Neuderth et al.

BMC Public Health (2016) 16:804

DOI 10.1186/s12889-016-3437-7

STUDY PROTOCOL Open Access

Work-related medical rehabilitation in

patients with musculoskeletal disorders:
the protocol of a propensity score
matched effectiveness study (EVA-WMR,
DRKS00009780)
Silke Neuderth1, Betje Schwarz2, Christian Gerlich3, Michael Schuler3* , Miriam Markus2 and Matthias Bethge2

Abstract
Background: Musculoskeletal disorders are one of the most important causes of work disability. Various rehabilitation
services and return-to-work programs have been developed in order to reduce sickness absence and increase
sustainable return-to-work. As the effects of conventional medical rehabilitation programs on sickness absence duration
were shown to be slight, work-related medical rehabilitation programs have been developed and tested. While such
studies proved the efficacy of work-related medical rehabilitation compared with conventional medical rehabilitation
in well-conducted randomized controlled trials, its effectiveness under real-life conditions has yet to be proved.
Methods/Design: The cohort study will be performed under real-life conditions with two parallel groups. Participants
will receive either a conventional or a work-related medical rehabilitation program. Propensity score matching will be
used to identify controls that are comparable to treated work-related medical rehabilitation patients. Over a period of
three months, about 18,000 insured patients with permission to undergo a musculoskeletal rehabilitation program will
be contacted. Of these, 15,000 will receive a conventional and 3,000 a work-related medical rehabilitation. We expect a
participation rate of 40 % at baseline. Patients will be aged 18 to 65 years and have chronic musculoskeletal disorders,
usually back pain. The control group will receive a conventional medical rehabilitation program without any explicit
focus on work, work ability and return to work in diagnostics and therapy. The intervention group will receive a
work-related medical rehabilitation program that in addition to common rehabilitation treatments contains 11 to
25 h of work-related treatment modules. Follow-up data will be assessed three and ten months after patients’
discharge from the rehabilitation center. Additionally, department characteristics will be assessed and administrative
data records used. The primary outcomes are sick leave duration, stable return to work and subjective work ability.
Secondary outcomes cover several dimensions of health, functioning and coping strategies.
Discussion: This study will determine the relative effectiveness of a complex, newly implemented work-related
rehabilitation strategy for patients with musculoskeletal disorders.
Trial registration: German Clinical Trials Register (DRKS00009780, February 10, 2016).
Keywords: Work-related medical rehabilitation, Effectiveness, Work ability, Return to work, Propensity score matching

* Correspondence: [email protected]
3
Department of Medical Psychology and Psychotherapy, Medical Sociology
and Rehabilitation Sciences, University of Würzburg, Klinikstraße 3, Würzburg
97070, Germany
Full list of author information is available at the end of the article

© 2016 The Author(s). Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0
International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and
reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to
the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver
(http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.
Neuderth et al. BMC Public Health (2016) 16:804 Page 2 of 11

Background musculoskeletal disorders in seven rehabilitation centers.

About 35 million people aged 15 to 64 (11.0 %) in the The results showed that the implementation was challen-
European Union are limited in the work they can do ging but feasible. Moreover, the observational study showed
because of a long-standing health problem or a basic a significant reduction of sick leave for WMR participants,
activity difficulty [1]. As musculoskeletal disorders are still and more work-related interventions predicted shorter
one of the most important causes of work disability, vari- sickness absence at the three-month follow-up [19]. How-
ous rehabilitation services and return-to-work programs ever, findings also indicated that allocation and treatment
have been developed and implemented in order to battle decisions (WMR vs. CMR) only partly followed the guide-
the consequences of musculoskeletal disorders for the line recommendations [20].
opportunity to participate in working life [2]. As shown by While studies proved the efficacy of WMR compared
systematic reviews, such programs and interventions with CMR in well-conducted randomized controlled
indeed may significantly reduce sickness absence and trials with high treatment fidelity and carefully selected
increase sustainable return-to-work, especially if they patients, the effectiveness of WMR under real-life condi-
include work-related elements [3–5]. tions has yet to be proved. Interventions that work in
In Germany, rehabilitation programs for musculo- efficacy studies may not necessarily also do well in real-
skeletal disorders are usually provided as three-week world applications [21, 22]. Though reliable efficacy studies
in- or outpatient interventions under the umbrella of the are a necessary condition for evidence-based practice,
German Pension Insurance (GPI). The treatment is multi- patients and other stakeholders may be most interested in
professional and follows evidence-based therapy recom- the effectiveness of real-world services. While a randomized
mendations [6]. However, change in sickness absence controlled trial is the gold standard and the most robust
duration was shown to be slight in meta-analyses of obser- way to prevent allocation bias in efficacy studies, other
vational studies [7, 8]. One randomized controlled trial designs and methods may be needed to evaluate the effec-
indicated no effect on sickness absence duration [9]. More- tiveness of an intervention, especially, as in the case of
over, several studies revealed that patients with severe WMR, where nationwide dissemination is completed and
restrictions of work ability (e.g. long-term sick leave, poor programs and allocation procedures are established. Never-
return-to work expectation) have a considerable risk of not theless, as the WMR guideline clearly describes the patients
returning to work despite participating in a conventional in need and proposes several screenings to identify these
medical rehabilitation (CMR) program [10, 11]. Therefore, patients [11], allocation bias is a severe challenge in deter-
work-related medical rehabilitation (WMR) programs have mining an unbiased treatment effect estimate. One proper
been developed and tested in recent years [10, 12]. evaluation design in this case is the use of observational
These programs were designed to particularly support data and a subsequent propensity score matched analysis to
patients with severe restrictions in terms of ability to control for confounding. We therefore designed a cohort
work. WMR programs regularly comprise a diagnostic study to analyze the relative effectiveness of WMR com-
assessment that compares job demands and patients’ work pared with CMR. As current allocation procedures regar-
capacity and offer therapeutic interventions such as work ding WMR seem to be far from perfect and WMR and
hardening and work-related functional capacity training, CMR patients have a considerable overlap, a propensity
work-related psychosocial groups and intensified social matched comparison was chosen to address allocation bias.
counseling [13–16]. A meta-analytic synthesis of rando- We hypothesize that WMR reduces sickness absence and
mized controlled trials that compared WMR and CMR improves sustainable return-to-work and work ability ten
programs in Germany provides robust evidence that in months after rehabilitation (primary outcomes) compared
patients with musculoskeletal disorders WMR programs with CMR. Moreover, we expect to see evidence of the
achieve higher rates of sustainable return-to-work and superior effectiveness of WMR regarding several secondary
reduce sickness absence after one year. WMR participants outcomes. The study protocol has been prepared according
had 2.4 times higher odds of sustainable return-to-work to the SPIRIT checklist (Standard Protocol Items: Recom-
after one year and also reduced sickness absence [17]. mendations for Interventional Trials) [23].
To support nationwide implementation of such pro-
grams and to improve rehabilitation outcomes, the GPI Methods
developed a WMR guideline. This guideline defines inclu- Study design
sion criteria for WMR (e.g. long-term sick leave, poor The study is a cohort study under real-life conditions with
self-rated return-to-work expectation, unemployment) two parallel groups. Participants will either receive a WMR
and important diagnostic and therapeutic measures in or a CMR program in one of the 256 approved rehabilita-
WMR (see Treatment section for a more detailed tion departments. The allocation ratio is determined by the
description of WMR) [13, 18]. The implementation of actual allocation under real-life conditions of rehabilitation
the WMR guideline was evaluated in patients with service provision in Germany. The current ratio of WMR
Neuderth et al. BMC Public Health (2016) 16:804 Page 3 of 11

