Assessment of Patient-Reported Outcomes After Polytrauma: A Systematic Review

Background: Tools assessing patient-reported outcomes (PROs) after polytrauma have not been systematically reviewed so far. Therefore, we sought to: (1) collect instruments that assess health-related quality of life (HRQoL), activities of daily living (ADL), and social participation during follow-up after polytrauma, (2) describe their use (e.g. frequency of application during follow-up), and (3) investigate other relevant PROs assessed in the included studies (e.g. pain, depression, anxiety). Methods: The search was conducted in MEDLINE, Embase, CINAHL, PsycINFO, CENTRAL, the trials registers ClinicalTrials.gov, and WHO ICTRP. All original empirical research published between January 2005 and April 2018 was included. Data extraction, narrative content analysis, and a critical appraisal were performed. Results: The search yielded 3496 hits and 54 publications were included in the review. Predominantly, HRQoL was assessed (in 48 of 54 publications), with Short Form-36 Health Survey (SF-36) applied most frequently. ADL and (social) participation were rarely assessed. The follow-up period lasted between 8 days and 17.5 years; assessments were performed between one and ve times. Most used were a single assessment of PROs, a follow-up period of 1 to 1.5 years, and a postal survey. Other relevant PRO areas reported were function, mental disorders, and pain. Conclusions: There is a large variation in the assessment of PROs after polytrauma, impairing comparability of outcomes. First efforts to standardise the collection of PROs have been initiated, but further harmonisation between central players, such as trauma registers, is required. Denition and continuous collection of core outcome sets would improve comparability of study results, creating the basis for more targeted treatment. Additional knowledge on rarely reported PRO areas like (social) participation and social network may lead to their consideration in health services provision. ndings, impairs the comparability of after polytrauma. this standardised approach for recording PROs First and (social)


Introduction
Severe injuries represent a leading cause of death and permanent disability (1). In the central European region, severely injured people are referred to as having a 'polytrauma' or being 'polytraumatised' (2). A polytrauma is de ned as at least two severe injuries in different body regions or organ systems, at least one of which is potentially life-threatening by itself or in combination with others, with an Injury Severity Score (ISS) ≥ 16 (1)(2)(3)(4). According to the TraumaRegister DGU®, a German registry that covers patients with severe injuries, the number of patients with polytrauma (ISS ≥ 16) was 15,651 in 2019 (5). Most affected were men (72%) with a mean age of 46.5 years (6).
The survival rates after polytrauma have increased lately, bringing the quality of survival into the focus of research. One year after polytrauma, survivors often fail to reach pre-traumatic status due to persisting medical problems (7). Disabilities such as respiratory problems, paraplegia or tetraplegia, prevent 40% from returning to their former workplace (8,9). They also have an impact on socio-economic aspects and quality of life (9)(10)(11). Therefore, it is crucial to reduce the short-term and long-term burden after polytrauma (9,11). In this context, the patients' perspective on polytrauma outcomes focussing on psychological, physical and social factors is getting more important for the patients themselves, their relatives, and healthcare professionals (11,12). Evaluating these patient-reported outcomes (PROs) is essential to complete the holistic assessment of relevant clinical outcomes after injury from a patient's perspective.
To assess the impact of polytrauma on health-related quality of life (HRQoL) and other important PROs, one must consult the person affected directly (12,13).
HRQoL is already being recorded in registers like the Australian Victoria State Trauma Registry. The German TraumaRegister DGU® will also complement its assessment with HRQoL (14). Activities of daily living (ADL), de ned as basic activities to ful l independent living, along with physical and psychological needs, are meaningful outcome parameters in case of post-polytrauma impairment. Participation is a health component according to the International Classi cation of Functioning, Disability and Health (ICF) (15). In recent years, it has gained more attention in political context (16). Several working groups have controversially discussed which tools should be applied to assess PROs after polytrauma, at which time points, and in which way (13,(17)(18)(19)(20)(21)(22)(23)(24)(25); however, guidelines with clear recommendations are lacking (13), and no systematic review could be identi ed.
Therefore, the aim of this systematic review was to: 1. collect instruments assessing HRQoL, ADL, and social participation during follow-up after polytrauma 2. describe their application in detail (e.g. duration of follow-up period, frequency of application, and time point of measurements within the followup period) 3. investigate which other relevant PROs are additionally assessed in included studies (e.g. pain, depression, anxiety, and cognitive function) without detailed analysis.

