1. Introduction
The global landscape of the COVID-19 pandemic has evolved significantly, transitioning from a state of emergent crisis to an enduring endemic phase [
1,
2]. Since January 2022, the SARS-CoV-2 Omicron variant and its subvariants have been dominant among SARS-CoV-2 infections in Germany [
3,
4]. Various measures, in particular the development of vaccines and mandated hygiene measures, have led to a reduction in mortality rates and in the health care burden for patients [
2,
4,
5]. However, approximately 154,000 COVID-19-related deaths still occurred in Germany in 2022 despite a base immunization rate of 76.4% [
6]. Therefore, identifying vulnerable patient populations for severe outcomes of COVID-19 is still of immediate interest [
2]. Among the variables influencing the disease severity of SARS-CoV-2 infections, the immune competence of patients emerges as a critical determinant, as immunocompromised (IC) patients demonstrate increased susceptibility to severe courses of the disease [
1,
7,
8,
9,
10,
11]. An ongoing individual immunocompromised state for, but not limited to, SARS-CoV-2 could be explained by a multi-factorial genesis on humoral and/or cellular mediated immunity [
8], such as reduced antibody concentrations after antigen exposure [
12], less neutralization activity of antibodies [
13] or an accelerated decline in immunity after vaccination [
14]. Despite the preventive efficacy of (active) vaccination among non-IC patients, for patients with insufficient immune competence, the availability of alternative antiviral treatment options, such as antiviral agents and monoclonal antibodies (passive immunization), offer additional options for the prevention and treatment of aggravated COVID-19 cases. [
15]. In addition, there is also evidence for an ongoing need for intensified prevention and treatment of vulnerable IC patients regarding hospitalization [
2,
16]. In the Omicron era, the INFORM study revealed, for broadly defined IC patients in the UK, a base immunization rate of 93.6% and a rate of 84.5% for at least three active vaccinations, yet these patients still experience a confounder-adjusted 2.16 times higher risk of hospitalization [
1]. Certain IC patient sub-cohorts such as solid organ transplant patients, experienced up to 14-fold increased risks regarding severe outcomes. Further, depending on the definition of IC patients, at least 4% of the population of Germany can be classified as immunocompromised [
11], leading to an anticipated insufficient vaccination response of approximately 2% of the German population (1.5 million individuals) [
17]. Current data on the in-hospital outcome for German cohorts, including hospitalized IC patients, are still rare [
4,
18], especially for patient populations with critical clinical courses who present with severe acute respiratory infections (SARI) related to infections with the endemic SARS-CoV-2 Omicron subtype.
This retrospective, cross-sectional, observational real-world data analysis within a German hospital cohort aims to evaluate the in-hospital course of IC patients infected during a period dominated by SARS-CoV-2 Omicron subtypes. Therefore, the primary objective of our study is the in-hospital outcome for a composite endpoint of severe courses in patients with COVID-19-related SARI, stratified for immune status. Additionally, the impact of SARS-CoV-2 infections within the IC cohort will be systematically evaluated as a secondary objective.
2. Methods
Data collection: In this retrospective, observational study, we analyzed administrative data from 84 hospitals (primary to tertiary care centers and university hospitals) within the national Helios group in Germany. The main diagnosis and comorbidities were identified according to the International Statistical Classification of Diseases and Related Health Problems (ICD-10-GM [German Modification]) [
19]. In-hospital cases of pre-defined cohorts undergoing full inpatient treatment and admitted between 1 January 2022 and 31 December 2022 were included.