to CMR patients is approximately one to five. The investi- pain management and relaxation training. CMR programs
gators will not have any influence on allocation decisions. focus on the functional limitations of the musculoskeletal
Current utilization of WMR is still considerably below the system and aim to restore physical abilities to promote
estimated number of patients who need it. Therefore, com- participation in work and daily living. However, in con-
parable controls are available who receive a CMR, though trast to WMR programs, they do not integrate an explicit
their severe restrictions in terms of work ability would also focus on work, work ability and return to work in diagnos-
justify participation in a WMR program. Although the tics and therapy.
researchers have no influence on the allocation procedure
they will have the opportunity to model treatment alloca- Intervention
tion by observed data. Thus propensity score matching will Participants of the intervention group will receive a WMR
be used to identify controls that are comparable to treated program according to the guideline for WMR [13, 18] as
WMR patients and to estimate the unbiased effects of the well as to the current treatment standards and guidelines
relative effectiveness of WMR compared with CMR. As for the rehabilitation of musculoskeletal disorders. WMR
randomized controlled trials confirmed the relative efficacy programs last approximately 2.4 days longer than CMR
of WMR in trials with high treatment fidelity and carefully programs since they contain 11 to 25 h of work-related
selected patients, and implementation of WMR is com- treatment modules [17]. Like CMR programs WMR
pleted, we assume the superiority of WMR. programs follow a multimodal approach that comprises
Baseline data will be assessed after approval of the sports and exercise therapy, physiotherapy, occupational
rehabilitation program but before the patients begin their therapy, massage and other physical therapies, social and
rehabilitation. Follow-up data will be assessed three and psychological counseling, patient education, pain manage-
ten months after their discharge from the rehabilitation ment and relaxation training. However, WMR programs
center. Moreover, administrative data records will be used. more explicitly focus on work, work ability and return to
Additionally, department characteristics will be assessed work. Thus, they include intensified work-related diagnos-
by a departmental survey. No one will be blinded before, tics as well as work-related functional capacity training,
during or after the trial. work-related psychosocial groups and intensified social
counseling. Intensified work-related diagnostics identify
Study setting individual needs by comparing work-related physical and
All included rehabilitation centers are located in Germany. psychosocial functional capacity with the patient’s job
Rehabilitation services may be provided as inpatient or demands. Assessment of functional capacity is performed
outpatient programs. Most of the departments are inpatient by means of a short functional capacity evaluation and
centers. There are some outpatient departments that are questionnaires. Demands will be assessed by job analysis.
mainly located in major cities. Participation in a rehabili- Matching of capacity and demands is supported by
tation program was approved by the Federal GPI. In both standardized assessments [24–26]. Work-related func-
groups, interventions will be performed by rehabilitation tional capacity training exercises specific movements and
physicians, psychologists, physiotherapists, sports thera- postures according to individual workplace conditions.
pists, social workers, occupational therapists and other Work-related psychosocial groups deal with the mutual
health professionals. dependence of behavioral and emotional health expe-
The duration of the rehabilitation program is initially riences and the workplace environment. Additionally,
determined by the Federal GPI (usually about three preventative measures like stress management or conflict
weeks). The rehabilitation center and the patient may management in dealing with psychosocial stressors are
arrange to extend the program. By request the patient taught. Finally, intensified social counseling examines the
may stop the rehabilitation program ahead of schedule. individual work-life situation and provides socio-legal
guidance and advice concerning further assistance within
Treatment the social security system.
Control
Participants of the control group will receive a CMR Participants
program according to current treatment standards and Patients are aged 18 to 65 years and have chronic muscu-
guidelines for the rehabilitation of musculoskeletal dis- loskeletal disorders, usually back pain. They have approval
orders. CMR programs last approximately three weeks. for rehabilitation either in an own rehabilitation institu-
The daily quantum of therapy amounts to three or four tion of the Federal GPI or an institution that is leadingly
hours. Following a multimodal approach, CMR programs occupied by the Federal GPI. Patients have requested
include sports and exercise therapy, physiotherapy, occu- rehabilitation because of health-related restrictions of
pational therapy, massage and other physical therapies, work ability. Need for rehabilitation was acknowledged by
social and psychological counseling, patient education, a registered doctor and approved by the Federal GPI.
Neuderth et al. BMC Public Health (2016) 16:804 Page 4 of 11