Methods
This systematic review was conducted in line with the Preferred Reporting Items for Systematic Review and Meta-Analysis (PRISMA) guideline (26) (Additional le 1). The review protocol was registered on 8 April 2017 at PROSPERO (CRD42017060825).

Data Sources and Search Strategy
The search was conducted in MEDLINE, Embase, CINAHL (Cumulative Index to Nursing and Allied Health Literature), PsycINFO, CENTRAL (Cochrane Central Register of Controlled Trials), the trials registers ClinicalTrials.gov and WHO ICTRP (WHO International Clinical Trials Registry Platform), covering the period from 1 January 2005 until 5 April 2018. In addition, reference lists of eligible studies were examined. Textwords (tw) and MeSH-terms (mh) like 'polytrauma' (tw), 'multiple trauma' (mh), 'quality of life' (mh), 'activities of daily living' (tw), 'social participation' (mh) and 'pain' (mh) were used. The development of the search strategy was outlined in the study protocol (27) and is provided in Additional le 2.

Inclusion and Exclusion Criteria
The publications included primarily investigated PROs of patients aged 18-75 years with an Injury Severity Score (ISS) ≥ 16 and/or an Abbreviated Injury Scale (AIS) ≥ 3 during a clearly de ned follow-up period. All original empirical research published in English or German was included. Publications examining special types of injury or illness -e.g. low-energy injuries, single or mono injuries, burn injuries, war injuries, or cancer -were excluded. We also excluded some publication types (grey literature, books, abstracts, and editorials). Systematic reviews and meta-analysis were not included, too, but screened to identify further appropriate studies. More details about the inclusion and exclusion criteria were published in the study protocol (27).

Selection of Publications and Studies
Four teams with two reviewers each (MR-IG; KM-SK; MR-SOB; AI-AHF) screened the title and abstracts independently. To foster the process of eligibility decisions, the inclusion or exclusion of the rst 50 publications was discussed by two reviewers (MR-IG). In a next step, two reviewers (MR-IG; KM-SK; MR-SOB, KM-SOB) independently screened the full texts of the remaining publications for nal inclusion. A third reviewer (MR, SK, IG) solved disagreements regarding the inclusion of publications for full-text screening and their nal inclusion in this review. We identi ed publications relating to the same study and/or population. For this purpose, extensive research was conducted, and authors of the corresponding publications were contacted. The section 'Characteristics of included publications' outlines how we approached publications dealing with different research questions of the same study, comprising subpopulations drawn from a superior study population, reporting subprojects of a study project, or presenting results of the same research question at different follow-up periods.

Data Extraction and Analysis
Data extraction was performed according to the requirements of Cochrane reviews (28). For the rst and second aim, two researchers (MR-IG) used a piloted data extraction sheet independently comprising the following: rst author, publication year and country, study design, aim of the study (objective), study population, treatment, applied instruments (incl. modi cations), time of measurements, length of follow-up period, method of assessment (application), study results, and results of the critical appraisal. Regarding the third aim, the other reported PROs were extracted and categorised in a separate extraction sheet. Narrative content analysis was performed regarding the assessment of HRQol, ADL and social participation as well as further reported PROs.

Critical Appraisal
Teams with two reviewers each (MR, IG, SK, KM, SOB) independently assessed the methodological quality and the risk of bias in each study type using standardised checklists of the UK National Institute for Health and Care Excellence (NICE) (29,30), Scottish Intercollegiate Guidelines Network (SIGN) (31), or the Mixed Method Appraisal Tool (MMAT) (32). Discrepancies were resolved by discussion between the reviewers or by a third reviewer.

Patient and Public Involvement Statement
Patients were not involved in the design and implementation of this systematic review.

Search Results
In total, 3485 references were retrieved. After a title and abstract screening as well as a full-text screening, further 11 records were identi ed by backward citation tracking. Finally, 54 eligible publications were included for data extraction (Fig. 1).