For the primary objective, we analyzed patient cohorts with COVID-19-related SARI with and without IC-related diagnoses. For contextualization, as a secondary objective, we analyzed IC patients with and without COVID-19-related SARI. The sub-cohorts of IC patients without COVID-19-related SARI also excluded COVID-19 patients without SARI to reduce confounding. The presence of SARS-CoV-2 infection was determined by laboratory-confirmed infection using the ICD-10-GM code U07.1. Comorbidities were classified by ICD-10-GM based Elixhauser comorbidity score with the Agency for Healthcare Research and Quality (AHRQ) algorithm [
20,
21]. SARI was defined in our claims dataset per the method of SARI surveillance if any of the ICD-10 diagnosis codes from ICD J09-J22 were coded, including ICD J09-J18 for influenza and pneumonia, ICD J20 for acute bronchitis, ICD J21 for acute bronchiolitis, and ICD J22 for acute lower respiratory tract infection [
4,
22,
23].
IC patients were defined by the ICD-10 diagnosis claims database if at least one diagnosis met the classification of COVID-19-related broadly defined IC patients [
16,
24]. ICD-10 diagnoses will be coded during inpatient stay, if it is of significant relevance to the inpatient treatment. These ICD-10-based defined IC patients are assessed as clinically moderate to severe vulnerable conditions [
10], including solid tumors, solid organ transplants, end-stage kidney disease, HIV infection, congenital immunodeficiency, autoimmune diseases, hematological malignancy, and end-stage chronic liver diseases. Detailed ICD-10-based definitions for IC-cohorts are provided in
Supplementary Material Table S1.
In-hospital medical procedures were used to describe the extent of in-hospital care and were identified by coded operations and procedures codes (OPS) [
25]. Severe in-hospital cases were classified per the WHO stages of COVID-19 [
26], including the following stratification: treatment in an intensive care unit (ICU) (OPS-codes 8-980, 8-98f or length of stay on ICU > 0 days), mechanical ventilation (OPS-codes 8-70x, 8-71x or length of mechanical ventilation > 0 days) or in-hospital death (determined by hospital discharge type in administrative data). Cases discharged to another hospital were excluded for analysis of mortality. In addition, the healthcare burden was analyzed by length of stay and hospitalization costs based on the total cost of healthcare insurance.
The study followed the STROBE guidelines for observational analysis [
27]. Patients’ data were stored in a double-pseudonymized form, and the use of the data was approved by the local ethics committee of the University of Leipzig (AZ490/20-ek) and the Helios Kliniken GmbH data protection authority. Considering the retrospective analysis of double-pseudonymized administrative routine clinical data, individual informed consent was neither applicable nor obtained.
Statistical analysis: Administrative data were extracted from QlikView (QlikTech, Radnor, PA, USA), and all analyses were performed within the R statistical computing environment. Baseline characteristics were presented as proportions (for categorical variables) and either means with standard deviations (for continuous variables) or medians with interquartile range (IQR). To describe the patient characteristics of the cohorts and comorbidities, we employed χ2-tests for categorical variables and a two-sample
t-test for numerical variables. We used logistic generalized linear mixed models (GLMM; [
28]) with a logit link function to compare the proportions of selected treatments and outcomes in the different cohorts. In all mixed models, we specified varying intercepts for the random factor (hospitals). The numeric variables, length of stay, and hospitalization costs were analyzed via linear mixed models (LMM). Because these variables were positively skewed, we transformed them via inverse hyperbolic sine to roughly approximate normal distributions. We apply a two-tailed 5% error criterion for significance for all tests.
Multivariable analyses of outcomes were performed via logistic GLMM and LMM. Categorical variables entered the analyses as simple-coding contrasts, while the continuous variable age was centered at its mean. All continuous variables were scaled to unit variance.
3. Results
Out of 87 hospitals in the German-wide Helios hospital network, 84 contributed (
n = 146,324) cases for the retrospective data analyses. A flow chart of cohort identification is presented in
Figure 1. A total of 16,809 cases were identified as COVID-19-related SARI cases. Of these, 14,772 (87.9%) cases were non-IC patients, and 2037 (12.1%) were IC patients. Additionally, 129,515 IC hospitalized patients were identified without COVID-19 and SARI. We have not further specified cases without COVID-19 and without an underlying IC disease, as this would include all patients in the hospital network and is not related to the research question.
