In this qualitative analysis, a deductive-inductive framework approach was used. The applied framework was developed by Liu et al [15] aiming at categorizing online patient complaints into a PCC perspective. The development was guided by the best fit framework synthesis technique [16] and tailored accepted PCC-framework models to the data source of online patient complaints. According to Coulter [17], PCC is a form of care that meets and responds to patients’ wants, needs, and preferences, and is prevalent where patients are autonomous and able to decide for themselves. The main dimensions affecting PCC are: (1) respect for patient values, preferences, and expressed needs; (2) coordination and integration of care; (3) information and education; (4) physical comfort; (5) emotional support and alleviation of fear and anxiety; (6) involvement of family and friends; (7) continuity and transition; and (8) access to care [15,18]. Grounded by this concept, Liu et al [15] integrated the dimensions of PCC into the Donabedian structure-process-outcome model containing the following four constructs: (1) prerequisites, (2) the care environment, (3) patient-centered processes, and (4) expected outcomes [19]. In a second step, the taxonomy was tested by assigning themes derived from the quantitatively selected patient online complaints data into the framework. Figure 1 illustrates the applied framework.

The sample of online posted patient reports was guided by a theoretical selection process. Reports on hospital stays posted on the most commonly used German hospital rating website Klinikbewertungen (Clinic Reviews) [20] were included if written between March 2020 and September 2021. This time period represents the phase in which initial hygiene protection standards (eg, mandatory masks, distance regulations, test obligation) were implemented and maintained across German rehabilitation clinics [21]. The selected hospital rating website is a third-party, patient-centered commercial platform that operates independently of governmental entities. Whether hospitals encourage patients to rate their hospital stay on this platform cannot be answered with certainty.

Given that this was an exploratory study, patient reports were included regardless of their medical indications. To contrast results, reports of IRS were included if referred rehabilitation clinics were located in the federal states of Brandenburg and Saarland, as these states demonstrated the highest and lowest decrease of applications for IRS provided by the Federal German Pension Fund, respectively (Saarland=58.3%, Brandenburg=23.9%) [12]. The federal state of Saarland is located in southwest Germany with a population density of 186 citizens/km² [22]. Brandenburg is located in northeast Germany with a density of 85 residents/km² [22]. Despite infrastructural differences, differences in IRS application rates may imply different coping strategies of resident patients of the respective states or may further reflect different political strategies at the federal policy level.

In this study, open-access online patient reports were used. Therefore, no ethical approval was required. However, we carefully anonymized all cited reports in the manuscript to avoid a linkage of patients’ user names of the hospital rating websites with referenced citations.

Data on patient reports were extracted using a web-scraping technique based on the computing package “Rvest” of the R Project for Statistical Computing [23]. The web-scraping code of included data is provided in Multimedia Appendix 1. Scraped data were transferred to Microsoft Excel (Redmond, USA) and imported into the qualitative analysis software MAXQDA (Berlin, Germany).

Within this domain, attributes of the patient-centered professional emerged to be of major significance for patients utilizing the referred hospital rating website. As a main result, patients felt a decreased sensitivity and a lack of empathy in interpersonal interactions between themselves and the medical staff. They perceived some physicians and therapists to exploit the naturally prevalent hierarchy among them and interpreted this patronizing human interaction as an expression of an imminent gratification crisis.

Among reports, a discontinuous communication of curfew legislation supported a perceived sense of disempowerment. This was further endorsed by hygiene rules, which were rated to be arbitrary as they noticeably differed across rehabilitation clinics and impacted the perceived autonomy of a relevant number of patients. Complaints addressing a decrease of autonomy were particularly prevalent in psychiatric facilities. However, a majority of patients rated existing hygiene rules to be appropriate.

Referring to care continuity, intersectoral care was not always maintained. Patients criticized a lack of involvement of their family physician or their psychologist in charge. Apparently, this was reflected by the fact that diagnostic reports and treatment plans of ambulatory care were frequently not taken into account during IRS. Conversely, results of IRS were not transferred to the ambulatory health care practitioner. Moreover, patients reported to have limited access to structured ambulatory follow-up rehabilitation programs as responsible social workers were hard to reach. Despite stated constraints, established clinician-patient relationships had the potential to mitigate adverse effects of the pandemic on IRS as patients valued empathetic, personal contact.