Allocation to CMR and WMR is primarily decided by the measure was approved within the previous four weeks will
socio-medical service of the Federal GPI. The socio- be contacted in each wave. The letter contains information
medical service may consider the findings of standardized about the study with the request to participate and the
screening which is usually part of the application docu- baseline questionnaire. The questionnaire contains no
ments and estimates the risk of not returning to work [11]. personal data but a unique study identifier. No personal
data (e.g. name and address) will be submitted or published
Sample size estimation by the Federal GPI. One week after the initial postal contact
We expect to find small mean differences between WMR a reminder will be sent to all potential participants. All
and CMR in primary outcomes. For three months, of participants will be thanked and reminded to participate if
about 18,000 insured patients with permission for a they have not yet done so. By this method, we expect to
musculoskeletal rehabilitation program 15,000 will receive increase participant enrolment in order to reach our esti-
CMR and 3,000 will receive WMR. We expect a partici- mated target sample size and to strengthen external valid-
pation rate of 40 %, i.e. 7,200 patients (6,000 CMR, 1,200 ity. Informed consent is assumed if patients complete their
WMR) at baseline. At the 10-month follow-up, we expect questionnaire and send the questionnaire to the research
a dropout rate of 25 %, i.e. 900 patients with WMR and team. Additionally, participants will be asked to give
4,500 CMR left. Using a one-to-one match without permission to use administrative data from their pension
replacement, 1,800 patients (900 WMR, 900 CMR) will be insurance accounts and to link these data with the ques-
analyzed. Assuming a number of clusters (departments) tionnaire data.
k = 256 (with 176 departments providing CMR and 80 At the three- and ten-month follow-up participants who
departments providing additionally WMR), a cluster size complete the baseline questionnaire will receive follow-up
of m = 7.0, an intracluster correlation of rho = 0.05, a questionnaires from the Federal GPI (Table 1). Three
power of 0.80 and a Bonferroni corrected p-value = 0.016, weeks later the questionnaire will be sent again with a
an effect of SMD = 0.175 can be detected (Fig. 1). reminder to all participants who have not completed the
questionnaire yet.
Recruitment
Participants with chronic musculoskeletal complaints Data management
whose request for rehabilitation was approved will be con- Questionnaires will be scanned and verified by an elec-
tacted by postal mail. Mails will be delivered in three waves tronic data capture system and exported to statistical soft-
at four-week intervals. All persons whose rehabilitation ware packages for further analysis. Scanning and verifying

Federal German Pension Insurance

allocating 18,000 patients

CMR WMR
15,000 patients 3,000 patients

Study information Study information

and baseline questionnaire and baseline questionnaire

Exclude: Exclude:
60% no consent 60% no consent

6,000 consenting and returning 1,200 consenting and returning

baseline questionnaire baseline questionnaire

CMR WMR
at 176 rehabilitation departments at 80 rehabilitation departments

Follow-up questionnaires Follow-up questionnaires

3 months 3 months
10 months 10 months

Administrative Analyze: Analyze: Administrative

data 900 patients 900 patients data

Fig. 1 Flow of participants

Neuderth et al. BMC Public Health (2016) 16:804 Page 5 of 11

Table 1 Schedule of enrolment, interventions, and assessments

STUDY PERIOD

TIMEPOINT Allocation Pre- Beginning End of the 3-month 10-month

intervention of the intervention follow-up follow-up
intervention

ENROLMENT

Information letter X

Allocation (under real-life conditions) X

INTERVENTIONS

Conventional medical rehabilitation

Work-related medical rehabilitation

ASSESSMENTS

Patient baseline survey X

Patient 3-month follow-up X

Patient 10-month follow-up X

Department survey X

Administrative data on received X

therapeutic interventions

Administrative data on income and welfare X X

benefits

will be done by trained research assistants. They will be extracted from administrative records (i.e. individual
check electronically processed data item by item and pension insurance accounts) provided by the Federal GPI.
compare imported data with the original questionnaire A complete list of all measured constructs, measurement
data. Administrative data will be extracted by the Federal points and expected scaling is shown in Table 2.
GPI. Personal data will be removed and replaced by the
unique study identifier. Data will be transferred to the Primary outcomes
principal investigator (MB). Finally, questionnaire and The primary outcomes of this study are sick leave dur-
administrative data can be linked by the unique study ation, stable return to work, and subjective work ability.
identifier. Evaluation of the effectiveness of WMR relates to the
Data management will be done by the authors of the 10-month follow-up. These three outcomes will also be
protocol. Data access is limited to the authors and the assessed at the three-month follow-up and if appropriate
research assistants of the research team. at baseline.
To assess the duration of sick leave, the participants will
Outcomes and other measures be asked to report the number of weeks they have been
This study will assess three primary outcomes as well as off work for health reasons since discharge from the
secondary outcomes, moderator variables and variables rehabilitation center. At baseline this question is related to
for propensity score matching. Outcomes and other mea- the last 12 months. Stable return to work has been defined
sures will be assessed with patient questionnaires or will in accordance with Kuijer and colleagues [27] as a
Neuderth et al. BMC Public Health (2016) 16:804 Page 6 of 11

