Abstract
Background
Coronavirus disease 2019 (COVID-19) placed substantial pressure on intensive care units (ICUs), but long-period comparisons with non-COVID-19 ICU patients remain useful for distinguishing disease-associated patterns from the general burden of critical illness. We addressed the gap in single-center data linking COVID-19 status with mortality, healthcare-associated infections, admission physiology, and Therapeutic Intervention Scoring System-28 (TISS-28) workload in a mixed tertiary ICU.
Methods
This single-center retrospective cohort study included first admissions to the adult ICU of the University Clinical Hospital in Białystok, Poland, between January 1, 2017, and June 1, 2023. Patients were classified as COVID-19-positive (n = 355) or COVID-19-negative (n = 2971) according to reverse transcription-polymerase chain reaction testing. Baseline characteristics, comorbidities, healthcare-associated infections, admission laboratory and arterial blood gas variables, and TISS-28 variables were compared using Welch t-tests, chi-square tests, or Fisher exact tests, as appropriate. Effect sizes are reported with 95% confidence intervals. Logistic regression estimated the association between COVID-19 status and in-hospital mortality in unadjusted, baseline-adjusted, and exploratory complete-case models.
Results
In-hospital mortality was higher in COVID-19-positive patients than in COVID-19-negative patients (227/355 [63.9%] vs. 1320/2971 [44.4%]; odds ratio [OR] 2.22, 95% CI 1.77 to 2.79). This association remained after baseline adjustment (adjusted OR 2.03, 95% CI 1.60 to 2.58) and in the exploratory complete-case model including admission physiology and laboratory markers (adjusted OR 2.73, 95% CI 1.60 to 4.64). Among non-survivors, time to death was shorter in the COVID-19-positive group (10.7 [SD 7.4] vs. 14.3 [SD 20.0] days; mean difference − 3.63 days, 95% CI -5.07 to -2.18). Bacterial bloodstream infection was more frequent in COVID-19-positive patients (22.8% vs. 9.9%; OR 2.70, 95% CI 2.05 to 3.56). At admission, COVID-19-positive patients had lower C-reactive protein, procalcitonin, PaO2, creatinine, and lactate, but higher PaCO2, glucose, sodium, potassium, bicarbonate, and hemoglobin values. TISS-28 profiles differed by COVID-19 status, most notably for respiratory physiotherapy recorded at least once (97.5% vs. 62.0%; OR 23.63, 95% CI 11.76 to 47.52) and longer duration of respiratory physiotherapy (mean difference 4.89 days, 95% CI 3.57 to 6.22).
Conclusions
In this retrospective cohort, COVID-19-positive status was associated with higher in-hospital mortality, shorter time to death among non-survivors, more frequent bacterial bloodstream infection, and a distinct ICU workload profile. These findings should be interpreted as associations rather than causal effects because of the single-center design, long heterogeneous study period, incomplete severity-score data, and lack of shift-level staffing and high-dependency-unit data.
Keywords: COVID-19, SARS-CoV-2, Intensive care unit, ICU mortality, Bloodstream infection, TISS-28, Healthcare-associated infections, Critical care organization
Background
SARS-CoV-2 infection created a rapid and sustained increase in demand for hospital and ICU resources. Early large-scale clinical data categorized approximately 14% of reported COVID-19 cases as severe and 5% as critical, while global surveys reported that approximately one-third of hospitalized COVID-19 patients developed acute respiratory distress syndrome (ARDS) and that approximately three-quarters of COVID-19 patients admitted to the ICU had ARDS [1, 2]. 5 These estimates help explain why pandemic waves affected not only bed availability but also ventilator use, oxygen-delivery systems, infection-control procedures, and bedside staffing [3, 4].
ICU bed capacity per capita is an important system-level indicator because a staffed ICU bed represents a combination of physical infrastructure, monitoring equipment, ventilators, organ-support capability, trained nurses, and physicians with critical care competencies. OECD data emphasized that ICU capacity was an essential pandemic resource and that, across 29 OECD countries, there were 16.9 ICU beds per 100,000 population in 2021; however, the same report also noted that staffing was often a more pressing constraint than bed numbers alone [5]. Therefore, interpreting ICU outcomes during COVID-19 requires attention to organization of critical care services, not only to diagnostic category.
Critical care is organized differently across countries and hospitals. In some systems, medical ICUs are led by pulmonary and critical care medicine specialists, whereas surgical or perioperative ICUs are traditionally led by anesthesiologists. In many European countries, including Poland, intensive care is closely integrated with anesthesiology and intensive therapy, and national guidance for admission to anaesthesiology and intensive care units is directed at specialists in anaesthesiology and intensive care and physicians undergoing specialist training in that field [6]. These organizational differences may affect case mix, admission thresholds, specialist availability, nurse staffing models, and transfer pathways.
Most early COVID-19 ICU studies focused on COVID-19 patients alone, including predictors of mortality, inflammatory markers, ventilator-associated pneumonia, bloodstream infection, and urinary tract infection [7–11]. Although these studies were clinically informative, they did not fully show how COVID-19 ICU admissions differed from non-COVID-19 ICU admissions treated in the same institutional environment. Long-period comparative studies can help identify which outcomes and workload elements are associated with COVID-19 status rather than with critical illness in general, while also requiring cautious interpretation because of changes in variants, vaccination, clinical protocols, and ICU referral during the pandemic [12, 13].
The primary hypothesis of this study was that COVID-19-positive ICU patients would have higher in-hospital mortality than COVID-19-negative ICU patients, even after adjustment for available baseline clinical characteristics. Secondary hypotheses were that COVID-19 status would be associated with a distinct pattern of healthcare-associated infection, admission laboratory and gas-exchange findings, and TISS-28-defined therapeutic workload.
Methods
Study design, setting, and reporting framework
This single-center retrospective cohort study was conducted in the adult ICU of the University Clinical Hospital in Białystok, Poland, a tertiary university hospital that functions as a major referral and teaching hospital in north-eastern Poland. The hospital is affiliated with the Medical University of Bialystok and is described by the university as one of the largest and most advanced healthcare operators in north-eastern Poland, with broad specialist services and teaching and research activity [14, 15]. The study is reported in accordance with the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines.
During the study period, critical care services were organized within an anaesthesiology and intensive therapy structure rather than as a pulmonary-led medical ICU. The unit functioned as a mixed high-acuity ICU for medical, surgical, trauma, postoperative, neurological, and infectious-disease patients. The ICU operated 30 beds in the study center, corresponding to approximately 2.6 ICU beds per 100,000 people in Podlaskie Voivodeship if related to the regional population of about 1.14 million residents. This ratio should not be interpreted as the regional ICU supply because other hospitals in the region also provide ICU services; instead, it describes the approximate population-to-bed context of the study center.
The ICU was staffed primarily by physicians trained in anaesthesiology and intensive therapy, including specialists and residents in that specialty. Specialists from other disciplines, including pulmonology, internal medicine, infectious diseases, nephrology, cardiology, neurology, and surgical specialties, were consulted when clinically indicated. During the COVID-19 pandemic, multidisciplinary consultation was particularly relevant for respiratory failure, antimicrobial decisions, renal replacement therapy, cardiovascular instability, and complications requiring surgical or interventional management. The unit did not function as a dedicated pulmonary medical ICU.
At the national level, Polish guidance recognizes the role of specialists in anaesthesiology and intensive care and physicians undergoing specialist training in anaesthesiology and intensive therapy in admission decisions and ICU care [6]. These guidelines define intensive therapy as the use of advanced, highly specialized therapeutic methods by qualified personnel, and they recommend APACHE II, SAPS II, and SOFA as advisory prognostic scales rather than as sole admission criteria [6].
No separate high-dependency unit (HDU) dataset was available in the institutional ICU database for the study period. Patients requiring invasive organ support or continuous ICU-level monitoring were admitted to the ICU; patients requiring lower-intensity monitoring were managed in specialty wards according to local clinical pathways. Therefore, HDU availability, step-up or step-down HDU utilization, and HDU outcomes before, during, and after the pandemic could not be analyzed. Similarly, planned ICU nurse staffing followed local and national intensive care standards and included nurses with formal specialist training in anaesthesiology and intensive care, but shift-level nurse-to-patient ratios before, during, and after pandemic waves were not captured in the retrospective dataset. This limitation was added because nurse staffing is a key determinant of safe critical care delivery.
