Introduction
The first wave of the COVID-19 pandemic, spreading globally in the spring of 2020, affected all areas of the health sector. While disciplines directly managing patients with proven SARS-CoV‑2 infection, such as pneumonology or intensive care, were hardest hit initially, the multidisciplinary management of cancer patients was also immediately affected.
In particular, the risk of limited availability of intensive care units led to a call from the German Ministry of Health to delay all non-urgent surgeries in all hospitals to free up intensive care capacity on March 12, 2020. Whereas cancer treatment is not considered elective or non-urgent, national and international expert groups devised recommendations on the interdisciplinary management of cancer patients under conditions of limited resources [
1].
A common pattern of these recommendations was the consideration to avoid major surgery in cancer patients who would be expected to require postoperative intensive care and instead prefer non-surgical treatments including radiotherapy [
2]. Leading German radiation oncology institutions published recommendations on the management of radiation oncology departments during the COVID-19 pandemic, including strategies to handle shortages of staff and the continuation of radiotherapy in patients with suspected or proven SARS-CoV‑2 infection [
3]. In addition, these recommendations mention hypofractionated schedules that would result in fewer administered fractions during the pandemic [
4].
One comprehensive study from Great Britain revealed a considerable decrease in the number of radiotherapy courses across all entities in 2020 when compared to 2019 [
5]. It is unclear to what extent the use of radiotherapy in cancer treatment was affected during the first wave of the COVID-19 pandemic in Germany. Recently published survey results from German comprehensive cancer centers (CCCs) indicated that radiotherapy availability was never affected during the first wave as opposed to, e.g., diagnostic imaging, systemic therapy, or cancer surgery [
6]. A questionnaire survey among radiation oncology institutions in Germany, Austria, and Switzerland performed in April and May 2020 documented a decrease in the number of patients treated with radiotherapy, which was unrelated to the incidence of SARS-CoV-2-positive cases or the type of radiation oncology institution [
7]. The survey respondents reported a tendency to change the fractionation mostly in palliative radiotherapy concepts and to postpone radiotherapy in curative indications.
The German Medical Informatics Initiative (MII) [
8], comprising the four consortia DIFUTURE [
9], HiGHmed [
10], MIRACUM [
11], and SMITH [
12], previously established an infrastructure to support federated analyses across the participating German university hospitals [
13]. Based on this infrastructure, we are now able to analyze possible changes in radiotherapeutic inpatient settings caused by the COVID pandemic.
Our analysis focuses on cervical cancer (CC) and head and neck cancer (HNC), as in both entities, primary radiochemotherapy is an alternative option to primary surgical treatment according to stage and risk factors [
14‐
16]. In addition, as many institutions administer radiochemotherapy in an inpatient setting, both entities might well serve as surrogates to analyze COVID-19 effects in an inpatient collective. In CC, radical hysterectomy is the mainstay of curative treatment for early stages in Germany, but radiotherapy is stated as an alternative in the S3 guidelines [
17]. Furthermore, in many situations of HNC, including laryngeal cancer requiring laryngectomy, radical radiotherapy or chemoradiation are organ-conserving alternatives to radical surgery [
18].
Thus, additional use of radiotherapy in the management of CC or HNC could become apparent via increased inpatient delivery of radiotherapy or chemoradiation, with a simultaneous reduction in the number of cases treated with surgery. A large proportion of institutions administer radiotherapy, especially brachytherapy or chemoradiation, in an inpatient setting, which is, however, not the case for the treatment with radiotherapy alone. Based on this reasoning, we aim to assess the effect of the lockdown on fractions and admissions in an inpatient setting in radiotherapy institutions.
CC and HNC are two cancer types that are typically treated by definitive chemoradiotherapy and in which no delays of therapy can be accepted, with respect to the rapid proliferation of these tumors. Concomitant radiotherapy and chemotherapy are usually performed by the radiation oncologist. Therefore, treatment of these two diseases is less dependent on biases caused by multidisciplinary treatment compared to many other tumor entities, supporting the choice of these cancer types with relatively robust rates of incidence, diagnosis, and treatment within radiation oncology departments.
Discussion
In this retrospective study, we analyzed changes in the therapeutic management of inpatients with CC and HNC across 14 German university hospitals following the lockdown announcement on March 16, 2020, in Germany [
21]. A significant decrease in performed radiotherapy fractions for malignant neoplasms of the cervix uteri was observed across all participating sites in 2020 in the 20 weeks following the lockdown announcement on March 16, 2020, in Germany to August 2, 2020, compared to the average of the two previous years. This effect was especially driven by a reduction of Megavoltage radiation therapy (OPS 8‑522, 8‑523), whereas no differences were observed for brachytherapy (OPS 8‑524, 8‑525). Notably, even though on April 28, 2020, the German Federal Ministry of Health announced the gradual reactivation of hospital capacity for elective treatments from May onwards [
20], the observed reduction was even more pronounced when analyzing the time period from May 4, 2020, to August 2, 2020, in more detail. These numbers are in accordance with our analysis of related hospital admissions.
