 10.1% during Omicron (OR = 1.10, 95% CI = 1.03 – 1.19, p = 0.015).

Figure 1

Graph A: Weekly incidences of community-acquired and hospital-acquired infections; Graph B: Proportion of hospital-acquired infections to all hospitalized patients with SARS-CoV2. The smooth curves were generated via LOESS (locally weighted scatterplot smoothing).

Outcomes of hospital-acquired SARS-CoV-2 infections

Among patients with hospital-acquired infections, 42.0% developed a severe acute respiratory infection (SARI), 34.7% were treated in ICU, 13.4 % were ventilated and 20.9% died (table I). Compared to patients with community-acquired infections, they were diagnosed less frequently with SARI (OR =0.57, 95% CI = 0.54-0.60) and received less frequently mechanical ventilation (OR = 0.92, 95% CI = 0.85-0.99), but were treated more often in ICU (OR = 1.99, 95% CI = 1.88-2.12). Mortality was higher among patients with HAI (OR = 1.52, 95% CI = 1.43-1.64). However, patients with HAI were on average 10.0 years older and had a 7.0-points higher ECI (p <.01) than patients with CAI. When age, sex and ECI were controlled for (table II , multivariable analyses), hospital-acquired infections were associated with a higher risk for ICU (OR = 1.44. 95 % CI = 1.35-1.53), but a lower risk for ventilation (OR = 0.597, 95% CI = 0.55 – 0.65) and death (OR = 0.80, 95% CI= 0.74 - 0.87).

Table 2

Multivariable analyses of risk factors for intensive care, mechanical ventilation and mortality, among all inpatients with a SARS-CoV-2.

	OR (95 % CI)	P-value
Intensive care
Male sex	1.627 (1.560-1.695)	< .001
Age	1.038 (1.010-1.064)	.004
Elixhauser Comorbidity Index	1.814 (1.770-1.854)	< .001
Hospital acquired vs. community acquired	1.439 (1.350-1.530)	< .001
Alpha vs.wildtype	1.290 (1.210-1.375)	< .001
Delta vs. wildtype	1.054 (1.000-1.114)	.065
Omicron vs. wildtype	0.577 (0.550-0.608)	< .001
Mechanical ventilation
Male sex	1.834 (1.740-1.928)	< .001
Age	0.961 (0.930-0.991)	.010
Elixhauser Comorbidity Index	1.940 (1.890-1.990)	< .001
Hospital acquired vs. community acquired	0.597 (0.550-0.649)	< .001
Alpha vs.wildtype	1.548 (1.440-1.661)	< .001
Delta vs. wildtype	1.224 (1.150-1.303)	< .001
Omicron vs. wildtype	0.323 (0.300-0.347)	< .001
Mortality
Male sex	1.620 (1.540-1.708)	< .001
Age	1.057 (1.050-1.059)	< .001
Elixhauser Comorbidity Index	1.057 (1.050-1.059)	< .001
Hospital acquired vs. community acquired	0.802 (0.740-0.866)	< .001
Alpha vs.wildtype	0.914 (0.840-0.994)	.037
Delta vs. wildtype	1.007 (0.940-1.078)	n.s.
Omicron vs. wildtype	0.362 (0.340-0.388)	< .001

OR, Odds Ratio; 95% CI, 95 % confidence interval; n.s, not significant.

The outcomes of HAI differed between the various periods of variant dominance, similarly to community-acquired ones (table I). ICU and mortality rates were highest during the Wildtype period (ICU: 36.6 %, mortality: 27.9 %). Outcomes became more favourable with each new variant. During the Omicron eriod, ICU rate dropped to 32.1 % and mortality rate to 11.4 %. Compared to infections during the Wildtype period, hospital-acquired infections during Omicron had the least odds for intensive care (OR = 0.78, 95% CI = 0.69 – 0.88, p < .001), ventilation (OR = 0.47, 95% CI = 0.39 – 0.56, p < .001), SARI (OR = 0.29, 95% CI = 0.25 – 0.33, p < .001) and death (OR = 0.33, 95% CI = 0.28 – 0.40, p < .001).