The ICU routinely provided invasive mechanical ventilation, non-invasive respiratory support, high-flow nasal cannula or continuous positive airway pressure (CPAP) support, artificial airway care, respiratory physiotherapy, arterial and central venous catheterization, vasoactive drug infusion, renal replacement therapy, dialysis catheter insertion, invasive and minimally invasive hemodynamic monitoring, cardiopulmonary resuscitation, parenteral and enteral nutrition, and selected neurological monitoring including intracranial pressure monitoring when clinically required. These interventions were captured through TISS-28 documentation [16].
Study population
The analytic cohort included 3326 patients admitted to the adult ICU between January 1, 2017, and June 1, 2023. For patients with more than one ICU admission, only the first ICU admission was included, and the unit of analysis was the ICU hospitalization. Overall, 355 patients (10.7%) were classified as COVID-19-positive, and 2971 patients (89.3%) were classified as COVID-19-negative. Figure 1 presents the participant selection flowchart.
Fig. 1.
Flowchart of patient screening and inclusion. ICU, intensive care unit; RT-PCR, reverse transcription-polymerase chain reaction; SARS-CoV-2, severe acute respiratory syndrome coronavirus 2
The comparison across 2017–2023 was selected to contrast ICU patients with confirmed SARS-CoV-2 infection against the broader non-COVID-19 ICU population treated in the same institution. This design increases contextual interpretability but also introduces heterogeneity because the study period covers pre-pandemic admissions, pandemic waves, changes in ICU referral pathways, evolving SARS-CoV-2 variants, vaccination rollout, and changes in COVID-19 treatment. Calendar-period adjustment was not possible because precise admission dates and variant or vaccination data were not available in the anonymized analytic extract. The results were therefore interpreted as associations rather than causal effects.
Definitions
COVID-19-positive status was defined as confirmed acute SARS-CoV-2 infection by RT-PCR testing of nasopharyngeal and oropharyngeal swabs or lower respiratory tract specimens in a patient requiring ICU admission. COVID-19-negative status indicated absence of confirmed COVID-19 in the institutional dataset.
In-hospital mortality was defined as death recorded during the index hospitalization. The variable “time to death” was analyzed among non-survivors only because discharge-to-final-disposition dates for all survivors were not available in the dataset. ICU length of stay was calculated as the number of ICU days during the index ICU hospitalization.
Ventilator-associated pneumonia (VAP) was defined as pneumonia occurring after at least 48 h of invasive mechanical ventilation, with compatible clinical features, worsening respiratory status or radiological findings, and microbiological and/or clinical documentation supporting a lower respiratory infection. Pneumonia present before or at the time of intubation was not classified as VAP. This operational definition was aligned with the definitions used in our previous ICU COVID-19 infection studies and with device-associated infection surveillance principles [7, 8, 17].
Bloodstream infection (BSI) was defined as growth of a recognized bacterial or fungal pathogen in at least one blood culture in a patient with compatible clinical signs, when the organism was not considered a contaminant. For common commensals, two or more blood cultures collected on separate occasions were required. BSI was interpreted in relation to other infection sites to distinguish primary and secondary events where possible [9, 17].
Catheter-associated urinary tract infection (CAUTI) was defined as symptomatic urinary tract infection in a patient with an indwelling urinary catheter in place for more than two consecutive days on the date of event or removed the previous day, with compatible clinical signs and a urine culture meeting surveillance criteria. Asymptomatic bacteriuria was not counted as CAUTI [10, 17]. The available all-cohort extract included urinary tract infection status; detailed CAUTI subphenotyping was not available for all patients.
Data collection and laboratory procedures
Patient-level data were extracted from the institutional database into Microsoft Excel by a study investigator. The dataset was reviewed, cleaned, and coded before statistical analysis. Baseline variables included age, sex, height, weight, body mass index (BMI), selected comorbidities, ICU length of stay, and in-hospital mortality. Infection variables included bacterial BSI, urinary tract infection, and VAP. Workload variables were derived from TISS-28 entries and analyzed both as duration variables, representing the number of ICU days on which an intervention was recorded, and as binary variables, indicating whether the intervention was recorded at least once during ICU stay.
All diagnostic testing was performed in a single laboratory at the Medical University of Bialystok, Poland, to reduce technical variability. SARS-CoV-2 detection used RT-PCR assays, including the Allplex 2019-nCoV Assay (Seegene Inc., Seoul, South Korea) and the COVID-19 Real Time Multiplex RT-PCR Kit (Labsystems Diagnostics Oy, Vantaa, Finland), which were cross-validated by the laboratory. Detailed interpretation of RT-PCR results for SARS-CoV-2 diagnostics in Poland has been described previously [18]. Procalcitonin was measured by immunoassay, interleukin-6 by enzyme-linked immunosorbent assay, C-reactive protein by turbidimetry, and complete blood count parameters by automated hematology analysis.
Severity scores and unavailable variables
The reviewers requested ICU severity scores such as APACHE II, SOFA, or SAPS II. These scores were not routinely available in the anonymized all-cohort extract used for the present COVID-19 versus non-COVID-19 comparison. APACHE II data were available in a previously published COVID-19-only subset, but they were not available for the full 3326-patient cohort and therefore could not be included in the main comparative models without substantial loss of comparability [7]. This is a major limitation. The dataset also did not contain reliable variables describing duration of stay in other hospital departments before ICU admission, the specialty responsible for care before ICU admission, shift-level nurse-to-patient ratios, HDU use, or destination after ICU discharge. These requested analyses were therefore not performed and are addressed in the Discussion as limitations.
Statistical analysis
Continuous variables are presented as mean and standard deviation (SD), and categorical variables are presented as number and percentage. Between-group differences were assessed using Welch t-tests for continuous variables and chi-square or Fisher exact tests for categorical variables, as appropriate. Effect sizes were reported as mean differences with 95% confidence intervals (CIs) for continuous variables and odds ratios (ORs) with 95% CIs for categorical variables. P values were reported to support interpretation but were not used as the sole criterion for clinical relevance. Variables with small absolute differences but statistically significant p values were interpreted cautiously.
Multivariable logistic regression was used to estimate the association between COVID-19 status and in-hospital mortality. Covariates were prespecified on clinical grounds and according to availability in the analytic dataset; no automated or univariable p-value-based variable selection was used. The baseline-adjusted model included age, sex, arterial hypertension, diabetes mellitus, recorded obesity, heart and/or respiratory failure, acute myocardial infarction or ischemic stroke, and steroid therapy. An exploratory complete-case model additionally included admission C-reactive protein, white blood cell count, procalcitonin, PaO2, PaCO2, creatinine, glucose, and lactate. Continuous predictors were entered as linear terms and categorical predictors as indicator variables. The reference categories were COVID-19-negative status, female sex, and absence of each binary condition or treatment. Patients with missing data for any covariate in a model were excluded from that model, and no imputation was performed. Model discrimination was summarized using the area under the receiver operating characteristic curve (AUC); sensitivity and specificity were calculated using a predicted-probability cut-off of 0.50. A validated admission-date or pandemic-wave variable was not available for calendar-period adjustment. Results are reported as odds ratios (ORs) with 95% confidence intervals (CIs) and two-sided Wald p values. Analyses were performed using jamovi, version 2.3.18 [19]; p < 0.05 was considered statistically significant.
Results
Baseline characteristics, ICU length of stay, and mortality
The study included 3326 ICU patients, of whom 355 (10.7%) were COVID-19-positive and 2971 (89.3%) were COVID-19-negative. The two groups were similar in age and sex distribution. Mean age was 64.6 years (SD 14.3) in the COVID-19-positive group and 63.5 years (SD 16.8) in the COVID-19-negative group (mean difference 1.13 years, 95% CI -0.48 to 2.74). Female patients accounted for 41.1% and 38.2% of patients, respectively (OR 1.13, 95% CI 0.90 to 1.41).