The observed decrease in case numbers for CC with associated radiotherapy goes hand in hand with little, non-significant changes in hospitalizations treated surgically. In light of the restricted operating theater and intensive care capacity, the unlikely finding of a decrease is surprising but may be explained by a high priority of surgical cancer treatment at a time of reduced capacity for elective interventions.
The delayed decrease of radiotherapy-related hospitalizations might be a consequence of a curtailed oncological screening for CC relation to the pandemic. This reasoning would entail the presumption of a lag between the initial diagnosis and initiation of treatment of several weeks. As an alternative interpretation, these findings might be a consequence of intended treatment postponements. However, few radiation oncology institutes in Germany reported a postponement of treatment as a consequence of the pandemic [
7]. From an outcome perspective, there exist no valid data on the effects of a delayed treatment in CC [
24].
In contrast to other countries [
25], there was no official suspension of CC screening during the corona lockdown in Germany. However, out of fear and because of the call to reduce contacts, women might have abstained from screening during this period. Such a reduced willingness might have detrimental effects on patients and treatment success [
26]. This is in line with data based on German practices where, among other disciplines, gynecology practices showed a strong reduction in case numbers by 21.7 to 30.8% between March and May 2020, [
27]. Another analysis based on the same data source as our study but addressing inpatient admissions in general, found that the decline in case numbers started immediately after introduction of the lockdown restrictions [
13].
In contrast, for HNC, a significant increase in performed radiotherapeutic fractions was observed in our cohort in 2020 in the 20 weeks following the lockdown announcement on March 16, 2020, as well as in the 13-week period from May 4, 2020 onwards, in comparison with the average of the two previous years. In our analysis of related inpatient hospital admissions, an increase could be observed for admissions in which radiotherapeutic procedures were performed, whereas no differences could be observed for radiotherapeutic admissions with additional chemotherapy.
The increase in case numbers and fractions found in HNC was accompanied by a numerical decrease in cases with surgery, which was, however, not statistically significant. Here, respective changes in radiotherapy occurred after initiation of the lockdown measures with a delay of 1 to 2 weeks only. This delay is explainable by radiotherapy planning prior to hospitalization. The shift in hospitalized cases might reflect a preference for non-surgical treatments during the lockdown. Such a reasoning might especially apply to head and neck cancer, where surgery is complex and imposes significant COVID-related risks for the surgical team [
3,
26]. One guideline recommended such a temporary shift from surgery to radiotherapy during the onset of the pandemic [
1]. In contrast to our findings, Spencer et al. found no relevant change in the number of courses and attendances in their study for HNC in the already mentioned British data [
5]. However, due to the centralized and unified character of the British health care system in the form of the NHS, measures and guidelines might have been introduced more coherently and stringently. In addition, the British study encompassed in- and outpatient data, with only a few centers failing to provide data.
Apart from this shift, other factors might have contributed to this finding: university hospitals with their large capacities might have received more patients in the aftermath of the first wave of the pandemic; diagnoses might be delayed, resulting in more advanced cases with different treatment approaches. In addition, the proportion of cases treated in an in- or outpatient setting might have changed. This applies to CC (outpatient treatment preferred during/after the lockdown) and HNC (inpatient treatment preferred during/after the lockdown). Finally, alternating chemotherapy regimens might have contributed to a change in admissions and fractions administered during hospitalization.
In a survey performed among radiation oncologists in Germany, Austria, and Switzerland, most of the radiation oncology institutes (ROIs) reported no change in curative or palliative treatment [
7]. Fractionation schedules were changed in 25.7% (curative radiotherapy) and 42.1% (palliative radiotherapy) of the ROIs, while the general postponement of treatment played virtually no role. The authors also found that non-university ROIs were more willing to change their treatment pattern. This might well apply to our setting, which addressed only university institutions. The decrease in case numbers in the survey was independent of the regional incidence of COVID-19 and the type of institute (university vs. non-university).
Limitations
The major limitation of the present analysis lies in the selective consideration of inpatients. However, radiotherapy might have shifted from an in- to an outpatient setting in the wake of the lockdown. We tried to mitigate this effect by focusing on entities with a strong inpatient component of treatment such as the regular use of concomitant radiochemotherapy. In addition, as the lockdown restrictions specifically targeted the inpatient setting while sparing outpatient cancer treatment and screening, a shift from in- to outpatient treatment would still in part be detectable in the hospital setting.
