Change of Functioning in Persons With Severe Musculoskeletal Injuries Over Time: A Multilevel Analysis


 Background: The World Health Organization (WHO) recommends the WHO Disability Assessment Schedule (WHODAS) 2.0 as a generic assessment instrument to collect data on functioning and disability. The questionnaire was developed specifically to capture the activities and participation domain as defined by the International Classification of Functioning, Disability and Health (ICF). Evidence on the most relevant factors predicting WHODAS 2.0 outcome in the context of musculoskeletal injuries are controversial. This study aims to assess change in functioning of patients with severe musculoskeletal injuries undergoing inpatient rehabilitation over time.Methods: A longitudinal multicentre study was conducted, following up 571 participants with severe musculoskeletal injuries over the course of a first inpatient rehabilitation stay until 3 months after discharge. At admission, data on sociodemographic, health-related aspects, functioning and contextual factors were collected. WHODAS 2.0 assessed functioning. Data were analysed using a multilevel model approach.Results: The mean WHODAS 2.0 declined from admission to discharge and 3-month follow-up, indicating an improvement in functioning. Multilevel analyses revealed age, duration of inpatient rehabilitation, severity of the injury, injury localizations, number of comorbidities, emotional functioning, pain, being informed about the injury, subjective prognosis on return to work and agreement on treatment targets as factors influencing change in functioning over time.Conclusions: In a rehabilitation setting, a healthcare professional can promote an increase in functioning, for example, by ensuring that there are treatment targets defined and agreed on with the patient and that the patient feels sufficiently informed about the injury. The identified factors could potentially be used for a short screening at admission to rehabilitation to estimate the patient’s change of functioning over time. Trial registration number and date of registration: DRKS00014857; July 04, 2018.

problems in functioning and contextual factors of persons with musculoskeletal injuries to the components of the classi cation mentioned beforehand.
To know about (long-term) consequences of musculoskeletal injuries and associations with functioning and disability is essential for an optimal rehabilitation treatment and management (6, 13). Particularly patient-reported outcomes play an important role in assessing functioning and health (15)(16)(17), as well as in evaluating the outcome quality of rehabilitative treatment (17,18). WHO's recommended generic assessment instrument to collect data on functioning and disability is the WHO Disability Assessment Schedule (WHODAS) 2.0 (19). The questionnaire has been developed speci cally to capture the activities and participation domain as de ned by the ICF and covers six domains of life (cognition, mobility, self-care, getting along, life activities, participation).
Thus, it remains unclear in which activities and participation domains patients are limited and restricted and whether or how these limitations and restrictions change over time.
The objective of the study is therefore to assess the change in functioning of patients with severe musculoskeletal injuries undergoing inpatient rehabilitation over time using the WHODAS 2.0 and by applying multilevel analyses. The speci c aims are: 1. to analyse how functioning of patients with severe musculoskeletal injuries changes over the course of a rst inpatient rehabilitation stay until three months after discharge; and 2. to identify the factors in uencing these changes over time.

Design
We performed a longitudinal multicentre study within the project "Predicting the Rehabilitation Outcome after trauma based on the ICF" (icfPROreha), which was carried out in ten rehabilitation centres throughout Germany. The project aims to identify factors in uencing RTW and quality of life in patients with severe musculoskeletal injuries at the time of admission to inpatient rehabilitation by collecting data on sociodemographic, health-related aspects, functioning, quality of life and contextual factors. In this study, the WHODAS 2.0 was applied to a large population with severe musculoskeletal injuries in inpatient rehabilitation for the rst time.
Within the project an expert panel was implemented including 24 experts with different backgrounds related to clinical and rehabilitative practice, administration and research (physicians, therapists, rehabilitation managers, representatives of accident insurances) which guided the conceptualization of the project and methodolgy.

Participants
Our study sample consists of patients being admitted to rehabilitation at one of the ten participating rehabilitation centres between August 2018 and December 2019. We included patients who ful lled the following inclusion criteria: (a) aged 18-65 years (working age), (b) diagnosis of severe musculoskeletal injury, (c) rst inpatient rehabilitation treatment after injury, (d) trauma or accident occurred a maximum of 16 weeks ago, (e) aim of the study was understood and (f) participants provided informed consent.
We excluded patients with (a) lesions of the major nerves including spine injuries, (b) neurological symptoms, (c) traumatic brain injuries and (d) insu cient knowledge of the German language to ll in patient-reported outcome measures.