In-hospital mortality was substantially higher among COVID-19-positive patients than among COVID-19-negative patients (63.9% vs. 44.4%; OR 2.22, 95% CI 1.77 to 2.79). The absolute mortality difference was 19.5% points. Mean ICU length of stay was not significantly different (13.8 vs. 15.3 days; mean difference − 1.53 days, 95% CI -3.08 to 0.02; p = 0.052). Among non-survivors, time to death was shorter in the COVID-19-positive group (10.7 vs. 14.3 days; mean difference − 3.63 days, 95% CI -5.07 to -2.18). Detailed baseline data are presented in Table 1.
Table 1.
Baseline characteristics, ICU length of stay, and mortality according to COVID-19 status
| Variable | COVID-negative | COVID-positive | Effect size (95% CI) | p-value |
|---|---|---|---|---|
| Length of ICU stay, days | 15.3 (24.8); n=2927 | 13.8 (12.1); n=355 | -1.53 (-3.08 to 0.02) | 0.052 |
| Time to death among non-survivors, days | 14.3 (20.0); n=1320 | 10.7 (7.4); n=227 | -3.63 (-5.07 to -2.18) | <0.001 |
| Age, years | 63.5 (16.8); n=2910 | 64.6 (14.3); n=355 | 1.13 (-0.48 to 2.74) | 0.168 |
| Height, cm | 167.9 (16.4); n=1988 | 169.4 (12.1); n=50 | 1.54 (-1.97 to 5.05) | 0.383 |
| Weight, kg | 80.1 (21.3); n=1982 | 85.0 (16.9); n=52 | 4.88 (0.09 to 9.67) | 0.046 |
| Body mass index, kg/m2 | 28.1 (7.1); n=1967 | 29.6 (4.6); n=50 | 1.54 (0.20 to 2.88) | 0.025 |
| In-hospital death | 1320/2971 (44.4%) | 227/355 (63.9%) | 2.22 (1.77 to 2.79) | <0.001 |
| Female sex | 1135/2971 (38.2%) | 146/355 (41.1%) | 1.13 (0.90 to 1.41) | 0.285 |
| Male sex | 1836/2971 (61.8%) | 209/355 (58.9%) | 0.88 (0.71 to 1.11) | 0.285 |
Values are presented as mean (SD); n for continuous variables or n/N (%) for categorical variables. Effect sizes are mean differences for continuous variables and odds ratios for categorical variables. For time to death, the denominator includes non-survivors only. ICU, intensive care unit; CI, confidence interval; OR, odds ratio; SD, standard deviation
Comorbidities and clinical conditions
Several cardiometabolic comorbidities were more common in COVID-19-positive patients, including arterial hypertension (53.5% vs. 34.8%; OR 2.15, 95% CI 1.73 to 2.69), diabetes mellitus (22.0% vs. 16.6%; OR 1.42, 95% CI 1.08 to 1.86), and gout (3.1% vs. 1.0%; OR 3.03, 95% CI 1.51 to 6.09). In contrast, heart and/or respiratory failure, acute myocardial infarction or ischemic stroke, and steroid therapy were recorded more often in the COVID-19-negative group. Recorded obesity and BMI-defined obesity were less frequent among COVID-19-positive patients; this finding should be interpreted cautiously because obesity was likely under-recorded and BMI was missing in many patients, particularly in the COVID-19-positive group. Comorbidity results are shown in Table 2.
Table 2.
Comorbidities and clinical conditions according to COVID-19 status
| Variable | COVID-negative | COVID-positive | Effect size (95% CI) | p-value |
|---|---|---|---|---|
| Arterial hypertension | 1035/2971 (34.8%) | 190/355 (53.5%) | 2.15 (1.73 to 2.69) | < 0.001 |
| Asthma | 57/2971 (1.9%) | 12/355 (3.4%) | 1.79 (0.95 to 3.37) | 0.068 |
| Diabetes mellitus | 492/2971 (16.6%) | 78/355 (22.0%) | 1.42 (1.08 to 1.86) | 0.011 |
| Gout | 31/2971 (1.0%) | 11/355 (3.1%) | 3.03 (1.51 to 6.09) | 0.004 |
| Ischemic heart disease | 177/2971 (6.0%) | 25/355 (7.0%) | 1.20 (0.77 to 1.85) | 0.419 |
| Atrial fibrillation | 365/2971 (12.3%) | 39/355 (11.0%) | 0.88 (0.62 to 1.25) | 0.479 |
| Renal failure | 194/2971 (6.5%) | 27/355 (7.6%) | 1.18 (0.78 to 1.79) | 0.442 |
| Recorded diagnosis of obesity | 327/2971 (11.0%) | 16/355 (4.5%) | 0.38 (0.23 to 0.64) | < 0.001 |
| BMI-defined obesity | 629/2971 (21.2%) | 26/355 (7.3%) | 0.29 (0.20 to 0.44) | < 0.001 |
| Heart and/or respiratory failure | 1285/2971 (43.3%) | 51/355 (14.4%) | 0.22 (0.16 to 0.30) | < 0.001 |
| Acute myocardial infarction or ischemic stroke | 136/2971 (4.6%) | 3/355 (0.8%) | 0.18 (0.06 to 0.56) | < 0.001 |
| Acute infectious and/or rheumatic disease | 427/2971 (14.4%) | 43/355 (12.1%) | 0.82 (0.59 to 1.15) | 0.248 |
| Steroid therapy | 872/2971 (29.4%) | 25/355 (7.0%) | 0.18 (0.12 to 0.28) | < 0.001 |
| Chronic renal failure | 29/2971 (1.0%) | 1/355 (0.3%) | 0.29 (0.04 to 2.11) | 0.364 |
Values are n/N (%). Effect sizes are odds ratios for COVID-positive versus COVID-negative patients. BMI-defined obesity was calculated from the available BMI variable and is limited by missing BMI values. CI, confidence interval; OR, odds ratio
Healthcare-associated infections
Healthcare-associated infection patterns differed between groups. Bacterial BSI was more frequent in COVID-19-positive patients (22.8% vs. 9.9%; OR 2.70, 95% CI 2.05 to 3.56), corresponding to an absolute difference of 13.0% points. Urinary tract infection was also more frequent in the COVID-19-positive group (16.6% vs. 11.3%; OR 1.56, 95% CI 1.15 to 2.11). VAP was numerically more frequent in COVID-19-positive patients (31.8% vs. 27.0%; OR 1.26, 95% CI 0.99 to 1.60), but the 95% CI crossed 1.0. These results are presented in Table 3.
Table 3.
Healthcare-associated infections according to COVID-19 status
| Variable | COVID-negative | COVID-positive | Effect size (95% CI) | p-value |
|---|---|---|---|---|
| Bloodstream infection, bacterial | 293/2971 (9.9%) | 81/355 (22.8%) | 2.70 (2.05 to 3.56) | < 0.001 |
| Urinary tract infection | 337/2971 (11.3%) | 59/355 (16.6%) | 1.56 (1.15 to 2.11) | 0.004 |
| Ventilator-associated pneumonia | 803/2971 (27.0%) | 113/355 (31.8%) | 1.26 (0.99 to 1.60) | 0.056 |
Values are n/N (%). Effect sizes are odds ratios for COVID-positive versus COVID-negative patients. BSI, bloodstream infection; CI, confidence interval; OR, odds ratio; VAP, ventilator-associated pneumonia
Admission laboratory and gas-exchange variables
Admission biomarker and gas-exchange profiles differed by COVID-19 status. COVID-19-positive patients had lower admission C-reactive protein (92.4 vs. 142.1 mg/L; mean difference − 49.73 mg/L, 95% CI -60.99 to -38.47) and procalcitonin (2.4 vs. 7.4 ng/mL; mean difference − 4.96 ng/mL, 95% CI -6.27 to -3.66), despite higher mortality. Admission interleukin-6 was higher in COVID-19-positive patients but with a wide confidence interval crossing zero (495.8 vs. 391.3 pg/mL; mean difference 104.55 pg/mL, 95% CI -45.03 to 254.14).
