Data sources

Since November 4, 2020, positive SARS-CoV-2 patients and controls without infection from all age groups have been recruited into the SUEP cohort in NAPKON at 57 locations in various areas of the healthcare system throughout Germany. These locations include university hospitals, non-university hospitals and general practitioners’ practices. Among demographic and clinical data, the SUEP database also records whether patients received in-hospital IER.

Study design

IER utilization data was extracted from the SUEP cohort of hospitalized cases and stratified for prior intensive care treatment in a first step. For clinical analyses, a SUEP database export of the cases with an existing 3-month-follow-up (3MFU) was retrieved. Patients were included in this analysis if they were at least 18 years old, hospitalized for COVID-19 and had completed at least the 3MFU of the NAPKON-SUEP project. A descriptive analysis of this sample was carried out. Patients were characterized by age, height and weight, clinical frailty scale [26], and comorbidities. Cases with missing data in the 3MFU were not included in the data export.

Statistical analysis was performed using SPSS (version 28, IBM, Chicago, Illinois). Mean values and standard deviations were calculated for metric items as part of the descriptive analysis and to provide a better overview. Absolute frequencies and percentages were given for categorical and dichotomous variables. Proportions of utilization rates were calculated as the number of patients receiving the service divided by the total sample size. Corresponding 95% confidence intervals (CIs) were computed using the exact method by Clopper-Pearson based on the binomial distribution to ensure accurate interval estimation given the low event counts. Calculations were conducted using R (version 4.5.0, http://www.r-project.org).

Reporting was based on the STROBE guidelines for reporting cohort studies.

Ethics and registration

For the NAPKON-SUEP cohort, a primary ethics vote was obtained from the ethics committee of the department of medicine at Goethe University Frankfurt, Germany (local ethics ID approval 20–924). All other study sites received local ethics votes from the respective ethics committees. Approval for this analysis was granted by the ethics committee of the medical faculty at the Friedrich Schiller University Jena, Germany (reference no. 2021–2493-Daten). The NAPKON-SUEP was registered with ClinicalTrials.gov (identifier: NCT04768998).

IER utilization rates of hospitalized and ICU patients

The analysis of the entire database enabled an examination of IER utilization rates stratified by prior ICU treatment. At the time of data export, 2,644 patients of the entire SUEP cohort had completed acute inpatient treatment, of which 564 (21%) had received intensive care treatment during the course of their illness. IER on a specialized ward, i.e., acute rehabilitation unit (ARU), was documented for a total of n = 21 patients (0.79% [95% CI: 0.51%, 1.22%] of hospitalized cases), and therefore fulfilled the formal IER criteria. N = 24 patients (0.91% [95% CI: 0.60%–1.34%]) were visited by mobile rehabilitation teams, not fulfilling formal IER criteria. Fig 1 shows the detailed flow chart. Of the 564 patients in the SUEP cohort who had ICU treatment, 2.13% [95% CI: 1.16%, 3.63%] received in-hospital IER on an ARU.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Flow chart.

IER utilization data stratified by prior ICU treatment (SUEP: German abbreviation: Cross-Sectoral Platform, IER: in-hospital interdisciplinary early rehabilitation, ICU: intensive care unit).

https://doi.org/10.1371/journal.pone.0334941.g001

Clinical Analysis of ER Sample

The data export with 3MFU data, selected according to the criteria described in the methods section, comprised 634 hospitalized patients with an existing 3MFU. A population of n = 36 (5.67% [95% CI: 4.09%, 7.80%]) patients received ER services, including IER as well as mobile early rehabilitation services (flow chart in Fig 2). They were on average 65.4 (± 16.1) years old and n = 13 (36.1%) stated that they were female.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Flow chart.

