It is assumed that efficient telemedical access to an interdisciplinary expert forum (the Extremity Board) with simultaneous therapy decisions in the treatment of fractures with open soft tissue damage and postoperative complications will achieve significant reductions in the length of inpatient stay, the total duration of treatment, and the complication and reoperation rate [23]. The measures optimize therapy pathways, standardize them, and make them more cost-effective. The establishment of the Extremity Board leads to higher guideline adherence, standardization, and documentation quality and, consequently, to reduced antibiotic consumption and a reduction in unnecessary diagnostics.

For the ergonomic evaluation of the EXPERT project, the job demands–resources (JD-R) model [24-26] on the influence of work resources and work requirements on work behavior will be used. In this theoretical model, which is backed up by a wide range of findings [27-29], work resources and work requirements are considered to be independent factors influencing work engagement and perceived stress from work. High cognitive, emotional, and physical demands such as time pressure, workload, complexity of tasks, and emotional stress lead to high demands, which has long-term negative consequences for employees and the company (eg, employee burnout, low job satisfaction, and low productivity [30]).

On the other hand, work resources such as social support, feedback, autonomy, and meaningfulness have a motivating effect because they satisfy basic needs for self-efficacy, attachment, security, and control [31,32]. Therefore, diverse work resources lead to more work engagement, which has positive consequences for employees and the company (eg, high job satisfaction and high productivity [30]).

The JD-R model also postulates that work requirements and work resources interact with each other, which affects the perception of stress and motivation. Thus, the availability of resources can mitigate the negative impact of the requirements. Resources seem to have a positive effect on motivation and work engagement, especially when work demands are high [33]. When work demands are high, employees increasingly perceive existing resources as helpful and useful, which helps them cope with the demands. This results in a subjectively lower burden (coping hypothesis [34,35]).

On the basis of the theoretical assumptions and empirical findings, the following hypotheses can be derived for the EXPERT project (Figure 1): (1) work requirements and workload increase in the intervention phase of the EXPERT project (hypothesis 1a); (2) work-related resources such as knowledge, competencies, and team support increase during the intervention phase (hypothesis 1b); and (3) the technical competence of employees increases during the intervention phase (hypothesis 1c).

Above- or below-average resources are recorded with the median of differences (pretest-posttest estimates of resources), leading to other hypotheses: (1) above-average resources are associated with lower workload, and below-average resources are associated with higher workload (hypothesis 2); (2) above-average resources are associated with higher work engagement (hypothesis 3a); (3) when resources increase, work engagement increases (hypothesis 3b); (4) if resources increase (exclusion of technical competence as a resource), then readiness for technology increases while the burden decreases (hypothesis 4a); (5) as technical competence (resource) increases, the burden decreases (hypothesis 5a); (6) when technical competence (resource) decreases, the burden increases (hypothesis 5b); and (7) above-average work engagement leads to higher readiness for technology, and below-average work engagement leads to lower readiness.

For the organization and evaluation of the new form of care, a stepped-wedge design was chosen [36,37]. In a stepped-wedge design, there are different clusters that start recruitment at the same time in the control phase. The individual clusters then move gradually into the intervention phase in randomized cohorts. In the EXPERT project, there is a transition phase between the control and intervention periods, which serves to implement and test the new form of care in the individual clinics as well as train the staff.

Randomization in the EXPERT project will take place at the hospital level. There are 31 hospitals participating in the study. Each clinic forms a cluster that belongs to 1 of 4 cohorts. The clusters within a cohort always enter the next study phase (transition or intervention phase) at the same time. The study will be conducted over a period of 24 months (18-month recruitment period+6-month follow-up of the last enrolled person). Data collection regarding the individual course of treatment takes place from the time of enrollment in the patient sample until 6 months later. Figure 2 shows the planned course of the study in the stepped-wedge design.

For the ergonomic evaluation, the “difference-in-difference” method will be used within the stepped-wedge design [38]. This methodological approach includes a treatment group and a control group, each of which is examined at 2 measurement points (t0 and t1). In the treatment group, but not in the control group, an intervention takes place between the measurement points. The first difference that is measured refers to a possible change from measurement time t0 to measurement time t1. This is calculated for both the treatment and control groups. The second difference is the difference between these 2 differences (ie, the difference within the control group is subtracted from the difference within the treatment group). The result can be interpreted as a causal effect of an intervention [38,39].