COVID-19-positive patients had lower admission PaO2 and higher PaCO2, consistent with a respiratory-failure-dominant phenotype. Admission glucose was higher in COVID-19-positive patients (177.2 vs. 147.3 mg/dL; mean difference 29.94 mg/dL, 95% CI 22.65 to 37.23). Several statistically significant electrolyte or creatinine differences were small in magnitude and should not be overinterpreted as clinically decisive. Admission laboratory and gas-exchange variables are shown in Table 4.
Table 4.
Admission laboratory and gas-exchange variables according to COVID-19 status
| Variable | COVID-negative | COVID-positive | Effect size (95% CI) | p-value |
|---|---|---|---|---|
| CRP at admission, mg/L | 142.1 (100.3); n = 1673 | 92.4 (80.1); n = 240 | -49.73 (-60.99 to -38.47) | < 0.001 |
| WBC at admission, x10^3/uL | 14.3 (12.3); n = 2216 | 14.0 (8.7); n = 240 | -0.25 (-1.46 to 0.97) | 0.689 |
| Procalcitonin at admission, ng/mL | 7.4 (14.9); n = 1468 | 2.4 (8.2); n = 232 | -4.96 (-6.27 to -3.66) | < 0.001 |
| Interleukin-6 at admission, pg/mL | 391.3 (659.0); n = 233 | 495.8 (830.4); n = 176 | 104.55 (-45.03 to 254.14) | 0.170 |
| PaO2 at admission, mmHg | 103.6 (28.7); n = 994 | 95.8 (24.2); n = 293 | -7.81 (-11.11 to -4.51) | < 0.001 |
| PaCO2 at admission, mmHg | 41.2 (8.1); n = 995 | 46.0 (10.1); n = 293 | 4.81 (3.55 to 6.07) | < 0.001 |
| pH at admission | 7.4 (0.1); n = 995 | 7.4 (0.1); n = 293 | 0.00 (-0.01 to 0.01) | 0.970 |
| Creatinine at admission, mg/dL | 1.6 (1.2); n = 989 | 1.5 (1.1); n = 292 | -0.18 (-0.32 to -0.03) | 0.020 |
| Glucose at admission, mg/dL | 147.3 (43.3); n = 995 | 177.2 (59.0); n = 293 | 29.94 (22.65 to 37.23) | < 0.001 |
| Sodium at admission, mmol/L | 140.0 (5.7); n = 995 | 141.0 (5.3); n = 293 | 0.98 (0.27 to 1.68) | 0.007 |
| Potassium at admission, mmol/L | 4.2 (0.6); n = 995 | 4.4 (0.6); n = 293 | 0.19 (0.12 to 0.27) | < 0.001 |
| Total hemoglobin at admission, g/dL | 11.1 (2.0); n = 994 | 12.2 (2.0); n = 293 | 1.10 (0.84 to 1.36) | < 0.001 |
| Bicarbonate at admission, mmol/L | 24.6 (5.6); n = 995 | 27.0 (4.9); n = 293 | 2.32 (1.66 to 2.98) | < 0.001 |
| Lactate at admission, mmol/L | 3.3 (4.3); n = 992 | 2.5 (1.9); n = 293 | -0.80 (-1.14 to -0.45) | < 0.001 |
Values are mean (SD); n. Effect sizes are mean differences for COVID-positive minus COVID-negative patients. CI, confidence interval; CRP, C-reactive protein; IL-6, interleukin-6; PaCO2, arterial partial pressure of carbon dioxide; PaO2, arterial partial pressure of oxygen; PCT, procalcitonin; WBC, white blood cell count
TISS-28 intervention burden
TISS-28 data indicated that COVID-19 status was associated with a different distribution of therapeutic workload rather than a uniform increase in all interventions. Respiratory physiotherapy was recorded at least once in nearly all COVID-19-positive patients (97.5% vs. 62.0%; OR 23.63, 95% CI 11.76 to 47.52) and was recorded for a longer mean duration (14.3 vs. 9.5 days; mean difference 4.89 days, 95% CI 3.57 to 6.22). Single vasoactive drug use was also more frequent in COVID-19-positive patients (85.4% vs. 72.2%; OR 2.24, 95% CI 1.60 to 3.13).
By contrast, several surgical, procedural, and transport-related elements were less common in COVID-19-positive patients, including drain care, frequent dressing changes, procedures outside the ICU, massive fluid loss or high fluid administration, and parenteral nutrition. These differences probably reflect case mix and infection-control constraints rather than COVID-19 itself. Binary TISS-28 variables are summarized in Tables 5 and 6, and selected duration variables are summarized in Tables 7 and 8.
Table 5.
TISS-28 monitoring, respiratory, and hemodynamic interventions recorded at least once according to COVID-19 status
| Variable | COVID-negative | COVID-positive | Effect size (95% CI) | p-value |
|---|---|---|---|---|
| Monitoring | 2949/2971 (99.3%) | 355/355 (100.0%) | 5.42 (0.33 to 89.61) | 0.160 |
| Laboratory blood sampling | 2926/2971 (98.5%) | 351/355 (98.9%) | 1.35 (0.48 to 3.77) | 0.566 |
| One drug administration | 19/2971 (0.6%) | 0/355 (0.0%) | 0.21 (0.01 to 3.53) | 0.253 |
| Two-drug administration | 406/2971 (13.7%) | 1/355 (0.3%) | 0.02 (0.00 to 0.13) | < 0.001 |
| Multiple-drug administration | 2938/2971 (98.9%) | 355/355 (100.0%) | 8.11 (0.50 to 132.58) | 0.043 |
| Mechanical ventilation | 2832/2971 (95.3%) | 332/355 (93.5%) | 0.71 (0.45 to 1.12) | 0.136 |
| HFNC or CPAP respiratory support | 644/2971 (21.7%) | 38/355 (10.7%) | 0.43 (0.31 to 0.61) | < 0.001 |
| Respiratory support without mechanical ventilation | 2008/2971 (67.6%) | 122/355 (34.4%) | 0.25 (0.20 to 0.32) | < 0.001 |
| Artificial airway care | 2856/2971 (96.1%) | 329/355 (92.7%) | 0.51 (0.33 to 0.79) | 0.002 |
| Respiratory physiotherapy | 1841/2971 (62.0%) | 346/355 (97.5%) | 23.60 (12.12 to 45.93) | < 0.001 |
| Single vasoactive drug | 2146/2971 (72.2%) | 303/355 (85.4%) | 2.24 (1.65 to 3.04) | < 0.001 |
| Multiple vasoactive drugs | 1896/2971 (63.8%) | 216/355 (60.8%) | 0.88 (0.70 to 1.10) | 0.272 |
| Arterial catheter | 2914/2971 (98.1%) | 354/355 (99.7%) | 6.92 (0.96 to 50.16) | 0.026 |
| Central venous catheter | 2846/2971 (95.8%) | 351/355 (98.9%) | 3.85 (1.42 to 10.49) | 0.004 |
| Cardiopulmonary resuscitation | 316/2971 (10.6%) | 61/355 (17.2%) | 1.74 (1.29 to 2.35) | < 0.001 |
| Calibrated minimally invasive hemodynamic monitoring | 116/2971 (3.9%) | 4/355 (1.1%) | 0.28 (0.10 to 0.76) | 0.008 |
| Non-calibrated/minimally invasive hemodynamic monitoring | 1/2971 (0.0%) | 0/355 (0.0%) | 2.79 (0.11 to 68.50) | 1.000 |
Values are n/N (%). Effect sizes are odds ratios for COVID-positive versus COVID-negative patients. CI, confidence interval; CPAP, continuous positive airway pressure; HFNC, high-flow nasal cannula; TISS-28, Therapeutic Intervention Scoring System-28
Table 6.