Clinical analysis sample with 3MFU (3MFU: 3 Month Follow Up, SUEP: German abbreviation: Cross-Sectoral Platform, ER: early rehabilitation, IER: in-hospital interdisciplinary early rehabilitation).

https://doi.org/10.1371/journal.pone.0334941.g002

When looking at the documented comorbidities of the 36 patients, cardiovascular diseases were predominantly documented (n = 24; 66.7%). This was followed—in descending order of frequency—by chronic neurological or psychiatric diseases (n = 13; 36.1%), chronic lung diseases (n = 10; 27.8%), and diabetes mellitus (n = 7; 19.4%). The characteristics of the sample are shown in Table 2.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Characteristics of the interdisciplinary in-hospital early rehabilitation subcohort (IER and mobile rehabilitation team).

https://doi.org/10.1371/journal.pone.0334941.t002

COVID-19 IER utilization rates compared with other data

This multicenter cohort study provides insight into the frequency of IER in acute hospitals for patients with COVID-19 in the NAPKON-SUEP cohort. The IER utilization rate in the acute hospital was 0.8% across the SUEP cohort and 2.1% among the subgroup of patients who received intensive care treatment during the course of their illness.

Previous scientific analyses on IER in Germany predominantly report clinical outcomes, while utilization rates are not precisely analyzed [6–8]. According to one estimate, the need for early rehabilitation therapy in acute hospitals is generally estimated between 1–3% of all acutely hospitalized patients [1].

In comparison, our observed COVID-19 sample shows a markedly lower utilization rate, suggesting a significant undersupply of IER. This interpretation is further supported by data from sepsis survivors in Germany, where the IER utilization rate is approximately 4.4% [5]. Notably, although this rate is more than five times higher than that observed in our COVID-19 cohort, it is still regarded as insufficient—highlighting an ongoing “unmet need” for IER, even in high-risk populations [5]. Drawing on findings from this interdisciplinary sepsis cohort, it can be hypothesized that early rehabilitation may help reduce long-term mortality—potentially extending up to 10 years—in patients with severe COVID-19 [5]. These comparisons emphasize the importance of incorporating early rehabilitation into acute hospital care for patients with significant functional impairments [5].

ICU patients’ utilization rates

Across all inpatient cases in Germany, the overall ICU treatment rate was approximately 9% in the year 2022 [27]. By comparison, COVID-19 patients had more than double this ICU admission rate (20%) in the year 2020 in Germany, suggesting a greater overall rehabilitation need [28]. In the SUEP cohort, 21% of patients—closely matching the national figure—required intensive care, further underscoring the expected need for ER in this population [28]. Despite the potential limitations in data quality, the analysis supports the following conclusion: the presumed need for ER within the acute hospital setting was not met for many COVID-19 patients—particularly not for those with prior ICU treatment.

Possible barriers to IER utilization

The utilization of IER services in acute care hospitals faced several barriers during the COVID-19 pandemic, some of which can be generalized, whereas others are specific to the German healthcare context. Especially during the pandemic, structural limitations like restricted bed capacity and staff shortages may have hindered the timely initiation of IER. The prioritization of critical cases often led to the reallocation of rehabilitation resources and personnel, reducing accessibility. In some cases, entire units were temporarily closed due to infection control measures. However, as data in the SUEP cohort were collected beyond the acute phase of the pandemic, these pandemic-specific disruptions cannot fully explain the under-utilization. Additional contributing factors may include patient-level factors, such as failure in to identify individuals eligible for IER and a consequent lack of referral. This may reflect knowledge or awareness deficits in clinicians, limitations in training, or inadequate resource availability.

Health system barriers

IER in acute hospitals is increasingly acknowledged as an essential component of high-quality, integrated functioning-oriented patient care. Both the German Medical Association (Bundesärztekammer, BÄK) and the World Health Organization (WHO) advocate for the establishment of structured early rehabilitation services within acute care settings. A recent appeal by the German Medical Council entitled High-performance medicine requires high-performance rehabilitation explicitly calls for the implementation of early rehabilitative care at all acute hospitals treating patients with complex medical needs [29]. While sequential rehabilitation capacities—particularly in neurological and geriatric specialties—are well developed in Germany, there remains a significant structural gap in providing (interdisciplinary) early rehabilitation services directly within acute care hospitals.