Availability of therapeutic and nursing care was mainly attributed to the care environment. Patients reported having limited access to therapeutic and nursing procedures. In some cases, this limited availability of care led to a termination of the inpatient rehabilitation stay.

Despite availability issues, it became apparent that hygiene legislations were more likely to be accepted if they were easily integrated into organizational routines. This was also seen as having the advantage to create a more familiar environment as, for instance, therapeutic care groups decreased in size. One factor not directly attributable to the pandemic was the available food, which was perceived to be inconsistent with nutrition education events offered during rehabilitation.

In general, the domain of expected outcomes was of minor significance for patients in German rehabilitation clinics. However, it became apparent that a distinct communication of patient-relevant outcomes and their respective change after rehabilitation was positively associated with patient satisfaction. The communication of changes in outcomes seemed to be more straightforward to be implemented in somatic care facilities. Furthermore, available emotional support during IRS was perceived to facilitate the individual healing process by having a direct impact on activating self-efficacy and self-management potential.

The domain of “peer relationship” inductively emerged during the process of analysis. Empowering peer-to-peer relationships was valued to have the potential to mitigate adverse effects of the pandemic on IRS. Thus, some patients reported that their stay remains unforgotten mainly due to their peers, who compensated for negative inconveniences. Patients also appealed to the personal responsibility of their peers. In their understanding, only active engagement allows expectations of rehabilitation success.

A comprehensive summary of online reported patient experiences addressing PCC domains and attributes is provided in Table 3.

For patients receiving IRS, aspects of the care environment, staff prerequisites, and patient-centered processes were predominantly relevant to evaluate their inpatient stay. SARS-CoV-2–related adaptation mechanisms affecting these domains comprised discontinuous communication and elaborate implementation of infection protection standards, which were perceived to threaten the personal autonomy of action. These mechanisms were amplified by tangible gratification crises of medical staff. However, the prevalence of established and emotional supportive relationships to clinicians and peer groups provided the potential to mitigate the adverse effects of hygiene protection standards on IRS. Moreover, a distinct communication of therapeutic outcome variation seemed to support patient satisfaction. These insights provide the opportunity to develop informed strategies fostering resilient organizations that sustainably embody PCC within the setting of rehabilitative care.

Our findings are partly in line with those of Liu et al [15] who demonstrated country-specific differences in patient complaint behaviors. Although British and Canadian reviewers tend to complain about staff prerequisites, Germans are more likely to criticize the care environment and patient-centered processes [25]. Considering the context of inpatient rehabilitation clinics, Sander et al [26] conducted an initial examination of web-based patient reports to investigate determinants associated with recommending inpatient rehabilitation clinics, and identified perceived therapy successes as well as process of care parameters to be associated with clinic recommendations. Although aspects of patient-centered processes and expected outcomes were quantitatively subordinate, the qualitative in-depth analysis of the present study indicates a relationship between positively reported expected outcomes and patient satisfaction. This is also in line with Kraska et al [27], who demonstrated outcome quality to be a predictor for patient satisfaction during hospital stays in Germany.

Considering the suitability of the applied PCC framework [15], this analysis revealed that the inductively originated domain of “peer relationships” has been of relevance for inpatient rehabilitation programs. As the overall accuracy of the applied taxonomy to the setting of PCC in inpatient rehabilitation was rated high, the “peer relationships” domain may be a meaningful extension of the taxonomy for settings in which patients have the opportunity to interact with their peers over a longer period of time.

In this analysis, patients perceived a significant number of medical staff to present aspects of a developing gratification crisis reflecting generic psychological distress. This finding is supported by Dobson et al [28], who identified health care workers to face moderate levels of depression and anxiety. This observation is particularly prevalent for the nursing profession, which experienced high levels of burnout and emotional exhaustion during the SARS-CoV-2 pandemic [29,30]. To facilitate the resilience of health systems for future pandemic events, it will be of interest to meet health care providers’ needs not only to foster employee health but also to support quality of care. As psychological distress of medical staff was perceived to be an amplifying factor for reduced patient autonomy, it is relevant to refer to the scoping review of Klemmt et al [31] supporting the influence of medical staff on patient autonomy, while further emphasizing domains of the rehabilitation system, the rehabilitation facility, and patients themselves to have a bidirectional influence on autonomy. Following their conclusion, it is important to be aware that IRS not only aims to foster social inclusion as a summative outcome but should also be requested for structures and stakeholders during rehabilitation.