Throughout the analyzed period, we observed a fluctuating trend in hospitalization with three major waves of SARS-CoV-2 infection in 2022: as illustrated in
Figure 1 for IC and non-IC patients, the first wave of infection ran from January to April, the second from May to August and the third from September to December (
Figure 2).
3.1. Baseline Characteristics of Patients with COVID-19-Related SARI, with and without IC
Baseline characteristics for COVID-19-related SARI cases are presented in
Table 1. The IC patient cohort was significantly younger (−1.9 years,
p < 0.001), had significantly more comorbidities (+9.3 Elixhauser comorbidity score,
p < 0.01), and included 3% more men (
p = 0.015). Highest comorbidity incidences in COVID-19-related SARI cases were fluid and electrolyte disorders (52% in non-IC, 54% in IC,
p = 0.214), uncomplicated arterial hypertension (45% in non-IC, 42% in IC,
p = 0.007), renal failure (35% in non-IC, 39% in IC,
p < 0.001), cardiac arrhythmias (34% in non-IC, 28% in IC,
p < 0.001) and congestive heart failure (31% in non-IC, 29% in IC,
p = 0.050). A list of incidences for Elixhauser comorbidity clustered diagnoses is provided in the
Supplementary Material Table S2.
3.2. In-Hospital Outcome for Patients with COVID-19-Related SARI with and without IC
The primary objective of evaluating the severe in-hospital course for COVID-19-related SARI revealed that IC patients exhibited a significantly higher prevalence rate of 42%, compared to 32% in patients without an IC diagnosis, resulting in an odds ratio (OR) of 1.4 (
p < 0.001, see
Table 2). This increased risk was based on an increased prevalence in all severity stratifications: ICU treatment (OR = 1.3,
p < 0.001), increased prevalence of mechanical ventilation (OR = 1.2,
p = 0.020) and increased in-hospital mortality (OR = 1.5,
p < 0.001). The length of hospital stay was prolonged by 4.6 days (41%) for IC vs. non-IC patients affected with COVID-19-related SARI and the cost of hospitalization was 49% higher in these cases (IC patients: vs. non-IC patients: 12,611.90€ vs. 8318.80€,
p < 0.001).
Table 2 summarizes the in-hospital outcomes for the primary research objective.
In the diagnosis-stratified sub-cohorts of IC patients with COVID-19-related SARI, we found different outcomes for the prevalence of severe in-hospital courses compared to non-IC patients (
Table 2). The risk for severe outcomes (composite endpoint) of IC sub-cohorts is presented in
Figure 3. Across all IC sub-cohorts, there was a trend towards a higher prevalence of severe outcomes, either within the composite endpoint itself or within its specific sub-domains. A significantly higher prevalence of ICU treatment was observed for IC patients with solid organ transplants, end-stage chronic liver diseases, and autoimmune diseases. ICU treatment with mechanical ventilation was at a significantly higher prevalence for patients with end-stage chronic liver diseases. Patients with solid tumors and end-stage chronic liver diseases showed a significantly increased in-hospital mortality compared to non-IC patients.
3.3. Baseline Characteristics of Hospitalized IC Patients, with and without COVID-19-Related SARI
A total of n = 129,515 hospitalizations of IC patients and n = 2037 cases of IC patients with COVID-19-related SARI occurred during the analyzed Omicron-dominant period per our IC definition. IC patients had a mean hospitalization case rate per month of n = 10,793 during the defined Omicron period.
In comparison to IC patients with COVID-19-related SARI, the IC patient cohort without COVID-19-related SARI was significantly younger (66.3 vs. 71.2 years,
p < 0.001), had 4% fewer males (55% vs. 59%
p < 0.001) and had a lower Elixhauser Comorbidity score (15.3 vs. 21.4
p < 0.001). Patient characteristics for total IC patients are shown in
Table 3.