TISS-28 device, renal/metabolic, nutrition, and procedural interventions recorded at least once according to COVID-19 status
| Variable | COVID-negative | COVID-positive | Effect size (95% CI) | p-value |
|---|---|---|---|---|
| Standard dressing care | 2575/2971 (86.7%) | 350/355 (98.6%) | 10.77 (4.42 to 26.19) | < 0.001 |
| Frequent dressing changes | 235/2971 (7.9%) | 7/355 (2.0%) | 0.23 (0.11 to 0.50) | < 0.001 |
| Drain care | 1091/2971 (36.7%) | 48/355 (13.5%) | 0.27 (0.20 to 0.37) | < 0.001 |
| Massive fluid loss or high fluid administration | 916/2971 (30.8%) | 37/355 (10.4%) | 0.26 (0.18 to 0.37) | < 0.001 |
| Pulmonary artery catheter | 9/2971 (0.3%) | 0/355 (0.0%) | 0.44 (0.03 to 7.55) | 0.610 |
| Renal replacement therapy | 655/2971 (22.0%) | 78/355 (22.0%) | 1.00 (0.76 to 1.30) | 0.974 |
| Urine output measurement | 2915/2971 (98.1%) | 354/355 (99.7%) | 6.80 (0.94 to 49.28) | 0.028 |
| Forced diuresis | 2598/2971 (87.4%) | 313/355 (88.2%) | 1.07 (0.76 to 1.50) | 0.697 |
| Dialysis catheter insertion | 474/2971 (16.0%) | 77/355 (21.7%) | 1.46 (1.11 to 1.91) | 0.006 |
| Treatment of acidosis/alkalosis | 866/2971 (29.1%) | 58/355 (16.3%) | 0.47 (0.35 to 0.64) | < 0.001 |
| Parenteral nutrition | 1680/2971 (56.5%) | 157/355 (44.2%) | 0.61 (0.49 to 0.76) | < 0.001 |
| Enteral nutrition | 2131/2971 (71.7%) | 308/355 (86.8%) | 2.58 (1.88 to 3.55) | < 0.001 |
| Single ICU intervention | 1482/2971 (49.9%) | 179/355 (50.4%) | 1.02 (0.82 to 1.27) | 0.847 |
| Multiple ICU interventions | 365/2971 (12.3%) | 20/355 (5.6%) | 0.43 (0.27 to 0.68) | < 0.001 |
| Procedures outside the ICU | 1276/2971 (42.9%) | 68/355 (19.2%) | 0.31 (0.24 to 0.41) | < 0.001 |
| Intracranial pressure monitoring | 84/2971 (2.8%) | 4/355 (1.1%) | 0.39 (0.14 to 1.07) | 0.059 |
Values are n/N (%). Effect sizes are odds ratios for COVID-positive versus COVID-negative patients. CI, confidence interval; ICU, intensive care unit; ICP, intracranial pressure; TISS-28, Therapeutic Intervention Scoring System-28
Table 7.
Selected TISS-28 respiratory and hemodynamic intervention duration variables according to COVID-19 status
| Variable | COVID-negative | COVID-positive | Effect size (95% CI) | p-value |
|---|---|---|---|---|
| Mechanical ventilation | 13.3 (16.1); n = 2959 | 12.4 (10.2); n = 355 | -0.86 (-2.08 to 0.35) | 0.163 |
| HFNC or CPAP respiratory support | 1.4 (4.8); n = 2959 | 0.5 (1.7); n = 355 | -0.92 (-1.17 to -0.68) | < 0.001 |
| Respiratory support without mechanical ventilation | 7.8 (15.2); n = 2959 | 1.8 (3.4); n = 355 | -5.99 (-6.64 to -5.33) | < 0.001 |
| Artificial airway care | 15.5 (19.0); n = 2959 | 12.9 (11.6); n = 355 | -2.64 (-4.03 to -1.25) | < 0.001 |
| Respiratory physiotherapy | 9.5 (14.4); n = 2959 | 14.3 (11.6); n = 355 | 4.89 (3.57 to 6.22) | < 0.001 |
| Single vasoactive drug | 7.2 (10.4); n = 2959 | 8.2 (7.7); n = 355 | 0.98 (0.09 to 1.87) | 0.030 |
| Multiple vasoactive drugs | 5.2 (9.1); n = 2959 | 3.8 (5.2); n = 355 | -1.44 (-2.08 to -0.81) | < 0.001 |
| Massive fluid loss or high fluid administration | 1.6 (5.0); n = 2959 | 0.4 (1.8); n = 355 | -1.19 (-1.45 to -0.94) | < 0.001 |
| Arterial catheter | 16.2 (18.2); n = 2959 | 14.6 (11.4); n = 355 | -1.61 (-2.97 to -0.25) | 0.020 |
| Central venous catheter | 16.2 (18.6); n = 2959 | 14.6 (11.5); n = 355 | -1.56 (-2.93 to -0.18) | 0.027 |
| Cardiopulmonary resuscitation | 0.1 (0.4); n = 2959 | 0.2 (0.4); n = 355 | 0.07 (0.02 to 0.12) | 0.004 |
Values are mean (SD); n is the number of patients with available TISS-28 duration data. Duration variables represent the number of ICU days on which the intervention was recorded. Effect sizes are mean differences for COVID-positive minus COVID-negative patients. CI, confidence interval; CPAP, continuous positive airway pressure; HFNC, high-flow nasal cannula; ICU, intensive care unit; TISS-28, Therapeutic Intervention Scoring System-28
Table 8.
Selected TISS-28 renal/metabolic, nutrition, and procedural intervention duration variables according to COVID-19 status
| Variable | COVID-negative | COVID-positive | Effect size (95% CI) | p-value |
|---|---|---|---|---|
| Renal replacement therapy | 2.4 (7.0); n = 2959 | 1.7 (4.9); n = 355 | -0.61 (-1.17 to -0.04) | 0.036 |
| Urine output measurement | 16.5 (18.9); n = 2959 | 14.7 (11.5); n = 355 | -1.85 (-3.23 to -0.47) | 0.009 |
| Forced diuresis | 12.7 (16.9); n = 2959 | 9.7 (9.7); n = 355 | -3.00 (-4.17 to -1.82) | < 0.001 |
| Dialysis catheter insertion | 0.2 (0.5); n = 2959 | 0.3 (0.5); n = 355 | 0.06 (0.00 to 0.12) | 0.043 |
| Treatment of acidosis/alkalosis | 0.6 (1.2); n = 2959 | 0.2 (0.7); n = 355 | -0.34 (-0.43 to -0.26) | < 0.001 |
| Parenteral nutrition | 7.8 (14.5); n = 2959 | 2.9 (5.5); n = 355 | -4.85 (-5.63 to -4.08) | < 0.001 |
| Enteral nutrition | 11.5 (16.5); n = 2959 | 10.7 (10.4); n = 355 | -0.79 (-2.02 to 0.45) | 0.210 |
| Single ICU intervention | 1.1 (1.7); n = 2959 | 1.0 (1.3); n = 355 | -0.16 (-0.31 to -0.00) | 0.046 |
| Multiple ICU interventions | 0.1 (0.4); n = 2959 | 0.1 (0.3); n = 355 | -0.08 (-0.11 to -0.04) | < 0.001 |
| Procedures outside the ICU | 0.9 (1.6); n = 2959 | 0.3 (0.9); n = 355 | -0.53 (-0.64 to -0.41) | < 0.001 |
| Intracranial pressure monitoring | 0.0 (0.2); n = 2959 | 0.0 (0.1); n = 355 | -0.02 (-0.03 to -0.01) | 0.004 |
Values are mean (SD); n is the number of patients with available TISS-28 duration data. Duration variables represent the number of ICU days on which the intervention was recorded. Effect sizes are mean differences for COVID-positive minus COVID-negative patients. CI, confidence interval; ICU, intensive care unit; ICP, intracranial pressure; TISS-28, Therapeutic Intervention Scoring System-28
Multivariable mortality models
In the unadjusted logistic regression model, COVID-19-positive status was associated with higher odds of in-hospital death (OR 2.22, 95% CI 1.77 to 2.79; p < 0.001). After adjustment for age, sex, arterial hypertension, diabetes mellitus, recorded obesity, heart and/or respiratory failure, acute myocardial infarction or ischemic stroke, and steroid therapy, the association remained (adjusted OR 2.03, 95% CI 1.60 to 2.58; p < 0.001). In the exploratory complete-case model that additionally included admission CRP, WBC, procalcitonin, PaO2, PaCO2, creatinine, glucose, and lactate, the adjusted OR was 2.73 (95% CI 1.60 to 4.64; p < 0.001). In the baseline-adjusted model, age, heart and/or respiratory failure, acute myocardial infarction or ischemic stroke, and steroid therapy had statistically significant coefficients. In the exploratory complete-case model, age, WBC, PaCO2, creatinine, glucose, and lactate had statistically significant coefficients. These covariate estimates describe conditional associations and should not be interpreted as causal effects, particularly for steroid therapy because treatment timing was not consistently distinguished. The complete-case model included 672 patients and 311 deaths; because of the substantial reduction in the analysis population, it should be interpreted as a sensitivity analysis that remains vulnerable to selection bias and residual confounding. Full coefficient estimates and descriptive model-performance measures are shown in Table 9.