In the EXPERT project, the clinics that have not yet implemented the project measures serve as a control group, which will be evaluated at the same time as the clinics that are already implementing the new measures. Cohorts 1 and 2 are the treatment group, whereas cohorts 3 and 4 act as the control group.

Studies show that, in approximately 40% of cases, complications occur during the treatment in open fractures [40]. The new form of care is realistically intended to reduce the complication rate to 30%. A type-1 error (α) of 5%, a type-2 error (β) of 20% (80% power), and an intraclass correlation coefficient of ρ=0.05 are assumed [36]. Considering the transition phase and a dropout rate of 20%, the total number of cases required to demonstrate a treatment effect is 3366 patients (of whom n=1377, 40.91% would be in the control phase; n=1428, 42.42% would be in the intervention phase; and n=561, 16.67% would be in the transition phase). This equates to 17 patients per hospital per 3-month interval.

Only patients who can give consent and who have statutory health insurance will be included in the project and will give their written consent to participate and share their data in the use of telemedicine. The study will include patients with fractures of the extremities with open soft tissue damage (according to Gustilo et al [41]) or with postoperative complications (according to the criteria of the Association of the Study of Internal Fixation Anti-Infection Task Force) or people with unplanned resurgery within 6 months of initial surgery. Specifically, these are International Statistical Classification of Diseases and Related Health Problems, 10th Revision (ICD-10), diagnoses [42] related to open soft tissue damage in fractures or dislocations, amputations of the lower extremity, infections of joints or joint endoprostheses of the lower extremities [43], and pseudarthrosis and delayed fracture healing of the lower extremity (see the list in Textbox 1).

Patients who are not capable of giving consent, who do not give their consent to participate or to data transfer, or who do not have statutory health insurance will be excluded from the study. Also excluded will be persons to whom the aforementioned ICD-10 diagnoses do not apply (ie, who do not have a fracture of the extremities or extremities or have a closed fracture of the extremities without open soft tissue damage and who do not experience a postoperative complication within 6 months of initial surgery). The inclusion and exclusion criteria are summarized in Textbox 2.

Physicians and nurses who are or will be involved with the project take part in the work-related evaluation. With medium to small effect sizes (η²>0.05 and <0.15), a type-1 error (α) of 5%, and a type-2 error (β) of 20% (80% power), a minimum number of 120 people per measurement point is required for the work- and project-related evaluation (60 people each in the control and treatment groups). A total of 4 employees per clinic at measurement time t0 and 4 employees at measurement time t1 are included in the sample. This results in 31 × 8 = 248 elevations. Care is taken to interview physicians and nurses in equal parts, all of whom are or will be involved in the project measures.

Regarding the interviews for the assessment of the objectives and project implementation, 2 persons from the nursing service and 2 persons from the medical profession are to be interviewed at time t1 in the clinics that are already implementing the new form of care. Approximately 72 people from at least 18 clinics will take part in the interviews.

The primary outcome and various secondary clinical outcomes are recorded based on data from the clinical case file as well as routine data from the health insurance companies.

The primary clinical outcome is the complication rate of open fractures or the occurrence of postoperative complications, which is collectively defined as a dichotomous outcome. Either the occurrence or nonoccurrence of complications is observed. The following events are complications if they occur within 6 months of enrollment in the sample: (1) occurrence of pseudarthrosis (incomplete fracture healing; ICD-10 code M96.0); (2) occurrence of chronic osteomyelitis (ICD-10 code M86.69); (3) unplanned follow-up procedures, such as wound revisions and Negative Pressure Wound Therapy (ICD-10 code S81.801); (4) deviation from standardized antibiotic therapy (extension of the duration of therapy or change or expansion of the active substance; ICD-10 code T36.95); (5) occurrence of antibiotic-associated complications (eg, clostridia infection or allergy; ICD-10 code B96.7); (6) infection with resistant germs (eg, methicillin-resistant Staphylococcus aureus) that occurred during inpatient treatment and did not exist at the time of trauma (ICD-10 code A49.02); (7) need for unplanned inpatient treatment; and (8) death of the patient related to the trauma (not due to other causes).

Secondary clinical outcomes include the number of antibiotics administered, limb function, and health-related quality of life (HRQoL) [44,45].

Health economic outcomes are mapped through claims data from health insurance companies. These will be recorded for each patient included in the project up to 6 months after enrollment in the sample. Specifically, the following features will be recorded: (1) use of health services (eg, contact with physicians and therapists, operations, and inpatient stays), (2) cost of health care services (inpatient hospital costs, outpatient costs, rehabilitation costs, medical and medical aid costs, drug costs, and total costs), (3) number of days on which there is incapacity for work, and (4) number of days on which sickness benefits are received.