Discussion

We analysed 62,875 patients treated with SARS-CoV-2 in the hospitals of the Helios group. Among them, around 11 % had acquired the infection during the hospital stay. Earlier studies focussing on the first wave described proportions of hospital-acquired infections of 12% in Wuhan [14] and of 15% - 20% in England [[15], [16], [17], [18]]. On the other hand, a single centre study from Boston detected only one HAI among nearly 10,000 inpatients during the first weeks of the pandemic [19].

In order to estimate the total occurrence of hospital-acquired infections in Germany, the number of patients hospitalized since the beginning would be needed. However, due to incomplete reporting, the exact number is not known. Based on the total number of infections in the population and on the proportion of hospitalisations among the available reports [9], it can be roughly estimated that about 915,000 COVID-19 patients have been hospitalized since the beginning of the pandemic and until April 2022. Assuming a similar burden of hospital-acquired infections in all hospitals in Germany, up to 110,000 patients might have been affected by hospital-acquired SARS-CoV-2 infections in Germany since the beginning of the pandemic.

HAI have occurred despite comprehensive infection control programmes. The hospital architecture in Germany is far from ideal regarding infections with an airborne transmission route: most rooms are designed to accommodate 2 to 3 patients and mechanically ventilated rooms are common only in units for intensive care or stem cell transplants. A major problem is a low nurse-to-patient ratio. Germany introduced a law regulating this ratio only in 2019 and suspended it during the first months of the pandemic. The legally required standard of 1 nurse per 2 patients in ICU and 1 per 10 in a medical-surgical units is still lower than the given average in other European countries or the USA [20]. Low nurse-to-patient ratios have been associated with a higher risk for HAI [[21], [22], [23]].

Some of the measures taken e.g. universal masking, screening of patients and personal, increased hand hygiene and frequent room ventilation created high costs and discomfort, but usually no major harm. Others e.g. quarantine and isolation, visitors’ limitations, postponing elective treatments and the limitation of in-hospital meetings and teachings had severe consequences for patients and staff alike, such as anxiety [24], depression [25], delirium [26], suicidality[27] and insufficient or delayed medical treatment[25, 28]. Which prevention measures were most effective, which proved superfluous and what additional efforts (if any) should have been implemented, still remains to be evaluated. At this time it seems hardly conceivable that even more severe restrictions could have been imposed on patients, relatives and health care workers to prevent hospital-acquired infections.

Occurrence of HAI during different periods of variant predominance

The highest number of HAI per week occurred during the Omicron period in March/April 2022. This was also the period that saw the highest numbers of inpatients with SARS-CoV-2 and of infections in the population since the beginning of the pandemic [29]. More infected people among staff and visitors unfortunately led to more hospital-acquired infections. However, the proportion of HAI among all inpatients with SARS-CoV-2 was highest during the first and second wave of the pandemic, i.e. during the Wildtype period, decreased during the Alpha period and increased again with Delta and Omicron. At the same time, with each new variant the Coronavirus became more contagious. Estimations on the original virus reported a basic reproductive number (R0) - the average number of people on to whom an infected person will pass the virus in a totally naïve population and in the absence of preventive measures-of around 2-3 [30, 31]. With the Alpha variant, R0 reached 4-5, [32], 5-8 with Delta [33] and 9-10 with Omicron [32, 34]. The major decrease of the proportion of HAI during Alpha can probably be attributed to, to improvements in prevention measures and in outbreak management, but above all to the vaccination campaign, which focused during the first months of 2021 on health care workers and persons at risk, providing initially a high level of protection against the infection. The later rebound is probably the consequence of the increase in transmissibility associated with Delta and Omicron.