Table 9.
Logistic regression models for in-hospital mortality: model performance and odds-ratio estimates
| Variable | Unadjusted model |
Baseline-adjusted model |
Baseline + admission markers model |
|---|---|---|---|
| Model characteristics | |||
| Analysis population, n/deaths | 3326/1547 | 3265/1540 | 672/311 |
| Specificity (cut-off 0.50) | 0.928 | 0.677 | 0.803 |
| Sensitivity (cut-off 0.50) | 0.147 | 0.494 | 0.662 |
| AUC | 0.537 | 0.629 | 0.811 |
| Predictor estimates, OR (95% CI); p-value | |||
| COVID-19-positive status |
2.218 (1.765–2.787) p < 0.001 |
2.032 (1.599–2.581) p < 0.001 |
2.729 (1.604–4.643) p < 0.001 |
| Age, per 1-year increase | - |
1.020 (1.015–1.025) p < 0.001 |
1.031 (1.017–1.045) p < 0.001 |
| Male sex | - |
0.969 (0.835–1.125) p = 0.680 |
1.051 (0.709–1.559) p = 0.803 |
| Arterial hypertension | - |
1.088 (0.927–1.278) p = 0.300 |
1.336 (0.901–1.982) p = 0.150 |
| Diabetes mellitus | - |
1.179 (0.968–1.436) p = 0.102 |
0.751 (0.455–1.238) p = 0.261 |
| Recorded obesity | - |
0.847 (0.660–1.087) p = 0.192 |
0.894 (0.536–1.491) p = 0.668 |
| Heart and/or respiratory failure | - |
1.189 (1.012–1.398) p = 0.035 |
1.088 (0.671–1.763) p = 0.734 |
| Acute myocardial infarction or ischemic stroke | - |
1.488 (1.036–2.138) p = 0.032 |
1.179 (0.509–2.730) p = 0.701 |
| Steroid therapy | - |
0.664 (0.562–0.784) p < 0.001 |
1.075 (0.685–1.689) p = 0.753 |
| Admission CRP, per 1 mg/L | - | - |
1.001 (0.999–1.003) p = 0.314 |
| Admission WBC, per 1 × 10^3/uL | - | - |
1.037 (1.011–1.065) p = 0.006 |
| Admission procalcitonin, per 1 ng/mL | - | - |
1.011 (0.996–1.026) p = 0.145 |
| Admission PaO2, per 1 mmHg | - | - |
1.006 (0.999–1.014) p = 0.099 |
| Admission PaCO2, per 1 mmHg | - | - |
1.048 (1.022–1.074) p < 0.001 |
| Admission creatinine, per 1 mg/dL | - | - |
1.286 (1.055–1.568) p = 0.013 |
| Admission glucose, per 1 mg/dL | - | - |
1.005 (1.000-1.009) p = 0.037 |
| Admission lactate, per 1 mmol/L | - | - |
1.453 (1.275–1.656) p < 0.001 |
Outcome was in-hospital death (event = yes). COVID-19-negative status, female sex, and absence of each binary condition or treatment were reference categories. Odds ratios for continuous predictors are expressed per 1-unit increase. The unadjusted model included COVID-19 status only. The baseline-adjusted model additionally included age, sex, arterial hypertension, diabetes mellitus, recorded obesity, heart and/or respiratory failure, acute myocardial infarction or ischemic stroke, and steroid therapy. The baseline plus admission markers model included all baseline-adjusted variables plus admission CRP, WBC, procalcitonin, PaO2, PaCO2, creatinine, glucose, and lactate and was restricted to complete cases. Specificity and sensitivity were calculated at a predicted-probability cut-off of 0.50. Performance measures are descriptive and are not directly comparable across all models because the complete-case model used a smaller analysis population. AUC, area under the receiver operating characteristic curve; CI, confidence interval; CRP, C-reactive protein; OR, odds ratio; PaCO2, arterial partial pressure of carbon dioxide; PaO2, arterial partial pressure of oxygen; WBC, white blood cell count
Additional requested analyses not performed
The reviewers requested additional analyses of the duration of stay in other hospital departments before ICU admission, the specialties responsible for pre-ICU care, and the destinations after ICU discharge. These variables were not reliably available in the anonymized dataset used for this study. The manuscript therefore does not report these analyses. The absence of these variables is included as a limitation because pre-ICU trajectory and post-ICU destination may influence mortality, length of stay, infection risk, and workload interpretation.
Discussion
Principal findings
In this single-center retrospective cohort of 3326 critically ill patients, COVID-19-positive status was associated with higher in-hospital mortality, shorter time to death among non-survivors, a higher frequency of bacterial BSI, a respiratory-failure-dominant admission profile, and a distinct TISS-28 workload pattern. The association between COVID-19 status and mortality remained after adjustment for available baseline characteristics and in an exploratory complete-case model including admission physiology and laboratory markers. These findings should be interpreted as associations, not as proof that COVID-19 status alone caused the observed differences.
Clinical interpretation and effect size versus statistical significance
The mortality difference between groups was both statistically and clinically relevant: COVID-19-positive patients had an absolute mortality difference of 19.5% points and approximately twofold higher adjusted odds of in-hospital death. In contrast, several statistically significant admission laboratory differences, such as sodium, potassium, creatinine, or bicarbonate, were comparatively small and should be interpreted as descriptive markers of group differences rather than as clinically decisive predictors.
The lower admission CRP, procalcitonin, and lactate values in the COVID-19-positive group, despite higher mortality, suggest that conventional inflammatory or hypoperfusion markers may have different meaning in viral respiratory failure than in heterogeneous non-COVID-19 critical illness. One hypothesis is that many COVID-19-positive patients were admitted primarily because of severe respiratory failure, endothelial injury, immune dysregulation, and evolving ARDS, rather than early bacterial sepsis or distributive shock. This interpretation is supported by lower PaO2, higher PaCO2, and higher respiratory workload in the COVID-19-positive cohort, but it remains a hypothesis because the study lacks detailed severity scores, ventilator settings, radiographic data, and longitudinal biomarker trajectories.
Critical care organization, intensivists, and multidisciplinary care
The organization of critical care is central to interpreting these results. The study ICU was a mixed, high-acuity, anesthesiology-led intensive therapy unit, rather than a separate medical ICU led by pulmonary and critical care medicine specialists. In systems where pulmonary and critical care medicine specialists lead MICUs, respiratory-failure expertise, bronchoscopy pathways, pulmonary consultation, and post-ICU respiratory follow-up may be organized differently. In Poland, the anaesthesiology and intensive therapy model places intensivist responsibility primarily within the anaesthesiology and intensive care specialty, with other specialists consulted according to clinical need [6].
This context does not imply superiority of one model over another. Instead, it emphasizes that COVID-19 ICU outcomes and workload depend on staffing structure, specialist availability, admission criteria, surge protocols, and multidisciplinary coordination. Pulmonologists and internists may be particularly important for management before ICU admission and after ICU discharge, whereas anesthesiology and intensive therapy specialists provide continuous ICU-level organ support in the studied unit. This study could not quantify the contribution of each specialty because pre-ICU specialty and consultation data were unavailable.