Table 1 provides an overview of the clinical and health economic outcomes used in this study.

Patients’ HRQoL is assessed at the time of enrollment and 6 months later using two internationally used measuring instruments: (1) the German-language version of the royalty-free Veterans RAND 36-item Health Survey [44] and (2) the German-language version of the EQ-5D of the EuroQol Group [45,46].

The Veterans RAND 36-item Health Survey is a generic profile questionnaire and measures HRQoL regarding the dimensions “Physical Functioning,” “Role-Physical,” “Role-Emotional,” “General Health,” “Social Functioning,” “Vitality,” “Bodily Pain,” and “Mental Health.”

The EQ-5D as a generic index instrument contains the 5 dimensions “Mobility,” “Self-care,” “Usual activities,” “Pain/Discomfort,” and “Anxiety/depression” as well as a visual analog scale. The results of the EQ-5D can be used to create a health profile as well as calculate an index value [46]. As an instrument validated in Germany, the index for HRQoL recorded using the EQ-5D is particularly well suited as a health economic outcome measure [46].

The eligible patients are enrolled in the project by trained staff. Patients are informed about the objectives and course of the project and give their written consent to participate. Each case is summarized with all relevant information (eg, anamnesis, findings, and imaging) in a clinical case file using data-secure web-based software. The case files are checked for completeness and, after approval, transmitted digitally to the interdisciplinary expert group, the Extremity Board. This group of experts is recruited from the following disciplines: trauma surgery, plastic surgery, vascular surgery, radiology, angiology, and antibiotic stewardship (microbiology, hygiene, infectiology, and pharmacy) [43]. The experts meet at least weekly via web conference and, depending on the urgency, discuss the cases together. Agreed interdisciplinary therapy recommendations are written and transmitted back to the presenting clinic via the web-based software platform. In addition, guidelines will be defined in the course of the project on the basis of the findings gained by the Board to objectify decision-making paths, streamline processes, define therapy algorithms, and enable the supraregional applicability of these standards.

The clinical case files of all patients included in the EXPERT project will be merged with the respective patient-related health insurance data in an anonymized data file for each case while maintaining data protection. Claims data from health insurance companies provide information about the use of health services and the costs that arise in doing so. To obtain information about the individual course of treatment after the introduction of the new form of care, claims data for the following 6 months are included in the evaluation for each patient from the time of enrollment in the sample. Thus, the clinical case file and the associated health insurance data depict the course of treatment of a patient, which is included in the evaluation.

To be able to evaluate the extent to which the new care measures improve the chances of recovery and reduce costs, various clinical and economic outcomes are collected.

First, the primary outcome of the complication rate as well as the secondary and health economic outcomes are evaluated descriptively using absolute and relative frequency distributions as well as position and scattering measures (eg, median, variances, SD, and representation by means of box plots or histograms). The differences between patients in the control and intervention phases are checked using appropriate statistical methods (eg, Mann-Whitney U test [60,61], chi-square test, or odds ratio [62]). In addition, individual subgroups (eg, by sex and age) are analyzed to check the data for possible biases.

This descriptive analysis is followed by interference statistical analyses. By means of suitable regression analyses (eg, using generalized linear regression models), it can be examined which regressors have an influence on the primary and secondary outcomes. In addition to the consideration of temporal and intervention-related effects, the influence of potential confounders is also controlled for. This is because various patient- and trauma-associated factors are relevant risk factors for the development of a complication in the treatment of open fractures. For example, male sex, age of >60 years, BMI of >40 kg/m², use of nonsteroidal anti-inflammatory drugs, diabetes mellitus, and nicotine consumption are positively correlated with the development of an infection or fracture healing disorder [63]. This also applies to fractures in the context of multiple injuries (polytrauma) or in the context of agricultural accidents [64,65]. The quality of the statistical models is tested with the help of a residual analysis. Sensitivity analyses are carried out to check the stability of the evaluation and the assumptions made, for instance, the effects of changed influencing variables on the result.

As part of the health economic evaluation, the use as well as the costs of health services and absences from work are considered. For this purpose, a cost-effectiveness analysis is carried out whereby the incremental cost-effectiveness ratio will be calculated. The evaluation of routine statutory health insurance data is carried out according to the principles of good secondary data analysis [66], the recommendations of the “Methods for Health Services Research” memorandum [67], and the standards of the German Evaluation Society [68]. The data basis for the evaluation comprises the statutory health insurance routine data, the data from the case file, and the collected data on the health economic quality of life.