Outcome of hospital-acquired SARS-CoV-2 infections

21% of patients with a hospital-acquired infection died in association with it. Extrapolated to the estimated number of hospitalized patients with SARS-CoV-2, up to 23,000 might have died in connection with a hospital-acquired SARS-CoV-2 infection in Germany since beginning of the pandemic. This would correspond to the order of magnitude of fatalities anticipated to result in the same period from all other common hospital-acquired infections combined, such as surgical site infections, pneumonia and urinary tract infections, including those caused by multi-drug resistant organisms [35].

However, not all deaths of patients with a hospital-acquired SARS-CoV-2 infection will have been caused directly by the infection. When the risk factors for an adverse outcome age, gender and co-morbidities were accounted for, hospital-acquired infections were associated with lower odds for death than community-acquired infections. Furthermore, patients with HAI were less likely to be diagnosed with a severe respiratory infection. Both features indicate that there might have been a greater proportion among patients with HAI than among those with CAI, in whom the infection was a secondary if not incidental finding, perhaps detected in the course of routine testing and with no greater clinical relevance. These patients likely had other reasons for being hospitalized than a SARS-CoV-2 infection, leading to intensive care, mechanical ventilation and death independently from a SARS-CoV-2 infection.

The Omicron period was associated with a significantly more favourable disease course. This is concordant to other studies comparing the outcome of Omicron to earlier variants [36, 37]. In addition to an attenuated virulence of the coronavirus, this development is probably attributed to the increasing immunisation of the population. Furthermore, during the Omicron period, only a quarter of the patients with HAI were diagnosed with a severe respiratory infection, indicating an increasing proportion of secondary findings. However, in spite of the decreasing relevance of hospital-acquired infections and the more benign outcomes since the Omicron variant, the mortality rate of patients with HAI was still high.

Strength and limitations of the study

To the best of our knowledge, this is the first study to analyse the occurrence and the outcome of hospital-acquired SARS-CoV-2 infections in a large cohort in Germany as well as to show the changes that came with new variants. However, there are limitations to the interpretation of the results.

The main parameter of the study, the classification of the infections into community-versus hospital-acquired relies on an estimation. The criteria were set and all cases were evaluated individually, including patients’ history, lab and imaging results in unclear cases. Still, in some cases, the route of infection could not be traced with absolute certainty. The high number of infection control nurses involved probably caused a high inter-rater variability. The median incubation period of the early variants was 5 days [38], of Delta 4.3 [39] and of Omicron 3 days [40]. Patients might have been infected in the first day(s) after admission and still tested positive before day 7, leading to the misclassification as community-acquired infections. Patients with disease manifestation after discharge where registered only when readmitted. On the other hand, approximately 25% of infections have an incubation period of more than 7 days [38] and can therefore have been falsely allocated as hospital-acquired. False negative results in early stages of the disease or delayed testing, caused by limited access to testing early in the pandemic or by the lack of typical symptoms will also have resulted in a misclassification into hospital-acquired infections. As there was no universal and repetitive screening in the beginning of the pandemic, underreporting is probable.

Conclusions

The pandemic has represented an unprecedented challenge for health care workers and infection prevention & control teams in their efforts to avoid hospital spreading. In spite of extensive prevention measures, hospital-acquired SARS-CoV-2 infections have occurred since the beginning of the pandemic, affecting a highly vulnerable population group. Fortunately, although the transmissibility has increased with new variants, the proportion of hospital-acquired infection has decreased, indicating an improvement in infection prevention. Furthermore, the Omicron period was associated with better outcomes. However, the burden of hospital-acquired SARS-CoV-2 infections remains still high. Further research is urgently needed to define prevention measures adequate to lower this burden with the greatest benefit and least harm.

Funding statements

No funding to declare.

Declaration of Competing Interest

M.B., S.H., C.K., J.L., K.S., A.B., R.M., R.K., I.N. have no conflict of interest.