Nurse staffing and HDUs
ICU nursing capacity is a major determinant of safe critical care, especially during respiratory pandemics [20]. International experience during COVID-19 showed that nurse-to-patient ratios can deteriorate during surges and that staffing may constrain ICU capacity even when physical beds or ventilators are available [5, 21]. In this study, TISS-28 provided intervention-level workload information but not shift-level staffing data. Therefore, the study cannot determine whether nurse-to-patient ratios modified mortality, infection risk, or duration of interventions. This is important because a bed counted as an ICU bed is clinically meaningful only when it is staffed by personnel trained to deliver intensive monitoring and organ support.
The absence of HDU data is another limitation. HDUs or intermediate care units can provide step-up monitoring for deteriorating patients and step-down care for patients no longer requiring full ICU resources. During COVID-19 waves, HDUs may have reduced ICU pressure by managing patients receiving high-flow oxygen, non-invasive ventilation, or close monitoring outside the ICU. Because the institutional ICU dataset did not include HDU utilization or outcomes, the present analysis cannot assess whether HDU pathways influenced ICU admission thresholds, mortality, or length of stay. Future studies should link emergency department, ward, HDU, ICU, and post-ICU data to characterize the whole critical-care pathway.
Healthcare-associated infections and antimicrobial management
The higher frequency of bacterial BSI among COVID-19-positive patients is consistent with international evidence that critically ill COVID-19 patients are vulnerable to ICU-acquired infections. The Eurobact II analysis found different epidemiology of hospital-acquired BSI in COVID-19 and non-COVID-19 critically ill patients, including more frequent respiratory and primary sources in COVID-19 and higher mortality among COVID-19 patients with BSI [22]. A systematic review also reported a high occurrence of BSI in ICU COVID-19 patients [23].
Recent evidence reinforces the importance of secondary infections in severe COVID-19. Stoian et al. reported that critically ill SARS-CoV-2 patients had substantial bacterial and fungal superinfection burden, frequent use of broad-spectrum antibiotic classes, and multidrug-resistant pathogens, and emphasized early screening, accurate diagnosis, and minimizing invasive devices [24]. A related ICU analysis of COVID-19 and Clostridioides difficile coinfection identified antibiotic and proton pump inhibitor exposure as significant risk factors for CDI coinfection and death [25]. These studies support cautious antimicrobial management in critically ill COVID-19 patients, particularly because admission procalcitonin and CRP may not reliably exclude subsequent secondary infection [26].
The present study does not show that COVID-19 directly caused BSI or that any specific antimicrobial strategy improved outcomes. The association may reflect prolonged invasive device use, pandemic infection-control constraints, immune dysregulation, corticosteroid or immunomodulatory exposure, antimicrobial selection pressure, and survival long enough to acquire ICU-associated infections. Therefore, the findings support hypothesis generation and highlight the need for prospective studies with device-days, antibiotic-days, microbiology, severity scores, and time-dependent infection modeling.
TISS-28 workload profile
The TISS-28 results suggest that COVID-19 changed the distribution of ICU workload. Respiratory physiotherapy was nearly universal and lasted longer in COVID-19-positive patients, whereas drain care, frequent dressing changes, procedures outside the ICU, and high-volume fluid replacement were more common in COVID-19-negative patients. This pattern is clinically plausible: COVID-19 ICU care was dominated by respiratory failure, airway management, infection isolation, and bedside organ support, while the non-COVID-19 ICU population included more postoperative, trauma, and procedural cases.
These workload findings should be interpreted with caution. TISS-28 captures recorded interventions, but it may not fully measure time spent on donning and doffing personal protective equipment, prone positioning teams, family communication during visitor restrictions, cleaning protocols, or staff redeployment [27]. Therefore, the observed TISS-28 differences likely underestimate the full organizational burden of pandemic ICU care.
Strengths and limitations
The strengths of this study include a large ICU cohort, comparison of COVID-19-positive and COVID-19-negative patients treated in the same institution, standardized laboratory testing, infection outcomes, admission physiology, and detailed TISS-28 data. The addition of effect sizes and multivariable mortality models improves interpretability compared with p values alone.
The study also has important limitations. First, the retrospective single-center design limits generalizability. Second, the COVID-19 and non-COVID-19 cohorts were heterogeneous, and the study period from 2017 to 2023 included pre-pandemic care, pandemic waves, evolving SARS-CoV-2 variants, vaccination, and changing treatment protocols. Calendar-period adjustment, wave-specific analysis, and vaccination adjustment were not possible because these variables were unavailable in the anonymized extract. Third, full-cohort APACHE II, SOFA, or SAPS II scores were unavailable; therefore, residual confounding by illness severity is likely. Fourth, the expanded physiology/laboratory model was restricted to 672 complete cases, with markedly different complete-case availability between COVID-19-positive and COVID-19-negative patients, creating a risk of selection bias. Its AUC, sensitivity, and specificity are apparent in-sample measures and were not internally or externally validated. Fifth, the timing of steroid therapy relative to ICU admission was not consistently distinguished, so its regression coefficient must not be interpreted as a treatment effect. Sixth, the dataset did not include reliable shift-level nurse-to-patient ratios, HDU utilization, pre-ICU ward duration and specialty, post-ICU destination, vaccination status, SARS-CoV-2 variant, ventilator settings, or detailed antimicrobial exposure. Seventh, TISS-28 variables depended on retrospective documentation and may under-record bedside activity. Finally, multiple comparisons were performed; effect sizes and clinical relevance should therefore be considered alongside p values.
Conclusions
In this single-center retrospective cohort, COVID-19-positive status was associated with higher in-hospital mortality, shorter time to death among non-survivors, more frequent bacterial bloodstream infection, and a distinct ICU workload profile characterized by extensive respiratory physiotherapy and selected supportive interventions. COVID-19-positive patients also showed a different admission physiology and laboratory pattern, with lower CRP, procalcitonin, PaO2, creatinine, and lactate but higher PaCO2 and glucose. Because of residual confounding, missing severity scores, long-period heterogeneity, and unavailable staffing, HDU, pre-ICU, and post-ICU pathway data, these results should be interpreted as associations rather than causal effects or direct practice recommendations.
Acknowledgements
Not applicable.
Abbreviations
- ARDS
Acute respiratory distress syndrome
- AUC
Area under the receiver operating characteristic curve
- BSI
Bloodstream infection
- CAUTI
Catheter-associated urinary tract infection
- CI
Confidence interval
- COVID-19
Coronavirus disease 2019
- CPAP
Continuous positive airway pressure
- CRP
C-reactive protein
- ELISA
Enzyme-linked immunosorbent assay
- HDU
High-dependency unit
- HFNC
High-flow nasal cannula
- ICU
Intensive care unit
- IL-6
Interleukin-6
- OR
Odds ratio
- PaCO2
Arterial partial pressure of carbon dioxide
- PaO2
Arterial partial pressure of oxygen
- PCT
Procalcitonin
- RT-PCR
Reverse transcription-polymerase chain reaction
- SARS-CoV-2
Severe acute respiratory syndrome coronavirus 2
- SD
Standard deviation
- SOFA
Sequential Organ Failure Assessment
- STROBE
Strengthening the Reporting of Observational Studies in Epidemiology
- TISS-28
Therapeutic Intervention Scoring System-28
- UTI
Urinary tract infection
- VAP
Ventilator-associated pneumonia
- WBC
White blood cell count
Author contributions
Conceptualization, M.B. and J.R.Ł.; methodology, M.B. and S.S.; software, M.B.; validation, M.B., S.S., S.L.C., S.F. and A.S.H.; formal analysis, M.B.; investigation, M.B., S.S., S.F. and A.S.H.; resources, M.B., S.L.C., J.R.Ł. and S.F.; data curation, M.B. and S.S.; writing—original draft preparation, M.B.; writing—review and editing, M.B., S.S., S.L.C., J.R.Ł., S.F. and A.S.H.; visualization, M.B.; supervision, S.L.C., J.R.Ł. and S.F.; project administration, M.B. and J.R.Ł. All authors have read and agreed to the published version of the manuscript.
Funding
This study received partial funding from the Medical University of Bialystok. The funder had no role inthe study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Data availability
The datasets used and/or analysed during the current study are not publicly available because they contain patient-level clinical data and are subject to privacy, ethical, and institutional restrictions. De-identified data may be made available from the corresponding author upon reasonable request, subject to approval by the relevant institutional authorities and/or ethics committee, where required.