Measures
Within the project icfPROreha a comprehensive set of variables was used. This variable set was decided on and operationalized during a two-day structured and consensus-building process by the expert panel. The panel decided on the variables based on the results of scoping reviews and expert surveys performed by researchers of LMU Munich prior to the workshop. Standardized instruments, and more speci cally patient-reported outcomes, were used wherever possible. For the present study, we selected a reduced set of variables which was chosen based on published literature and discussions within the research team (see Table 1).
In this analysis, we used the following standardized patient-reported outcome measures: Functioning was assessed with the German version of the WHODAS 2.0 12-item self-administered or intervieweradministered version. The response options of the 12 items range from 0 = no di culties to 4 = extreme di culties/cannot do which sum up to a total score of 0 to 48, with zero indicating maximum functioning (19).
Self-e cacy was assessed with the General Self-E cacy Short Scale (Allgemeine Selbstwirksamkeit Kurzskala, ASKU). It was constructed for the general German-speaking population. The mean value of the three individual items makes up the total score (32). Resilience was measured with the Resilience Scale RS-11 which is a shortened version of the original scale by Wagnild and Young (1993) (33). The 11 items on a 7-point likert scale are added up to a total score (34). Heavy drinking and/or active alcohol abuse or dependence were identi ed using the Alcohol Use Disorder Identi cation Test Consumption Questions (AUDIT-C). The values of the three questions are summed up resulting in a possible score of 0-12 (35).

Data Collection
Patients ful lling the above-mentioned inclusion criteria were informed about the aim and procedure of the study and were invited to participate when admitted to one of the ten inpatient rehabilitation centres. Patients who were willing to participate signed informed consent. Gender, age and funding agency of rehabilitation were collected for the patients which did not agree to participate. Patients were recruited between August 2018 and December 2019. The study coordinators and physicians involved in the study underwent a training in recruitment strategy and data collection.
Patients lled in an electronic survey using a mobile device during the rst three days after admission (T1) and one to three days before discharge (T2). There was also the possibility to complete the survey using a paper-based form if there was a problem with the electronic device or a patient wished to do so. Three-month follow-ups (T3) were conducted by telephone interviews performed by trained interviewers. If the patients were contacted at least ve times by telephone and were not reached or if they speci cally requested it, a questionnaire was sent to them by postal mail.
Data was collected and anonymously stored on a protected server at LMU Munich using the web-based application Research Electronic Data Capture (REDCap) (36).

Descriptive Analysis
We examined the distributions of the variables (mean, standard deviations, absolute and relative frequencies) to ensure that the data met the assumptions for statistical tests. Some variables needed to be recoded due to low frequencies of response options. Table 1 shows the recoding of these variables.

Exploratory Analysis
We added each potential predictor without and with time interaction separately to the basic time model, and the model t was compared using the Akaike Information Criterion (AIC). The research team then decided on the variables to be included in a model consisting of so-called basic factors (such as age) and injury-related factors, as well as additional factors (personal factors, environmental factors).
We added the selected variables concerning basic factors and injury-related factors step by step to the model, beginning with the variables which showed the most improvement in AIC. To assess the effect of additional factors, we included them separately in the model, again using the AIC to compare the goodness of t of the models. To come up with a nal model, we included all selected basic, injury-related and additional factors in a full model and removed non-signi cant effects (p ≥ 0.05).
We imputed missing values in the standardized patient-reported outcome measures according to the respective instructions for analysis. Otherwise, we did not undertake missing data imputation. SAS software (version 9.4) (38) was used for all statistical analyses.