Despite illustrated challenges of IRS during the early phase of the SARS-CoV-2 pandemic, most patients felt safe and supported infection protection standards. This is underpinned by a survey of oncological patients treated in German rehabilitation clinics, 87% of whom reported to feel safe in facilities [32]. Although the implementation of infection-protection standards was associated with a tangible workload increase, 84% of staff members assisted the implementation [32].

First, one limitation of our study is the limited representativeness of findings for other rehabilitation settings within Germany as hygiene regulations differed across states. Moreover, online reported patient complaints as a scientific data source produce concerns of representativeness and subjectivity as sample characteristics are uncontrolled and widely unknown. However, an analysis of Facebook reviews demonstrated that contents of reviews do not correlate with inpatient quality assessment indicators but instead correlate with a standardized national survey of patient experiences in German obstetrics [33]. Moreover, a Dutch investigation of online patient ratings identified a positive correlation of these ratings with evaluation reports of the Dutch Healthcare Inspectorate referring to underperforming, high-risk hospitals [34]. Demonstrating initial representativeness concerns of online ratings, Dutch health care inspectors valued online ratings as an additional source of information after being confronted with negative ratings and emphasized to cautiously interpret them under referral to standardized quality and safety indicators [35].

Second, using online reported patient complaints is accompanied by unknown sample characteristics and thereby associated with hazards of selection bias. In this context, Han and colleagues [36] identified prognostic factors of patient characteristics associated with patient intentions and behaviors on physician rating websites. In their survey study, they identified health-related variables (seeking physician information online, usage of web-based medical consultation services, prevalence of a serious disease, good medical experiences) to be directed to the active rating behavior. Conversely, cognitive variables (altruism, self-efficacy to perform online ratings, trust in online ratings of peers) affected the rating intention. These results may help to further understand the patient population using this feedback opportunity.

Moreover, research activities of economic sciences identified online product ratings to be influenced by social dynamics. It is acknowledged that product ratings are not only affected by individual experiences but rather by prior ratings of one’s peer group [37,38]. At this stage of research, it is reasonable to question to what extent these dynamics equally occur on physician and hospital rating websites.

Along with these stated limitations, this analysis faces unique restrictions. As the distribution of reports across included clinics varied strongly, a cluster bias of included rehabilitation clinics with disproportionately strong patient rating activities of some facilities cannot fully be ruled out. In this regard, it will be of interest to further investigate which clinic-related parameters affected ratings of clinics with above-average report numbers. Additionally, the current selection of keywords reflecting PCC domains was made inductively and potentially implies an incomplete list of keywords supporting a distortion of distributed domains.

Despite these limitations, patient rating portals became increasingly popular over the last 10 years [8,39], which suggests that patients claim these portals as a trusted source of information. Beyond a growing number of physicians acknowledging patient reports for in-house quality improvement initiatives [40], by systematically investigating online reported patient reviews, this analysis provides the potential to integrate patient perspectives into the discussion on how to maintain PCC structures under the stress and strains of a pandemic. Integrating online-reported patient complaints offers the opportunity to extend scientific data sources by providing the advantage to reduce the social desirability bias of common qualitative research formats, as this analysis demonstrates that patients perceive a hospital rating website to be a protective platform supporting the exchange of individual experiences.

Taking the present results into account, future research direction should investigate country-specific differences in the perceived significance of PCC domains. For instance, it remains to be answered why online reports of German inpatient care recipients are currently dominated by reports about the care environment, whereas health-relevant outcome expectations seem to have a subordinate role.

This analysis reflects previous research as German patients predominantly reported feedback associated with the care environment. SARS-CoV-2–related reports were strongly affected by aspects of patient-centered processes addressing discontinuous communication and an organizationally demanding implementation of infection protection standards, which was in some cases perceived to threaten patient autonomy. This perceived threat in reduced autonomy was amplified by a tangible increase in staff workload. Developed peer relationships during the rehabilitation stay had the potential to mitigate these mechanisms.