The original item scaling of all 4 or 5 surveys respectively was retained, and all items were compiled into 1 questionnaire. In total, 4 versions of this questionnaire were created, which are identical in content but have been—with relation to the instructions—slightly adapted to the group (treatment group vs control group) and the time of measurement (before vs after the implementation of the new measures). For example, the instruction before treatment reads “Dear participant,...the clinic where you work...will soon start implementing the new measures” (treatment group) and “...will start implementing the new measures in the foreseeable future” (control group), whereas the instruction in the treatment group after implementing the new measures is “Dear participant,...the clinic where you work...is already implementing the new measures.” The questionnaire was implemented, and postcards with a brief explanation and a QR code with a link to each specific questionnaire version were prepared. Interview guidelines and protocol sheets for the semistructured interview were created, and several fact sheets were written to inform the employees about the interview and data protection aspects.

In the practical implementation of ergonomic data collection, the project staff at Steinbeis University are supported by project staff from University Hospital Münster (Universitätsklinikum Münster [UKM]). Project staff from UKM have contact with the clinics and recruit people there to take part in the questionnaire surveys and interviews.

Before the clinics of cohort 1 start implementing the new form of care in project month 7 (Figure 2), both the clinic staff of cohort 1 (treatment group) and the clinic staff of cohort 3 (control group) will be examined with the EXPERT questionnaire in project months 4 and 5 (measurement time Pre). In project months 12 and 13, clinic staff from both cohorts will repeat the questionnaire (measurement time Post). The same procedure will be followed for cohort 2 and cohort 4, 3 and 9 months later, respectively. Project staff from UKM will recruit enough people in the individual clinics for the surveys. Interested employees in the clinics who are or will be involved in the project can find out more about the survey in advance by means of an information text. The UKM project staff receive a questionnaire link as a QR code from the Steinbeis project staff for the corresponding measurement time tailored to the corresponding group and cohort and send this QR code to a contact person in the clinics. The contact persons pass on the QR code to the participants, who scan it on their mobile phones and answer the questions. Participants are requested to do so as soon as possible after receiving the link. The completed questionnaires are automatically stored anonymously on the Steinbeis University server and later evaluated in summary form.

The UKM staff draw attention to the project-related interviews in the clinics that are already implementing the measures. Interested employees can find out more about the interviews by means of an information text and can contact the Steinbeis project staff to make an appointment for the telephone interview. They then receive an email from Steinbeis with the confirmation of the appointment, the interview questions, and the (renewed) promise that their email address and telephone number will be deleted after completion of the project. The transcripts of the telephone interviews are anonymized using a code. The content collected in the interviews will be evaluated in summary form, quantitatively, and qualitatively.

Regarding the topics of workload, work engagement, work-related resources, and readiness for technology, the main effects and interactions between measurement time (before vs after start of the Extremity Board) and group of measures (treatment vs control) are examined via variance analysis; the result is a 2 × 2 design with the factors time (before vs after) and group (treatment vs control group).

Ergonomic aspects of the telemedicine technology are collected and descriptively evaluated at the second measurement point in approximately half of the clinics.

As staff turnover in hospitals is high, it can be assumed that the groups of employees will be composed differently at each time of measurement. Therefore, no repeated-measure design can be applied. The analyses will be performed as balanced grouping factors with the same number of participants in all conditions (before vs after and treatment vs control). The measurements are considered as a grouping factor of a multiple ANOVA.

The evaluation of the repertory grids yields 2D representations of the negative and positive constructs mentioned by the interviewees during the interviews with respect to the different implementation steps, namely, preparation, development of new standards, interdisciplinary collaboration, and therapy implementation.

The combination of a questionnaire that can be used quickly and easily on a mobile phone and a telephone interview with both predetermined, scaled questions and open-ended evaluations according to the repertory grid technique results in a comprehensive picture of changes in pretest-posttest working conditions and of the success factors accompanying the process both in quantitative and qualitative terms. On the one hand, the web-based questionnaire before and after the introduction of the new form of care quantitatively captures influences on workload, work engagement, work-related resources, and readiness for technology and backs them up through the “difference-in-difference” design. On the other hand, central success factors regarding project implementation will be collected both quantitatively and qualitatively and made visible by means of the repertory grid technology as “mental maps” of the employees in the 31 participating clinics.