Table 2 Measures, assessment, expected scaling and measurement occasions

Outcome Source and reference Scaling Baseline 3-month follow-up 10-month follow-up
Primary outcomes
Sick leave duration in weeks Own development metric X X X
Stable return to work Own development binary X X X
Subjective work ability WAS [28] metric X X X
Secondary outcomes
General health COPSOQ (1 item) [30, 31] metric X X X
Physical functioning IRES-24 [32] metric X X X
Depression PHQ-2 [33] metric/binary X X X
Anxiety GAD-2 [34] metric/binary X X X
Pain intensity CPQ [35, 36] metric X X X
Pain disability CPQ [35, 36] metric X X X
Fear of movement FABQ [37, 38] metric X X X
Physical activity Modified Godin-Scale [39] metric X X X
Medication use Own development nominal X X X
Self-management skills heiQ [40] metric X X X
Employment Own development binary X X X
Current sickness absence Own development binary X X X
Subjective prognosis of employment status SPE [41] metric/binary X X X
Implementation of work-related interventions [19] metric X
Consistency of the work-related strategy [19] metric X
Work-related benefit [19] metric X
Treatment satisfaction CSQ-8 [42] metric X
Income and welfare benefits GPI accounts metric X X
Therapeutic interventions GPI accounts metric X
Department characteristics Department survey metric/nominal X
Other measures
Risk scores for not returning to work SIMBO [11], WS [12] X
Somatization SCL-90-R [43] metric X
Pain generalization Own development metric X
Psychosocial stress [44] metric X
Socio-demographic data Own development metric/nominal X
Size of company [45] nominal X
Physical job demands [46] metric X
Effort-reward imbalance [47] metric X
Overcommitment [47] metric X
Support by co-workers and supervisors Own development metric X
Notes: SCL-90-R Symptom Check-List-90-R, SIMBO Screening-Instrument zur Feststellung des Bedarfs an medizinisch-beruflich orientierten Maßnahmen in der
medizinischen Rehabilitation (Screening to assess the need for work-related medical rehabilitation), COPSOQ Copenhagen Psychosocial Questionnaire, CPQ Chronic
Pain Grade Questionnaire, CSQ-8 Client Satisfaction Questionnaire, FABQ Fear-Avoidance Beliefs Questionnaire, GAD-2 Generalized Anxiety Disorder Questionnaire,
heiQ Health Education Impact Questionnaire, IRES-24 Indikatoren des Reha-Status (Indicators for rehabilitation status), PHQ-2 Patient Health Questionnaire,
SPE Subjective prognosis of employment status, WAS Work Ability Score, WS Würzburger Screening

minimum of four weeks of employment without sick work ability with the lifetime best. The 11-point scale
leave. Subjective work ability will be assessed by the Work ranges from zero (complete incapacity to work) to ten
Ability Score (WAS), which is the first item of the Work (lifetime’s best work ability). The WAS is highly correlated
Ability Index (WAI) [28]. It compares current subjective with the overall WAI score [29].
Neuderth et al. BMC Public Health (2016) 16:804 Page 7 of 11

Secondary outcomes scaling from zero (completely disagree) to 10 (completely

Secondary outcomes cover several dimensions of health, agree) instead of the original scaling. A composite score is
functioning and coping strategies (see below) and will be determined by averaging both item scores.
measured at all three measurement points. At the three-
month follow-up, participants will also evaluate how the Physical activity A modified version of the Godin
rehabilitation programs dealt with work-related issues as Leisure-Time Exercise Questionnaire [39] will be used to
well as satisfaction with the treatment. assess how often per week and how long per session
patients performed light, moderate and strenuous phy-
General health One item of the Copenhagen Psychosocial sical exercise. The total physical activity score in minutes
Questionnaire will be used to measure general health. The per week will be calculated by multiplying the total
11-point scale ranges from zero (worst imaginable health number of sessions per week (in each domain) by the mi-
state) to 10 (best imaginable health state) [30, 31]. nutes per session (in each domain). Additionally, a total
sum score will be calculated by multiplying the scores of
Physical functioning Physical functioning is measured each domain with the metabolic equivalent of tasks
by eight items of the IRES-24 (German: Indikatoren des (9 = strenuous, 5 = moderate and 3 = light physical exercise).
Reha-Status) [32], a widely used instrument in rehabili-
tation research in Germany. All items are measured on a Medication use Medication used to reduce pain (e.g.
five-point scale. Item values are averaged and rescaled to Paracetamol), to enhance mood (e.g. Citalopram) or to
a range from zero to 10 points, with lower values indi- treat other health complaints will be assessed by three
cating less functioning. new developed items. Response categories are no, regu-
larly (e.g. daily) and rescue medication.
Depression and anxiety The two-item versions of both
the Patient Health Questionnaire (PHQ-2) and the Self-management skills Self-management skills will be
Generalized Anxiety Disorder Questionnaire (GAD-2) assessed with three items of the scale Skill and technique
will be used to assess depression [33] and anxiety [34]. acquisition of the Health Education Impact Questionnaire
The PHQ-2 uses the first two items of the PHQ-9 (“little [40]. The items are measured on a four-point scale (1 =
interest or pleasure in doing things,” “feeling down, strongly disagree, 2 = disagree, 3 = agree, 4 = strongly
depressed or hopeless”) while the GAD-2 uses the first agree). The scale score is the unweighted mean of all
two items of the GAD-7 (“feeling nervous, anxious or on items, with higher values indicating higher subjective self-
edge", “not being able to stop or control worrying”). All management skills.
items are measured on a four-point scale (0 = not at all,
1 = several days, 2 = more than half of the days, 3 = nearly Employment To cover participation in working life the
all days). Sum scores for depression and anxiety range employment status (employed vs. unemployed) is asked for.
from zero to six points. Additionally, we will determine Moreover, we will assess if the patients are on sick leave.
binary outcomes by categorizing values of >2 as clinically
relevant depressive or anxiety disorder. Subjective prognosis of employment status Three items
assess if patients believe they will remain at work until
Pain intensity and disability Six items from the Graded retirement, if patients assume that their health will be
Chronic Pain Scale will be used to assess pain intensity permanently jeopardized, and if patients intend to sub-
and pain disability [35, 36]. Pain intensity items cover mit a request for a disability pension [41]. Items are
current pain and worst pain in the last four weeks and summarized to reflect a total score ranging from zero to
average pain in the last four weeks. Pain disability is re- three points. Higher values indicate a more unfavorable
lated to interference with daily activities, recreational, outcome. A score of at least two points reflects that
social and family activities and work activities. All items permanent work participation is deemed to be unlikely.
are scored from zero to 10 points, with higher scores in-
dicating stronger pain or more disability. Items will be Perceived vocational orientation of the rehabilitation
averaged and multiplied by 10. Thus, the composite scores program The realization of work-related goals and ther-
range from zero to 100 points. apies during rehabilitation will be assessed with a slightly
modified version of a previously used set of items from a
Fear of movement Two items of the Fear-Avoidance study that investigated the implementation of the WMR
Beliefs Questionnaire [37] (“My pain was caused by phy- guideline [19]. Participants report on 12 dichotomized
sical activity,” “I should not do physical activities which items as to whether they received WMR contents through-
might make my pain worse”) will be used to assess fear of out their rehabilitation program. Scores are aggregated
movement. Like Kent and colleagues [38], we will use to a total score ranging from zero to 12 points. This score
Neuderth et al. BMC Public Health (2016) 16:804 Page 8 of 11