Regarding the observational unit of cases, confounded results might occur if shorter but repeated hospitalizations became the preferred pattern during the lockdown. However, such an alteration appears unlikely, as it would contradict the lockdown restrictions calling for reduced hospitalizations [
19]. Thus, for assessment of temporal changes in radiotherapy, fractions are a more reliable endpoint than cases.
Furthermore, we have no detailed information on fractionation or dose concepts. In order to shorten treatments, hypofractionated or even ultra-hypofractionated radiotherapy might have become a frequently applied regime. An increased use of such ultra-hypofractionated concepts was especially striking in British data and the treatment of breast cancer [
5]. The German Radiation Oncology Society (
Deutsche Gesellschaft für Radioonkologie, DEGRO) recommended the application of hypofractionated concepts in order to reduce treatment time [
28]. Two sources of alternated fractionation play an important role in HNC. On the one hand there might be a decrease in hyperfractionated concepts and an increase in the frequency of hypofractionation [
18]. However, if such alterations were apparent in our data, we would underestimate the lockdown effect in terms of radiotherapy use, where we observed increased numbers during the lockdown period.
Admission varied considerably between institutions, decreasing the power to detect possible alterations caused by the lockdown. By applying mixed models, we could reduce the statistical variation between considered hospitals and estimate a generalized effect.
Another limitation may be the overlapping of cases between some therapy categories (Supplementary Table S1). For example, for CC, some cases of the category “radiotherapy without surgery, chemotherapy present” may also be included in the group “radiotherapy without surgery without brachytherapy.” However, we chose this approach to look at possible effects from different perspectives by analyzing the subgroups. Furthermore, the therapy category “surgery present” might include both “pure” surgical cases and those with additional radiochemotherapy during the same hospital stay. Although we assume that the latter is rather a minority, future analyses may aim at a stricter and more finely granulated separation between these therapy categories.
Further limitations introduced by the use of the claims dataset were also described in more detail in [
13].
Members of the MII research group
Julien Balig (Institute of Medical Systems Biology, Ulm University, Ulm, Germany); Jonas Bienzeisler (Institute of Medical Informatics, Medical Faculty, RWTH Aachen University, Aachen, Germany); Daniel Buergy (Department of Radiation Oncology, University Medical Centre Mannheim, Medical Faculty Mannheim, Heidelberg University, Heidelberg, Germany); Patrick Fischer (Institute of Medical Informatics Faculty of Medicine, Justus-Liebig-University Giessen, Giessen, Germany); Jonas Fortmann (Institute of Medical Informatics, Medical Faculty, RWTH Aachen University, Aachen, Germany); Timo Fuchs (Department of Nuclear Medicine, University Hospital Regensburg, Regensburg, Germany); Thomas Ganslandt (Department of Biomedical Informatics, University Medicine Mannheim, Mannheim, Germany); Matthias Gietzelt (Peter L. Reichertz Institute for Medical Informatics of TU Braunschweig and Hannover Medical School, Hannover Medical School, Hannover, Germany); Christian Haverkamp (Institute of Digitalization in Medicine, Medical Faculty and Medical Center, University of Freiburg, Freiburg, Germany); Kurt Marquardt (University Hospital of Giessen and Marburg, Marburg, Germany); Dennis Kadioglu (Medical Informatics Group, University Hospital Frankfurt, Frankfurt am Main, Germany); Irina Lutz (Data Integration Center Aachen, University Hospital RWTH Aachen, Aachen, Germany); Gerhard Mayer (Institute of Medical Systems Biology, Ulm University, Ulm, Germany); Achim Michel-Backofen (Institute of Medical Informatics, Data Integration Center, Faculty of Medicine, Justus-Liebig-University Giessen, Giessen, Germany); Harald Renz (Institute of Laboratory Medicine, member of the German Center for Lung Research (DZL), Universities of Giessen and Marburg Lung Center (UGMLC), Philipps University Marburg, Marburg, Germany); Christian Seidemann (University Hospital of Giessen and Marburg, Marburg, Germany); Holger Stenzhorn (Institute of Medical Biometry, Epidemiology and Medical Informatics, Saarland University Medical Center, Homburg, Germany); Ana Stolnicu (Institute of Medical Systems Biology, Ulm University, Ulm, Germany); Holger Storf (Medical Informatics Group, University Hospital Frankfurt, Frankfurt am Main, Germany); Gaetan Kamdje Wabo (Department of Biomedical Informatics, University Medicine Mannheim, Mannheim, Germany); Jochen Zohner (Institute of Medical Informatics, Data Integration Center, Faculty of Medicine, Justus-Liebig-University Giessen, Giessen, Germany)