Instruments Used to Assess (Social) Participation
One publication reported the assessment of participation by self-designed questions covering return to sports and work (52). Table 2 displays the number of publications and the instruments used to assess HRQoL, ADL, and (social) participation.

Further PROs Assessed
Beyond HRQoL, ADL, and (social) participation, assessments of further PROs were reported in the included publications. Most frequently these comprised function, pain, and mental disorders (including depression and/or depressive symptoms, anxiety, and PTSD). Only few PROs focused on social support, cognition, subjective health status, and social network. Table 3 provides an overview of the kind and number of further recorded PROs.

Main Findings
The present systematic review identi ed 54 publications contributing to the research questions. A great variety was observed among all investigated aspects. HRQoL was the most frequently recorded PRO, and SF-36, SF-12 and EQ-5D were the most frequently applied instruments to assess it. Data on ADL and (social) participation were rarely reported. Most questionnaires were sent out per mail. Measurements were predominantly performed at a single time point, and the follow-up period lasted mainly between 1 and 1.5 years. Function, mental disorders and pain were the most frequently assessed categories of the other PROs.

Instruments for the Assessment of HRQoL, ADL, and (Social) Participation
In 48 of the included publications, HRQoL was assessed -predominantly with generic and preference-based HRQoL instruments such as SF-36, SF-12, EQ-5D, and NHP. Thus, over a long period of time, results are available for the impact of polytrauma on HRQoL and its possible change over time.
Unsurprisingly, the SF-36, one of the oldest and most widely used instrument for assessing HRQoL (85), is the most frequently reported instrument in publications included in this review. Along with the SF-12, SF tools accounted for 61% of all identi ed instruments. This result is supported by observations of other reviews conducted in the eld of polytrauma (85-88).
In 1999, an international consensus conference on evidence-based guidelines for the systematic evaluation and assessment of HRQoL after polytrauma recommended the use of modular instruments comprising generic and trauma-speci c aspects (89). Generic instruments have been applied continuously over a long time period, have been translated into different languages, and can be applied regardless of the underlying disease. They are suitable for assessing HRQoL after polytrauma since it can affect different body regions, show different injury patterns, and lead to a variety of complications. In contrast, traumaspeci c instruments capture targeted relevant problems and measure the longitudinal change of HRQoL (90). Despite the recommendation, the POLO Chart is the only modular instrument comprising generic and trauma-speci c aspects reported in the included publications (90). HASPOC, TOP and ALOS were the trauma-speci c tools used. HASPOC is an instrument designed to evaluate the quality of rehabilitation. Since it comprises instruments assessing inter alia HRQoL, ADL, and function (91,92), we decided to subsume it into the category 'HRQoL'.
Apart from assessing HRQoL, we were particularly interested in the measurement of ADL and social participation. Regarding the assessment of ADL, ve different instruments and self-designed questions in six publications were observed in our review. (Social) participation was only reported once by Holtslag et al. (2006) (52). In summary, it was notable that ADL and (social) participation after polytrauma were seldom assessed. One reason might be that measures of HRQoL or function already include several questions related to ADL and (social) participation. Predominantly, (social) participation is researched in people with brain or spinal cord injuries. A general understanding and clear de nition of the participatory construct is still missing, which might be another reason for the low reporting rates of (social) participation. It has been argued that (social) participation should be one of the main aspects in rehabilitation after polytrauma (93), besides survival and function, and there are a number of instruments available to assess this PRO (94)(95)(96). This could also be assumed for ADL.