Descriptive Analyses
In total, 797 of 1060 eligible patients agreed to participate in the study by the end of the recruitment period. Those who refused to participate (n = 263, 24.8 %) did not differ in age and gender from those who decided to do so. In our analysis we included 571 participants (72.7 % male (n = 415), mean age 47.4 years (SD = 12.2)) from whom complete WHODAS 2.0 data over the three time points admission (T1), discharge (T2) and 3-month follow-up (T3) was available.
The mean duration of the inpatient rehabilitation was 5.4 weeks (SD = 3.4); 28.2 % (n = 160) of the patients suffered a polytrauma and 43.5 % (n = 247) were diagnosed with a severe injury. The characteristics of the study population is shown in Table 2. The most frequent diagnoses were fractures and dislocation of the foot (anklebone/heel bone/tarsals) and vertebra fractures, deformities and instabilities (9.2 % each) followed by fractures of the pelvic ring, deformity or instability (6.3 %). Table 3 (40) The mean WHODAS 2.0 total score was 21.0 (SD = 9.9) at admission, 14.9 (SD = 8.3) at discharge and 12.5 (SD = 7.6) at 3-month follow-up. Figure 1 shows details on the distribution of the WHODAS 2.0 mean score of the whole population for the different domains.

Modelling Change of Functioning over Time
The development of the WHODAS 2.0 score from admission over discharge to 3-month follow-up can be described with the following model: WHODAS score = 21.042-6.165*T2-8.524*T3

Exploratory Analyses
Out of our dataset which included 43 variables in total, 27 variables showed effects (p < 0.05) either alone or in interaction with time when they were added one by one to the model just including time (see Table 1 variables marked with #). Table 4 provides details on the improvement of model t for these variables, as well as the chosen variables to be included in further analyses based on consultations in the research team and their assignment to the blocks.
Although gender did not show signi cance (p ≤ 0.101) when being added to the model of time, it was decided to be entered as forced-in variable into further exploratory models. * Included in further analyses; # basic factor; † injury-related factor; § additional factor The results obtained from the multilevel analysis of the basic and injury-related factors are presented in Table 5. Each stepwise inclusion of both factors improved the AIC of the model. Except for the variables gender and severity of injury, all factors consistently showed signi cant effects in the models. Including the additional factors emotional functioning, pain, subjective prognosis on RTW, information about injury, sleep functions, rehabilitative goal, nancial assets and resilience separately in the multilevel model also resulted in an improvement of the AIC for each variable (see Table 6).

Final Model
A complete model including all predictors mentioned in Table 6 resulted in an AIC of 11356.6. The removal of the nonsigni cant effects (p ≥ 0.05) gender, sleep functions, nancial assets and resilience resulted in a drop of the AIC to 11352.1, which marked the best model. The nal model describing functioning over time therefore included time, age, duration of inpatient rehabilitation, severity of injury, injury localizations, number of comorbidities, emotional functioning, pain, information about injury, subjective prognosis on RTW and treatment targets de ned and agreed on (see Table 7).