Main comorbidities were less prevalent in IC patients compared to IC patients with COVID-19-related SARI, except for uncomplicated arterial hypertension. The most prevalent comorbidities were fluid and electrolyte disorders (18% in IC, 54% in IC SARI,
p < 0.001), uncomplicated arterial hypertension (44% in IC, 42% in IC SARI,
p = 0.104), renal failure (22% in IC, 39% in IC SARI,
p < 0.001), cardiac arrhythmias (15% in IC, 28% in IC SARI,
p < 0.001) and congestive heart failure (12% in IC, 29% in IC SARI,
p < 0.001). The data for comorbidities within the total IC cohort and stratified for IC sub-cohorts are provided in the
Supplementary Table S2.
3.4. In-Hospital Outcome for IC Patients with and without COVID-19-Related SARI
In-hospital outcomes for the cohort of IC patients with and without COVID-19-related SARI are shown in
Table 4. Results stratified by diagnosis-based IC sub-cohorts are shown in
Figure 4 for the composite endpoint of severe outcomes, and results for subdomains ICU treatment, mechanical ventilation, and mortality are provided in the
Supplementary Material Table S4. Analysis of the secondary objective for IC patients with and without COVID-19-related SARI showed an odds ratio of 4.0 for the severe course during hospitalization for IC patients with COVID-19-related SARI (
p < 0.001). Severe courses were seen in 42% of IC patients with COVID-19-related SARI compared to 14% in the total IC cohort. More specifically, the odds ratios for IC patients with COVID-19-related SARI were 3.1 for ICU treatment (
p < 0.001), 10.0 for mechanical ventilation (
p < 0.001), and 6.0 for in-hospital mortality (
p < 0.001). These findings were confirmed in all IC sub-cohorts stratified by IC diagnosis. Notably, patients with hematologic disease had the highest risk of severe in-hospital outcomes for COVID-19-related SARI compared to the total IC cohort (OR for severe course = 12.0,
p < 0.001).
Multivariable analyses for in-hospital outcomes in IC patients identified COVID-19-related SARI as the variable with the highest OR of 3.23 for the composite endpoint of severe outcomes, as well as sex (male, OR = 1.07, p < 0.01), age (OR = 1.07, p < 0.001), and Elixhauser comorbidity score (OR = 2.67, p < 0.01), showing significant association with severe outcomes. These findings from the multivariable analysis showed that significantly increased OR for COVID-19-related SARI on hospital outcomes in IC patients were concordant for ICU treatment (OR = 2.59, p < 0.01), mechanical ventilation (OR = 8.07, p < 0.001) and in-hospital mortality (OR = 4.44, p < 0.001).
An increased in-hospital length of stay was associated with male sex, age, Elixhauser comorbidity score, and the prevalence of COVID-19-related SARI. Cost for hospitalization showed an association with the male sex, COVID-19-related SARI, and Elixhauser comorbidity scores.
Table 5 summarizes the results of multivariable analyses for in-hospital courses of IC patients.
4. Discussion
The global course of the COVID-19 pandemic has undergone a paradigm shift, transitioning from an acute pandemic crisis to a protracted endemic state. In this dynamic development, a nuanced understanding of the disease’s impact on various patient cohorts, especially those with reduced immune competence, remains important [
2]. Our retrospective, cross-sectional observational analysis evaluated in-hospital outcomes for COVID-19-related SARI within a multi-centric German in-hospital cohort, providing evidence on the risk stratification of severe courses in IC patients during the emergence of the Omicron variant.
The temporal trend analysis for 2022 revealed recurrent waves of SARS-CoV-2 infection, with a congruent trend in IC and non-IC patients, since peak infection rates occurred in the same months. This observation suggests that the infection rates of IC patients are driven by the general trend of infection rates in Germany, a pattern consistent with findings reported in the United States [
10]. Ketkar et al. demonstrated a close link in the incidence of infection between non-IC and IC patients over the course of the pandemic until 2022 in the US [
29]. Preventive measures to protect all patients can also indirectly protect IC patients, as this pattern was repetitively shown over different parts of the pandemic and in different countries.