reflects the implementation of the work-related therapies. a risk of not returning to work and a need for WMR if a
Additionally, six items assess the perceived diagnostic and person is unemployed at the beginning of the rehabili-
therapeutic focus on issues of return to work and work tation program or scored one out of three points on a
ability, e.g. the experience of a consistent return to work three-item scale that assesses negative return-to-work
strategy. These items are measured on a five-point scale. expectations.
Scores will be summed to a total score ranging from zero
to 24 points. Finally, the subjective work-related benefit Somatization Somatization will be assessed by using
from participating in the rehabilitation program will be seven items from the Symptom Check-List-90-R [43]. All
assessed by eight items measured on a five-point scale. items are measured on a five-point scale (0 = not at all,
Scores will be aggregated to a total score ranging from 1 = a little bit, 2 = moderately, 3 = quite a bit, 4 = extremely).
zero to 32 points. Items are averaged to calculate a total score.

Treatment satisfaction Treatment satisfaction will be Pain generalization Three newly developed items will
assessed using the German version of the Client Satisfac- assess the experience of widespread pain and pain ampli-
tion Questionnaire [42]. This questionnaire includes eight fication. These items are measured on a four-point scale
items designed to assess various aspects of the patient’s (0 = totally disagree, 1 = disagree, 2 = agree, 3 = totally
satisfaction with the treatment. Items are measured on a agree). Items are averaged to calculate a total score.
four-point scale. The sum score ranges from eight to 32
points. Psychosocial stress Two items will assess family- and
job-related stress in the last two weeks [44]. Both items
Income and welfare benefits Income from regular are measured on a four-point scale (0 = not at all, 1 =
employment and the duration of welfare benefits (for several days, 2 = more than half of the days, 3 = nearly all
example, unemployment and sickness benefit) will be days). Items are added to a sum score ranging from zero
extracted from the GPI accounts. to six points.

Therapeutic interventions Work environment Several aspects of the work envi-

Therapeutic interventions will be extracted from the ronment will be assessed as they might moderate the
standardized rehabilitation discharge letters [19]. These treatment effect: amongst others, size of company [45],
letters are stored in the individual GPI accounts. The physical job demands [46], effort-reward imbalance [47],
documentation of the therapeutic interventions will indi- overcommitment [47] and support by co-workers and
cate adherence to the WMR guideline. supervisors.

Department characteristics Socio-demographic data We will ask participants for

Measurement of department characteristics covers amongst socio-demographic data (age, sex, native language, educa-
others data on average guideline adherence and the number tional level, partnership, children, and family life).
of treated patients. A survey of the departments will assess
additional data on the department level (e.g. staff, team Statistical analysis
cooperation, and infrastructure). Propensity score matching
Patients of WMR and CMR will be matched by propen-
Other measures sity scores to achieve balanced sample characteristics and
Additional measures will be assessed as potential effect to calculate unbiased estimates of relative effectiveness
modifiers and as variables for calculating the propensity [22, 48–54]. The propensity score is the conditional prob-
score. ability of receiving the treatment under evaluation (i.e.
WMR) given the vector of observed background variables.
Risk of not returning to work Two risk scores are Matching by propensity scores enables balanced charac-
frequently used to assess the need for WMR in Germany. teristics of the treated and the untreated sample if there is
The score of the SIMBO (German: Screening-Instrument sufficient overlap of the propensity scores of both groups.
zur Feststellung des Bedarfs an medizinisch-beruflich Compared with a conventional direct matching procedure
orientierten Maßnahmen in der medizinischen Rehabilita- the problem of multidimensionality in finding a corre-
tion) [11] ranges from zero to 100 points. Higher scores sponding control, for instance related to age, sex, sickness
indicate a higher risk of not returning to work and a absence duration, pain, depression and others, is thereby
stronger need for WMR. A SIMBO score of ≥23 points reduced to one dimension only.
was shown to be an optimal threshold to identify patients The propensity score will be estimated by a logistic
in need for WMR. The Würzburger Screening [12] states regression model including all variables that are associated
Neuderth et al. BMC Public Health (2016) 16:804 Page 9 of 11

with the treatment allocation. For every WMR participant Discussion

a patient with a similar propensity score will be selected This study will provide an estimation of the relative effect-
from the larger pool of CMR patients. Resampling will be iveness of a complex, newly implemented work-related
realized without replacement. If necessary to achieve rehabilitation strategy for patients with musculoskeletal
balanced data, a caliper of one-quarter of the propensity disorders. Findings will complement the existing evidence
score during resampling will be used. For sensitivity of the relative efficacy derived from randomized controlled
analysis, additional matching schemes will be tested (with trials by robust estimation of the effects under real-life
and without replacement, different calipers, varied number conditions of rehabilitation service provision in Germany.
of controls). As an indicator of the bias before and after The findings of this study will be published in peer-
matching owed to differences related to the observed reviewed journal articles and conference presentations.
sample characteristics the standardized percentage bias will
be calculated. This is the difference of the sample means in Trial status
cases and controls relative to the square root of the average Recruitment has started and is ongoing.
of the sample variances in both groups [55]. Multiple impu-
Abbreviations
tations will be used to fill in missing data before estimation CMR, conventional medical rehabilitation; GAD-2, Generalized Anxiety
of the propensity score. The propensity scores for each Disorder Questionnaire; GPI, German Pension Insurance; IRES-24, Indikatoren
record will be averaged across the completed datasets, and des Reha-Status; PHQ-2, Patient Health Questionnaire; WAI, Work Ability
Index; WAS, Work Ability Score; WMR, work-related medical rehabilitation
propensity score matching will be performed with these
averaged scores [56]. Acknowledgements
Not applicable.