The WHO’s 2023 Resolution on Strengthening Rehabilitation in Health Systems formally acknowledges rehabilitation as an essential health service and calls on member states to embed rehabilitation into health system planning at all levels. The resolution highlights the importance of early initiation of rehabilitation interventions to improve patient outcomes and reduce disability. Complementing this, the WHO’s Rehabilitation 2030 initiative provides detailed recommendations for implementing rehabilitation services, emphasizing timely, multidisciplinary, and patient-centered approaches starting in the acute phase. This also includes clear infrastructure development recommendations to implement specialized acute rehabilitation units: “Hospitals should include specialized rehabilitation units for inpatients with complex needs” [30].

These international frameworks underscore the urgent need to expand in-hospital IER capacities in Germany to ensure equitable access and optimize recovery trajectories for all patient populations—not only in the post-pandemic care for COVID-19 patients but also following other serious illnesses resulting in functional impairments [1].

Furthermore, it must be assumed that the documented mobile early rehabilitation teams in some hospitals do not meet the structural or therapeutic standards of dedicated acute rehabilitation units and instead provide services closer to early mobilization [29]. For this reason, such services were excluded from our utilization rate analysis.

ER provided sequentially after initial acute care—such as transfer to specialized neurological or geriatric rehabilitation hospitals or clinics—is likely not captured in the SUEP documentation. If included in an overall assessment, utilization proportions would potentially increase; however, this would not align with the definition of in-hospital IER as analyzed here.

Also, from a regulatory perspective, current reimbursement structures under the German Diagnosis-Related Groups (G-DRG) system may have be a barrier for implementing early rehabilitation services due to perceived financial constraints or inadequate reimbursement.

Taken together, two recurring structural weaknesses in the German early rehabilitation landscape can be identified: (1) the lack of integrated ER services within acute care hospitals, and (2) the limited availability of ER services for non-neurological, non-geriatric populations in the sequential rehabilitation phase following discharge from acute hospital care.

The opportunity presented by the forthcoming healthcare reforms should therefore be leveraged to address these structural deficits and enable the timely, i.e., earliest possible access to ER for patients with complex needs during their initial hospitalization.

IER sample analysis

Among the 36 patients in the SUEP cohort who received IER, a majority of n = 22 patients reported having been very fit, averagely active or still coping well with everyday life prior to the SARS-CoV-2 infection—consistent with the Clinical Frailty Scale (CFS). However, it should be noted that the CFS is only validated for the age group over 64 years [34]. Among the 19 patients aged 65 or over, the distribution of responses was varied, although the most common (n = 6) premorbid functional scale was “managing well”. The scale was frequently used as a triage tool during the COVID-19 pandemic to make informed clinical decisions for further treatment when treatment resources were scarce [31]. The meta-analysis by Pranata et al. (2021) confirmed that an increasing CFS score is linearly associated with higher COVID-19 mortality [32]. Marti-Pastor et al. (2023) also showed that mild to moderate frailty was associated with increased readmission and also served as a predictor independent of disease severity [33,34]. In the cohort examined, pre-existing functional limitations were relatively minor and therefore likely did not negatively impact the intensity of subsequent treatment, potentially leading to lower IER needs and utilization.

Limitations

While the analysis of the SUEP database promises robust data due to its sample size and multicenter design, there are major methodological limitations and analytical constraints. This analysis was only possible in the SUEP cohort of the NAPKON project. However, many patients who were treated at university hospitals, including hospitals with large IER units, were recruited into the high-resolution cohort (HAP), which is another NAPKON cohort. Unfortunately, no item concerning IER was included in the HAP database. For this reason, the actual utilization rate of IER services in hospital may differ from the results presented here, and a lack in generalizability of the results must be acknowledged.

It should also be mentioned that originally a matched-pairs design was planned to compare the quality of life of patients with and without IER at 3MFU and 12MFU. However, after an intensive review of the data export, it turned out that it contained incorrect data sets relating to other rehabilitation services. After the exclusion of data sets with missing data, the adjusted NAPKON export was then too small to analyze and the original comparative design had to be abandoned, which further weakens our study.

A relevant limitation of this analysis is certainly the small sample size, especially the low event count and consequently low utilization rates, which lead to leveraged bias. Therefore, further analyses beyond a descriptive presentation were not expedient.