Application of Instruments
Most frequently, a postal questionnaire was used. Compared to personal interviews, questionnaires sent by mail are quicker and less expensive. In addition, potential bias introduced by an interviewer is excluded. However, it is not possible to verify the person answering the questionnaire, comprehension questions cannot be clari ed, the full completion of the questionnaire cannot be guaranteed, and the proportion of non-responders is higher (97). The consultation of a proxy answering the questions was rarely reported. Injuries, e.g. traumatic brain injuries (TBI), may require the use of a proxy and could enable the collection of outcome data after polytrauma, despite a risk of bias (97,98). The present systematic review excluded publications reporting single or mono injuries like single TBIs, which might be the reason only few publications including proxies were detected.
The retrospective assessment of the pre-trauma situation helps enable a comparison of results and obtain a holistic view of the outcome after an injury. Gross et al. (2010,2012) as well as Gross and Amsler (2011) measured pre-fracture level of HRQoL over 2 years after trauma (47)(48)(49). However, there are limitations to this approach: Memories of the pre-traumatic situation can be distorted by various circumstances, which might increase the risk of recall bias (99). Large population screenings or the use of population norms might help to evaluate the pre-traumatic state of HRQoL (19,46). Baseline assessments should ideally be made in the rst week after the event. However, this is not always possible due to the severity of the injury. Another limitation might be that some instruments are not suitable for use in hospital. For example, changed behaviour patterns measured by the SIP can also result from the very fact of being an inpatient. Apart from HRQoL, functional recovery is also of great interest and an important factor for evaluating outcomes after a polytrauma. In comparison to HRQoL and due to better quanti cation, the retrospective evaluation of function is more feasible (19).
Most of the included publications reported a single assessment of the PROs. However, the course of recovery cannot be derived from a single measurement.
The aforementioned consensus conference recommended measurements at three, twelve and 24 months after the event (18). Other groups suggested assessments of PROs after one, two, four and twelve months (100) or six and twelve months after the injury, and beyond twelve months in case of longer rehabilitation needs (13).
The most frequently reported follow-up period were 1 to 1.5 years. Six publications conducted follow-up periods of more than ten years.

Further PROs Assessed
Further reported PROs assessed mainly function, mental disorders, and pain -outcomes that are highly affected by polytrauma. Social support, subjective health status, cognition, or social network were rarely recorded. This might be explained as follows: Instruments assessing HRQoL mainly capture subjective health status. While cognition is frequently affected by brain injuries, publications reporting study populations with single brain injuries were excluded from this review. Social network and social support seem to in uence recovery and return to work after trauma (101, 102) and should therefore receive more attention in future research.

Recommendations for the Assessment of PROs
The existing recommendations for the use and application of instruments -as well as time points of assessments and outcomes to be measured after polytrauma -differ substantially (13,18), but there are attempts to standardise outcome measurements (19). Classi cation systems might be helpful for a standardised assessment of PROs, facilitating the adequate selection of instruments for a speci c research objective (24). The internet platform PROMIS -Patient-reported Outcomes Measurement Information System -provides research tools and is continuously updated with the latest developments and newly translated instruments (103). An important step towards a standardised approach of collecting HRQoL was made in the TraumaRegister DGU® by the implementation of HRQoL assessment with the SF-12 (14).
Furthermore, several groups have focused on the de nition of special outcome variables to be recorded as default parameters. Hoffman et al. (2016) recommended the development of a 'minimum data set' based on the ICF (21). A core outcome set relevant for the respective outcome could be de ned. This procedure could guide decision-making for the collection and presentation of outcomes in studies (23). In addition, ICF linking enabling a check of which ICF components are mapped by an instrument could support the tool selection for particular research questions (17,20). Other authors also support the idea of core outcome sets (22,25).

Strengths and Limitations
The present systematic review has some limitations: Selection bias can be assumed as the search was restricted by the search period and language. However, we performed backward citation tracking and saw that only a few additional articles could be included.
This review was conducted thoroughly based on a registered protocol in line with the PRISMA-P quality and the PRISMA statement requirements. We provide a broad overview of instruments used for the collection of HRQoL, ADL and (social) participation, along with information on their application. In addition, this study offers an overview on further PROs reported in the context of polytrauma.

Conclusion
The present review reveals variations across study ndings, which impairs the comparability of outcomes after polytrauma. To improve this situation, a more standardised approach for recording PROs should be established. First efforts have been initiated, but further alignment between the central players, such as the trauma registers, is required. Continuous collection of core outcomes during the regeneration of persons with polytrauma could create the basis for a targeted use of treatments and their adaptation to the results of the outcome measurement. Moreover, there appears to be little research on (social) participation and on further PROs like social support and social network. PROs need more attention in polytrauma-related outcome research. After all, we need to know the patients' perspectives on relevant outcomes after polytrauma so as to consider their needs when providing health services.