Funding
Multilevel regression analyses The study is funded by the Federal German Pension Insurance,
Analyses of treatment effects in propensity score matched Hohenzollerndamm 46/47, 10713 Berlin, Germany (grant number:
samples can use the same statistical methods as those FV-1253-15-0483-01). Funding covers personnel, material and traveling
expenses. The Federal German Pension Insurance additionally contributes
used in experimental studies [48, 49]. Multilevel regres- by sending out study materials and patient questionnaires and by providing
sion analyses will be used to account for dependencies in administrative data on work participation and rehabilitation services of study
the data [57, 58]. Individual patient data will be concep- participants. Anonymous data processing and data analyses are guaranteed.

tualized as level-1 parameters and the rehabilitation de- Availability of data and material
partment as a level-2 variable. All models will include the Not applicable.
treatment variable (WMR vs. CMR), the baseline score of
Authors’ contributions
the respective outcome measure and a random intercept SN, MS, BS and MB developed the study design. All authors contributed
which reflects the rehabilitation department. For the pri- to drafting the manuscript. All authors finally read and approved the
mary outcomes, p-value will be fixed at 0.016 (Bonferroni manuscript for submission. All authors fulfill the authorship criteria of the
International Committee of Medical Journal Editors.
corrected for three outcomes). For the secondary out-
comes, p-value will be fixed at 0.05. Linear models will be Competing interests
used for continuous outcomes and logistic models for The authors declare that they have no competing interests.
binary outcomes.
Consent for publication
Not applicable.

Moderator analyses Ethics approval and consent to participate

The investigation conforms to the principles outlined in the Declaration of
Exploratory moderator analyses will be performed to Helsinki. The study protocol was approved by the Ethics Committee of the
clarify whether level-1 or level-2 characteristics modify the University of Lübeck (15-356), and participation in the study is voluntary.
treatment effect. The modeling of moderator effects of Eligible patients will receive written information on study aims, participation
and the right to refuse. The study has been registered on the German
level-1 characteristics will be realized by multiplicative Clinical Trials Register (DRKS00009780).
interaction terms of the treatment indicator and potential This paper contains the original study protocol. Any substantial modifications
moderators. Moderator effects of level-2 variables will be to the study protocol will be notified to the Ethics Committee of the
University of Lübeck for approval prior to implementation. These
tested by including the potential moderator as a level-2 amendments will be documented in detail in the German Clinical Trials
predictor of the level-1 treatment-effect. In the case Register and will be described transparently in trial reports.
of continuous moderators, these variables will be z-
Author details
standardized [59]. Level-1 moderators will be tested 1
University of Applied Sciences Würzburg-Schweinfurt, Faculty of Applied
in order to identify patients who particularly benefit Social Sciences, Münzstraße 12, Würzburg 97070, Germany. 2Institute of
from WMR. Level-2 moderators may suggest depart- Social Medicine and Epidemiology, University of Lübeck, Ratzeburger Allee
160, Lübeck 23562, Germany. 3Department of Medical Psychology and
ment characteristics that may impact the effectiveness Psychotherapy, Medical Sociology and Rehabilitation Sciences, University of
of WMR. Würzburg, Klinikstraße 3, Würzburg 97070, Germany.
Neuderth et al. BMC Public Health (2016) 16:804 Page 10 of 11