Discussion
The present study shows that the mean total WHODAS 2.0 score measuring functioning in persons with severe musculoskeletal injuries declines over the course of a rst inpatient rehabilitation stay until 3 months after discharge, with a larger difference between admission to and discharge from the hospital than between discharge and 3-month follow-up. Of initially 43 possible factors in uencing the development of functioning, the following ten variables were included in the nal model to predict change in functioning additionally to time: age, duration of inpatient rehabilitation, severity of injury, injury localizations, number of comorbidities, emotional functioning, pain, information about injury, subjective prognosis on RTW and treatment targets de ned and agreed on.
The improvement in functioning over time -as measured with the WHODAS -seems plausible, since rehabilitation does not only aim to restore body functions, but also to improve patient's activities and participation (13). The general tendency of the WHODAS 2.0 to decline after an injury is also supported by other studies (8, 20,29,41), however, these studies most often refer to a longer period of time.
Prior studies are controversial about the in uence of gender on the WHODAS 2.0 score in trauma patients (20,21,29).
In our analysis, gender did not show signi cant effects when it was added to the basic time model. Despite this, we decided to treat it as forced-in variable in the exploratory models based on expert consultation. Removing gender resulted in an improved model t. In line with this the nal model did not contain the variable. This could be due to the fact that in our cohort, comorbidity seems to be a stronger predictor for the development of functioning over time than gender. Increasing age was associated with worsened functioning at discharge and 3-month follow-up in our nal model, even though it had no statistically signi cant in uence on the baseline WHODAS 2.0 score. This nding is in line with previous research (20,21). A longer period of inpatient rehabilitation was associated with poorer functioning at baseline, which might be explained by the nature of the injuries: Injuries resulting in worse functioning probably also require a longer inpatient rehabilitation. In contrast, at discharge and 3-month follow-up every additional week of inpatient rehabilitation positively in uenced functioning which is also con rmed by Soberg and colleagues (29).
In our study, injury severity was measured according to the Index of Injuries for Inpatient Treatment of the German Social Accident Insurance (DGUV) (39). Nevertheless, having a severe injury was found to be associated with functioning. This is also con rmed in other studies measuring severity using the Injury Severity Score (ISS) (42) or New Injury Severity Score (NISS) (43) (20,21,29,44). Having three or more injury localizations as well as having three or more comorbidities both resulted in an increase of the WHODAS 2.0 total score. This ts well with the results of Davie et al. (41) who found multimorbidity to be associated with poorer functioning.
We found that decreased emotional functioning was associated with worse functioning at baseline in our nal model. between healthcare professionals and patients to decide on the best procedure for rehabilitation (49). In our study, an existing agreement on a treatment targets as well as feeling informed about the injury and its consequences were associated with improved functioning. These results support the importance of goal setting and the involvement of patients in the patient-centred decision process. However, feeling informed did not show consistent effects at discharge and 3-month follow-up. In fact, the model even showed an increase in WHODAS 2.0 at 3-month follow-up. This suggests feeling su ciently informed might change during the course of the inpatient rehabilitation. Further research is needed to fully explain the reasons for this discrepancy.
We would like to stress that only data collected by the end of March 2020 was used for the analyses in this paper. By the end of March 2020, restraining orders and curfews had been imposed by the federal and state governments in Germany because of the COVID-19 pandemic. In telephone interviews, many participants reported that the restrictions due to the COVID-19 pandemic negatively affected their mood and their answers to the WHODAS 2.0 questionnaire. Therefore, we decided to exclude all data collected after this time point.
One major strength of our study was that the study population was comparable in terms of age and gender to the German trauma population according to the German TraumaRegister DGU® (mean age 50.5; 70.2 % male) (50). Furthermore, the study had a large sample size and different study sites across Germany. We used the WHODAS 2.0, a short, valid and reliable instrument recommended by WHO, to measure functioning, as well as other standardized instruments wherever possible. Since the second part of our study was of exploratory nature, the entire project team was involved to ensure that the preselection of variables was done in a meaningful way by taking into account a clinical and content-related perspective and was not based on statistical analyses only.
Several limitations might have in uenced the reported results. First, one must be aware that the results only cover patients with musculoskeletal injuries and are probably not transferable to other patient populations, especially to patients with severe neurological injuries. Second, the order in which the variables were added to the model was arbitrary, and a different order could have led to different results. Besides, it is unclear whether participants who could not be contacted for their 3-month follow-up are missing at random. This could be a potential source for bias.
The self-administered and the interviewer-administered version of the WHODAS 2.0 were applied, therefore there might be an effect on the participants' answers. Finally, the ndings from this study may not be transferable to a long-term change over time in functioning assessed with the WHODAS 2.0. The results can however be updated once the entire project is completed and will include further time points up to 78 weeks after discharge from inpatient rehabilitation.
In conclusion, our study identi ed several factors which can be collected at admission to inpatient rehabilitation to predict functioning of patients with musculoskeletal injuries. The results highlight relevant factors which could be targeted for intervention in rehabilitation settings. Providing support groups for instance could help to limit negative in uences of pain and impaired emotional functioning on patients' every-day lives. In order to strengthen the improvement of functioning, it could be bene cial to timely de ne and agree on a treatment targets and to ensure that the patient feels su ciently informed about the injury and its consequences. Ethic Committees of LMU Munich and the involved rehabilitation centres gave ethical approvals for the study.

List Of Abbreviations
Participants gave written informed consent to participate in the study.

Consent for publication:
Not applicable Availability of data and materials: The data that support the ndings of this study are available from the corresponding author, MC, upon reasonable request.
Competing interests: The authors declare no con ict of interest.