Declarations
Ethics approval and consent to participate
The study was conducted in accordance with the Declaration of Helsinki. The Bioethics Committee of the Medical University of Bialystok, Poland, confirmed that this retrospective study did not constitute a medical experiment and therefore did not require ethics committee approval (protocol code: APK.002.82.2024; date: 22.02.2024). The requirement for informed consent was waived due to the retrospective nature of the study and the use of routinely collected clinical data.
Consent for publication
Not applicable.
Competing interests
The authors declare no competing interests.
Footnotes
Publisher’s note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
References
- 1.Wu Z, McGoogan JM. Characteristics of and important lessons from the coronavirus disease 2019 (COVID-19) outbreak in China: summary of a report of 72,314 cases from the Chinese Center for Disease Control and Prevention. JAMA. 2020;323(13):1239–42. 10.1001/jama.2020.2648. [DOI] [PubMed] [Google Scholar]
- 2.Tzotzos SJ, Fischer B, Fischer H, Zeitlinger M. Incidence of ARDS and outcomes in hospitalized patients with COVID-19: a global literature survey. Crit Care. 2020;24:516. 10.1186/s13054-020-03240-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Arabi YM, Myatra SN, Lobo SM. Surging ICU during COVID-19 pandemic: an overview. Curr Opin Crit Care. 2022;28(6):638–44. 10.1097/MCC.0000000000001001. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Phua J, Weng L, Ling L, et al. Intensive care management of coronavirus disease 2019 (COVID-19): challenges and recommendations. Lancet Respir Med. 2020;8(5):506–17. 10.1016/S2213-2600(20)30161-2. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Organisation for Economic Co-operation and Development. Health at a Glance 2023: OECD Indicators. Paris: OECD Publishing; 2023. 10.1787/7a7afb35-en. [DOI] [Google Scholar]
- 6.Krzych ŁJ, et al. Guidelines of the Polish Society of Anaesthesiology and Intensive Therapy defining the rules of qualification and criteria for admitting patients to anaesthesiology and intensive care units. Anaesthesiol Intensive Ther. 2022;54(3):219–25. 10.5114/ait.2022.118963. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Bartoszewicz K, Bartoszewicz M, Gradkowski W, et al. Analysis of prognostic factors in critically ill patients with COVID-19. PLoS ONE. 2024;19:e0302248. 10.1371/journal.pone.0302248. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Bartoszewicz M, Kosel J, Nadolny K, et al. Ventilator-associated pneumonia among patients with COVID-19: a retrospective cohort study. Emerg Med Service. 2023;10:210–6. 10.36740/EmeMS202304102. [DOI] [Google Scholar]
- 9.Bartoszewicz M, Czaban SL, Bartoszewicz K, et al. Bacterial bloodstream infection in critically ill patients with COVID-19: a retrospective cohort study. Ther Adv Infect Dis. 2023;10:20499361231207178. 10.1177/20499361231207178. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Dąbrowska P, Bartoszewicz M, Bartoszewicz K, et al. Catheter-associated urinary tract infections in critically ill patients with COVID-19: a retrospective cohort study. Ther Adv Infect Dis. 2024;11:20499361241278218. 10.1177/20499361241278218. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Liu F, Li L, Xu M, et al. Prognostic value of interleukin-6, C-reactive protein, and procalcitonin in patients with COVID-19. J Clin Virol. 2020;127:104370. 10.1016/j.jcv.2020.104370. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Bologheanu R, Maleczek M, Laxar D, et al. Outcomes of non-COVID-19 critically ill patients during the COVID-19 pandemic: a retrospective propensity score-matched analysis. Wien Klin Wochenschr. 2021;133:942–50. 10.1007/s00508-021-01857-4. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Todi S, Ghosh S. A comparative study on the outcomes of mechanically ventilated COVID-19 vs non-COVID-19 patients with acute hypoxemic respiratory failure. Indian J Crit Care Med. 2021;25:1377–81. 10.5005/jp-journals-10071-24009. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.University Clinical Hospital in Białystok. Klinika Anestezjologii i Intensywnej Terapii z Blokiem Operacyjnym [Internet]. Białystok (Poland) https://uskwb.pl/. Accessed 16 Jun 2026.
- 15.Medical University of Bialystok. Medical University of Bialystok Teaching Hospitals [Internet]. Białystok (Poland): Medical University of Bialystok; [cited 2026 Jul 1]. Available from https://www.umb.edu.pl/en/szpitale_kliniczne. Accessed 16 Jun 2026.
- 16.Miranda DR, de Rijk A, Schaufeli W. Simplified Therapeutic Intervention Scoring System: the TISS-28 items - results from a multicenter study. Crit Care Med. 1996;24(1):64–73. 10.1097/00003246-199601000-00012. [DOI] [PubMed] [Google Scholar]
- 17.Centers for Disease Control and Prevention. National Healthcare Safety Network (NHSN) patient safety component manual. Atlanta (GA): Centers for Disease Control and Prevention; 2024. Available from: https://www.cdc.gov/nhsn/. Accessed 16 Jun 2026.
- 18.Gierczyński R, Czerw A, Juszczyk G, et al. Quantitative analysis of RT-PCR test results for SARS-CoV-2 diagnostics across Poland during COVID-19 pandemic: comparison between early stage and major pandemic waves in 2020 and 2021 with reference to SARS-CoV-2 variants. Adv Med Sci. 2022;67(2):386–92. 10.1016/j.advms.2022.09.002. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.The jamovi project. jamovi (Version 2.3.18) [computer software]. Sydney (Australia): The jamovi project; 2022 [cited 2026 Jul 1]. Available from https://www.jamovi.org.
- 20.Rae PJL, Pearce S, Greaves PJ, et al. Outcomes sensitive to critical care nurse staffing levels: a systematic review. Intensive Crit Care Nurs. 2021;67:103110. 10.1016/j.iccn.2021.103110 [DOI] [PubMed] [Google Scholar]
- 21.Jeitziner MM, Moser A, Wendel-Garcia PD, et al. Critical care staffing ratio and outcome of COVID-19 patients requiring intensive care unit admission during the first pandemic wave: a retrospective analysis across Switzerland from the RISC-19-ICU observational cohort. Swiss Med Wkly. 2022;152:w30183. 10.4414/smw.2022.w30183. [DOI] [PubMed] [Google Scholar]
- 22.Buetti N, Tabah A, Loiodice A, et al. Different epidemiology of bloodstream infections in COVID-19 compared to non-COVID-19 critically ill patients: a descriptive analysis of the Eurobact II study. Crit Care. 2022;26:319. 10.1186/s13054-022-04166-y. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Ippolito M, Simone B, Filisina C, et al. Bloodstream infections in hospitalized patients with COVID-19: a systematic review and meta-analysis. Microorganisms. 2021;9:2016. 10.3390/microorganisms9102016. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Stoian M, Azamfirei L, Văcăraș R, et al. Incidence and risk factors of secondary infections in critically ill SARS-CoV-2 patients: a retrospective study in an intensive care unit. Biomedicines. 2025;13(6):1333. 10.3390/biomedicines13061333. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Stoian M, Azamfirei L, Văcăraș R, et al. COVID-19 and Clostridioides difficile coinfection analysis in the intensive care unit. Antibiotics. 2024;13(4):367. 10.3390/antibiotics13040367. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Williams P, McWilliams C, Soomro K, et al. The dynamics of procalcitonin in COVID-19 patients admitted to intensive care unit - a multi-centre cohort study in the South West of England, UK. J Infect. 2021;82:e24–6. 10.1016/j.jinf.2021.03.011. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Bruyneel A, Lucchini A, Hoogendoorn M. Impact of COVID-19 on nursing workload as measured with the Nursing Activities Score in intensive care. Intensive Crit Care Nurs. 2022;69:103170. 10.1016/j.iccn.2021.103170. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Data Availability Statement
The datasets used and/or analysed during the current study are not publicly available because they contain patient-level clinical data and are subject to privacy, ethical, and institutional restrictions. De-identified data may be made available from the corresponding author upon reasonable request, subject to approval by the relevant institutional authorities and/or ethics committee, where required.