Received: 16 July 2016 Accepted: 3 August 2016 23. Chan AW, Tetzlaff JM, Gotzsche PC, Altman DG, Mann H, Berlin JA, Dickersin K,
Hrobjartsson A, Schulz KF, Parulekar WR, et al. SPIRIT 2013 explanation and
elaboration: guidance for protocols of clinical trials. BMJ. 2013;346:e7586.
24. Zoer I, de Graaf L, Kuijer PP, Prinzie P, Hoozemans MJ, Frings-Dresen MH.
References Matching work capacities and demands at job placement in employees
1. Eurostat. Disability statistics - prevalence and demographics. 2016. with disabilities. Work. 2012;42:205–14.
http://ec.europa.eu/eurostat/statistics-explained/index.php/Disability_ 25. Toeppen-Sprigg B. Importance of job analysis with functional capacity
statistics_-_prevalence_and_demographics. Accessed 10 Aug 2016. matching in medical case management: a physician’s perspective.
2. Loisel P, Buchbinder R, Hazard R, Keller R, Scheel I, van Tulder M, Webster B. Work. 2000;15:133–7.
Prevention of work disability due to musculoskeletal disorders: the 26. Voss M, Schutte M, Wieland K. Synopsis of procedures used in the
challenge of implementing evidence. J Occup Rehabil. 2005;15:507–24. evaluation of the physically handicapped and work demands.
3. Schaafsma FG, Whelan K, van der Beek AJ, van der Es-Lambeek LC, Ojajarvi Rehabilitation. 1988;27:135–9.
A, Verbeek JH. Physical conditioning as part of a return to work strategy to 27. Kuijer PP, Gouttebarge V, Wind H, van Duivenbooden C, Sluiter JK,
reduce sickness absence for workers with back pain. Cochrane Database Frings-Dresen MH. Prognostic value of self-reported work ability and
Syst Rev. 2013;8:Cd001822. performance-based lifting tests for sustainable return to work among
4. Kamper SJ, Apeldoorn AT, Chiarotto A, Smeets RJ, Ostelo RW, Guzman J, construction workers. Scand J Work Environ Health. 2012;38:600–3.
van Tulder MW. Multidisciplinary biopsychosocial rehabilitation for chronic 28. Ilmarinen J. The Work Ability Index (WAI). Occup Med (Lond). 2007;57:160.
low back pain. Cochrane Database Syst Rev. 2014;9:Cd000963. 29. El Fassi M, Bocquet V, Majery N, Lair ML, Couffignal S, Mairiaux P. Work
5. van Vilsteren M, van Oostrom SH, de Vet HC, Franche RL, Boot CR, Anema ability assessment in a worker population: comparison and determinants of
JR. Workplace interventions to prevent work disability in workers on sick Work Ability Index and Work Ability Score. BMC Public Health. 2013;13:305.
leave. Cochrane Database Syst Rev. 2015;10:Cd006955. 30. Kristensen TS, Hannerz H, Hogh A, Borg V. The Copenhagen Psychosocial
6. Brüggemann S. The Rehabilitation Guideline Program of the BfA. Z Arztl Questionnaire–a tool for the assessment and improvement of the psychosocial
Fortbild Qualitatssich. 2005;99:47–50. work environment. Scand J Work Environ Health. 2005;31:438–49.
7. Hüppe A, Raspe H. Efficacy of inpatient rehabilitation for chronic back pain 31. Nuebling M, Hasselhorn HM. The Copenhagen Psychosocial Questionnaire
in Germany: a systematic review 1980–2001. Rehabilitation. 2003;42:143–54. in Germany: from the validation of the instrument to the formation of a
8. Hüppe A, Raspe H. Efficacy of inpatient rehabilitation for chronic back pain job-specific database of psychosocial factors at work. Scand J Public Health.
in Germany: update of a systematic review. Rehabilitation. 2005;44:24–33. 2010;38:120–4.
9. Hüppe A, Glaser-Möller N, Raspe H. Offering multidisciplinary medical 32. Wirtz M, Farin E, Bengel J, Jäckel WH, Hammerer D, Gerdes N. IRES-24
rehabilitation to workers with work disability due to musculoskeletal patient questionnaire: development of the short form of an assessment
disorders: results of randomized controlled trial. Gesundheitswesen. instrument in rehabilitation by means Mixed-Rasch analysis. Diagnostica.
2006;68:347–56. 2005;51:75–87.
10. Bethge M. Success factors of work-related orthopaedic rehabilitation. 33. Lowe B, Kroenke K, Grafe K. Detecting and monitoring depression with a
Rehabilitation. 2011;50:145–51. two-item questionnaire (PHQ-2). J Psychosom Res. 2005;58:163–71.
11. Streibelt M, Bethge M. Prospective cohort analysis of the predictive validity 34. Kroenke K, Spitzer RL, Williams JB, Lowe B. The Patient Health Questionnaire
of a screening instrument for severe restrictions of work ability in patients Somatic, Anxiety, and Depressive Symptom Scales: a systematic review.
with musculoskeletal disorders. Am J Phys Med Rehabil. 2015;94:617–26. Gen Hosp Psychiatry. 2010;32:345–59.
12. Lukasczik M, Wolf HD, Gerlich C, Löffler S, Vogel H, Faller H, Neuderth S. 35. Von Korff M, Ormel J, Keefe FJ, Dworkin SF. Grading the severity of chronic
Current state of vocationally oriented medical rehabilitation - a German pain. Pain. 1992;50:133–49.
perspective. Disabil Rehabil. 2011;33:2646–55. 36. Klasen BW, Hallner D, Schaub C, Willburger R, Hasenbring M. Validation and
13. Streibelt M, Buschmann-Steinhage R. A profile of requirements for the reliability of the German version of the Chronic Pain Grade questionnaire in
performance of work related medical rehabilitation from the perspective of primary care back pain patients. Psychosoc Med. 2004;1:Doc07.
the statutory pension insurance. Rehabilitation. 2011;50:160–7. 37. Pfingsten M, Kroner-Herwig B, Leibing E, Kronshage U, Hildebrandt J.
14. Bethge M, Herbold D, Trowitzsch L, Jacobi C. Work status and health-related Validation of the German version of the Fear-Avoidance Beliefs
quality of life following multimodal work hardening: a cluster randomised Questionnaire (FABQ). Eur J Pain. 2000;4:259–66.
trial. J Back Musculoskelet Rehabil. 2011;24:161–72. 38. Kent P, Mirkhil S, Keating J, Buchbinder R, Manniche C, Albert HB. The
15. Kleist B, Alliger K, Winter S, Beyer WF. Work Hardening bei chronischen concurrent validity of brief screening questions for anxiety, depression,
unspezifischen Rückenschmerzen in der stationären Rehabilitation. Praxis social isolation, catastrophization, and fear of movement in people with
Klin Verhaltensmed Rehab. 2001;14:145–50. low back pain. Clin J Pain. 2014;30:479–89.
16. Streibelt M, Bethge M. Effects of intensified work-related multidisciplinary 39. Godin G, Shephard RJ. A simple method to assess exercise behavior in the
rehabilitation on occupational participation: a randomized-controlled trial in community. Can J Appl Sport Sci. 1985;10:141–6.
patients with chronic musculoskeletal disorders. Int J Rehabil Res. 2014;37:61–6. 40. Osborne RH, Elsworth GR, Whitfield K. The Health Education Impact
17. Bethge M, Neuderth S. Medizinisch-berufliche Maßnahmen. In: Bengel J, Questionnaire (heiQ): an outcomes and evaluation measure for patient
Mittag O, editors. Psychologie in der medizinischen Rehabilitation. Ein education and self-management interventions for people with chronic
Lehr- und Praxishandbuch. Heidelberg: Springer; 2016. conditions. Patient Educ Couns. 2007;66:192–201.
18. Deutsche Rentenversicherung Bund. Anforderungsprofil zur Durchführung 41. Mittag O, Raspe H. A brief scale for measuring subjective prognosis of
der Medizinisch-beruflich orientierten Rehabilitation (MBOR) im Auftrag der gainful employment: findings of a study of 4279 statutory pension insurees
Deutschen Rentenversicherung. 3., überarbeitete Auflage. Berlin: Deutsche concerning reliability (Guttman Scaling) and validity of the scale.
Rentenversicherung Bund; 2012. Rehabilitation. 2003;42:169–74.
19. Bethge M, Löffler S, Schwarz B, Vogel H, Schwarze M, Neuderth S. Is the 42. Attkisson CC, Zwick R. The client satisfaction questionnaire. Psychometric
guideline for work-related medical rehabilitation successfully implemented? properties and correlations with service utilization and psychotherapy
Rehabilitation. 2014;53:184–90. outcome. Eval Program Plann. 1982;5:233–7.
20. Bethge M, Löffler S, Schwarz B, Vogel H, Schwarze M, Neuderth S. Specific 43. Schmitz N, Hartkamp N, Kiuse J, Franke GH, Reister G, Tress W. The
work-related problems: do they matter in access to work-related medical Symptom Check-List-90-R (SCL-90-R): a German validation study. Qual Life
rehabilitation? Rehabilitation. 2014;53:49–55. Res. 2000;9:185–93.
21. Glasgow RE, Lichtenstein E, Marcus AC. Why don’t we see more 44. Küch D, Arndt S, Grabe A, Manthey W, Schwabe M, Fischer D. UKS –
translation of health promotion research to practice? Rethinking the Ultrakurzscreening psychosozialer Problemlagen zur bedarfsorientierten
efficacy-to-effectiveness transition. Am J Public Health. 2003;93:1261–7. Angebotszuweisung in der somatischen Rehabilitation. In: Arbeitskreis
22. Schelvis RM, Oude Hengel KM, Burdorf A, Blatter BM, Strijk JE, van der Beek Klinische Psychologie in der Rehabilitation Fachgruppe der Sektion Klinische
AJ. Evaluation of occupational health interventions using a randomized Psychologie im Berufsverband Deutscher Psychologinnen und Psychologen
controlled trial: challenges and alternative research designs. Scand J Work (BDP) e. V., editors. Psychologische Betreuung im Krankheitsverlauf. Berlin:
Environ Health. 2015;41:491–503. Deutscher Psychologen Verlag; 2011. p. 106.
Neuderth et al. BMC Public Health (2016) 16:804 Page 11 of 11