The characteristics of our IC cohort showed a lower mean age, a higher number of comorbidities, and a slightly higher proportion of men compared to non-IC patients, aligning with other studies in hospitalized IC cohorts [
30,
31]. Notably, the most prevalent comorbidities in both cohorts included fluid and electrolyte disorders, uncomplicated arterial hypertension, renal failure, cardiac arrhythmias, and congestive heart failure. These latter four comorbidities were also found as most prevalent in IC patients by Singson et al. in a multi-state observational study conducted in the United States [
31]. The authors reported a proportion of patients with three or more comorbidities of 69.3% but did not provide a mean number or index for total comorbidities. Our IC cohort also showed a high proportion for comorbidities of 85% for at least five points on the Elixhauser comorbidity index.
Regarding the primary objective of our study, which involved analyzing severe in-hospital outcomes for patients with COVID-19-related SARI based on immune status, we provided evidence of persistently elevated rates for severe courses in IC individuals during the Omicron period. An odds ratio of 1.4 demonstrated an increased risk for IC patients compared to non-IC patients with COVID-19-related SARI. Nevejan et al. reported a higher OR of 2.1 for severe outcomes in a more narrow defined in-hospital IC cohort across a small number of hospitals [
30]. Our cohort reflects a broader perspective of in-hospital care, encompassing primary, secondary, and tertiary care providers, which may explain the lower risk for severe outcomes. The increased risk for severe outcomes is further supported by increased rates of ICU treatments, mechanical ventilation, and in-hospital mortality. The odds ratios for ICU treatment and death were 1.3 and 1.5, respectively, which are consistent with the odds ratios reported by Singson et al., who reported ORs of 1.4 for ICU treatment and 1.9 for mortality in vaccinated IC patients [
31]. Additionally, a meta-analysis conducted by Khoury et al. reported risk ratios of 1.9 for overall mortality among cancer patients [
32]. Similarly, in our study, patients with solid tumors or end-stage chronic liver disease had significantly increased risks for the severe courses, consistent with findings from previous studies [
32,
33]. Although patients with hematologic diseases and end-stage chronic kidney disease also showed a trend towards increased risk for severe outcomes, this did not reach significance, probably due to the smaller sample size for these less prevalent diseases. However, the trend is consistent with data provided from other observational studies [
1,
34].
Yin et al. also reported an increased hospital length of stay and mortality for IC patients, showing a comparable increase in mortality of 50% [
35]. Our data also confirm the mortality rate (OR 1.44) for hospitalized patients found by Turtle et al. in the United Kingdom [
36], although the majority of cases considered here were the early variants of SARS-CoV-2. Thus, compared to our data, the risk of IC patients is still relatively present compared to non-IC patients. Invasive ventilation showed heterogeneous results in specific IC subgroups in sub-analyses provided by Turtle et al. (2023), for example lower rates in cancer patients. While we found a clear trend toward higher rates of mechanical ventilation, not all sub-cohorts showed a statistically significant increased risk. Prolonged hospital stays and increased hospitalization costs underscore the increased healthcare burden in IC individuals infected with the Omicron variant. The EPOCH study found that hospitalization costs were, on average, 2.46 times higher for their IC composite cohort when comparing pre vs. post-COVID-19 hospitalization costs [
10]. In our analysis, hospitalization costs for IC patients with COVID-19 were 2.19 times higher compared to IC patients without COVID-19, confirming the findings of the EPOCH study. In addition, hospitalization costs at the onset of the COVD-19 pandemic in the year 2020 were at comparable levels for IC patients, as we found for our 2022 cohort [
18].
Age and sex (male) were identified as predictors for severe in-hospital outcomes in our COVID-19-related SARI cohort, consistent with findings reported by Nevejan et al. for IC patients and by Bonsignore et al. for non-IC patients [
30,
37]. Among patients who developed a COVID-19-related SARI, IC patients showed increased risk for severe in-hospital courses.
Therefore, while active vaccination reduces susceptibility to severe disease and our clinical treatment practices continue to accumulate knowledge, our data support the recommendation to provide continued special attention to vulnerable groups [
2,
15,
38].