45. European Commission. User guide to the SME definition. Luxembourg:

Publications Office of the European Union; 2015.
46. Bethge M, Spanier K, Neugebauer T, Mohnberg I, Radoschewski FM.
Self-reported poor work ability-an indicator of need for rehabilitation?
A cross-sectional study of a sample of German employees. Am J Phys
Med Rehabil. 2015;94:958–66.
47. Siegrist J, Wege N, Puhlhofer F, Wahrendorf M. A short generic measure
of work stress in the era of globalization: effort-reward imbalance.
Int Arch Occup Environ Health. 2009;82:1005–13.
48. Guo S, Fraser MW. Propensity score analysis: statistical methods and
applications. Los Angeles: Sage Publications; 2010.
49. Rosenbaum PR, Rubin DB. The central role of the propensity score in
observational studies for causal effects. Biometrika. 1983;70:41–55.
50. Dehejia RH, Wahba S. Propensity score-matching methods for
nonexperimental causal studies. Rev Econ Stat. 2002;84:151–61.
51. Descatha A, Leclerc A, Herquelot E. Use of propensity scores in occupational
health? J Occup Environ Med. 2013;55:477–8.
52. Saltychev M, Laimi K, El-Metwally A, Oksanen T, Pentti J, Virtanen M,
Kivimaki M, Vahtera J. Effectiveness of multidisciplinary primary prevention
in decreasing the risk of work disability in a low-risk population. Scand J
Work Environ Health. 2012;38:27–37.
53. Saltychev M, Laimi K, Oksanen T, Pentti J, Kivimaki M, Vahtera J. Does
perceived work ability improve after a multidisciplinary preventive program
in a population with no severe medical problems - the Finnish Public
Sector Study. Scand J Work Environ Health. 2013;39:57–65.
54. Bethge M. Effects of graded return-to-work: a propensity-score-matched
analysis. Scand J Work Environ Health. 2016;42:273–9.
55. Rosenbaum PR, Rubin DB. Constructing a control group using multivariate
matched sampling methods that incorporate the propensity score.
Am Stat. 1985;39:33–8.
56. Mitra R, Reiter JP. A comparison of two methods of estimating propensity
scores after multiple imputation. Stat Methods Med Res. 2016;25:188–204.
57. Kreft I, Leeuw J. Introducing multilevel modeling. London: Sage; 2006.
58. Donner A, Klar N. Design and analysis of cluster randomization trials in
health research. London: Arnold; 2000.
59. Smeets RJ, Maher CG, Nicholas MK, Refshauge KM, Herbert RD. Do
psychological characteristics predict response to exercise and advice for
subacute low back pain? Arthritis Rheum. 2009;61:1202–9.

Submit your next manuscript to BioMed Central

and we will help you at every step:
• We accept pre-submission inquiries
• Our selector tool helps you to find the most relevant journal
• We provide round the clock customer support
• Convenient online submission
• Thorough peer review
• Inclusion in PubMed and all major indexing services
• Maximum visibility for your research

Submit your manuscript at

www.biomedcentral.com/submit
 Reproduced with permission of copyright owner.
Further reproduction prohibited without permission.