The secondary objective, analyzing the impact of COVID-19-related SARI in IC cohorts, showed a fourfold increased risk of severe outcomes in IC patients with COVID-19-related SARI compared to IC patients without COVID-19-related SARI. Multivariable analyses confirmed this increased risk for severe outcomes, even after correcting for age, male sex, and comorbidity scores. SARS-CoV-2 infection and SARI also showed a significant effect when analyzed separately by multivariate analysis (SARS-CoV-2 infection OR = 1.31, p < 0.001; SARI OR = 3.23, p < 0.001). Therefore, SARS-CoV-2 infections and the development of COVID-19-related SARI have a significant impact on the in-hospital outcome of IC patients. Increased risks for severe in-hospital outcomes were not homogenous between all IC sub-cohorts. However, despite a small cohort of HIV patients, all IC sub-cohorts showed a pronounced increased risk for severe in-hospital outcome with OR ranging from 2.9 (solid organ transplants) to 12.0 (hematological diseases). This finding underlines the need for diagnosis-specific patient risk evaluations among hospitalized patients in the IC sub-cohort.
Our study results highlight both clinical and public health implications. The particular vulnerability of immunocompromised individuals requires an increased awareness during in-hospital treatments and underlines the importance of strategies like proactive vaccination initiatives [
39], preventive measures like pre-exposure prophylaxis (PREP) [
40], and general public hygiene rules.
Different active vaccines may also trigger different immune responses [
41]. Comorbidities and genetic predispositions also influence immunocompetence and symptom severity [
42]. Increased vaccination regimens did not necessarily result in improved humoral or cellular immune responses in IC patients [
43]. Moreover, different vaccines were proven for different effectiveness in IC patients in observational and more real-world driven studies [
44]. Further studies in IC patients could lead to specific recommendations in this context. In addition, antiviral therapies with passive immunization, for example, have been recommended in the German COVID-19 guidelines to be considered in individual cases for patients with a reduced immune response [
45]. Our data suggest that diagnoses and comorbidities can be used to specifically identify vulnerable patients and to protect patients in the context of hospitalizations and medical interventions. In summary, this study contributes additional evidence regarding the impact of the Omicron variant on immunocompromised cohorts. Analysis of in-hospital outcomes, differential risk factors, and disease-specific comorbidities contributes to a refined understanding of the determinants that drive disease severity in this vulnerable population. Further research efforts are necessary to continually refine our findings and identify and evaluate targeted interventions to optimize outcomes in immunocompromised individuals affected by SARS-CoV-2 infections.
5. Limitations
This study is based on administrative data primarily collected for remuneration purposes rather than research interests, which potentially could affect the encoded information. The quality of our results depends to a large extent on the accurate encoding of procedures and diagnoses at hospital discharge. Nevertheless, claims data are continuously monitored and checked at the hospital level by medical coding personnel and health insurance providers, and were introduced in 2022 as a surveillance method in Germany by the federal Robert Koch Institute [
46]. Moreover, reimbursement companies and health insurance providers continually evaluate administrative hospital discharge data to ensure data accuracy.
In addition, the ICD-10-based assessment of IC states did not include all patients undergoing specific therapies for immunosuppression due to the unavailability of information on medication. Furthermore, it is also not possible to estimate the severity of immune suppression. Therefore, an underestimation of IC prevalence is possible in our cohort. In addition, our cohorts of diagnosis-based IC patients varied in sample size, which led to tendencies for severe outcomes without reaching significant levels due to higher confidence intervals.
Due to the data structure and data protection regulations, especially case pseudonymization, linking patient data between different hospitals was not possible. This limitation implies the possibility of double-counting specific patients hospitalized multiple times or due to multiple SARS-CoV2 infections within our data periods. The claims-based analysis did not include vaccination information. However, since it is known that IC patients have higher vaccination rates than non-IC patients, our risk increases tend to be conservative and thus ultimately reflect the clinical real-world outcomes.