Lorbeer et al.

BMC Musculoskeletal Disorders (2023) 24:550 BMC Musculoskeletal

https://doi.org/10.1186/s12891-023-06661-x
Disorders

RESEARCH Open Access

Enhancing physical activity and reducing

symptoms of patients with osteoarthritis
of the knee: a randomized controlled trial
of the PrevOP‑Psychological Adherence
Program
Noemi Lorbeer1†, Nina Knoll1*†, Jan Keller1, Antonia Domke1, Sally Di Maio1, Gabriele Armbrecht2,
Hendrikje Börst2, Peter Martus3, Wolfgang Ertel4 and Ralf Schwarzer1,5

Abstract
Background This primary analysis evaluated the “PREVenting the impairment of primary Osteoarthritis by high-
impact long-term Physical exercise regimen—Psychological Adherence Program” (PrevOP-PAP), designed to support
patients with osteoarthritis of the knee (OAK) to engage in regular moderate-to-vigorous physical activity (MVPA) to
reduce OAK symptoms (WOMAC scores). Theory-based on the health action process approach (HAPA), the interven-
tion targeted volitional precursors of MVPA change: action and coping planning, maintenance and recovery self-
efficacy, action control, and social network formation. We hypothesized that compared to an active control condition,
increases in MVPA at the end of the 12-month intervention would translate into lower WOMAC scores at 24 months in
the intervention condition.
Methods Participants with radiographically verified moderate OAK (N = 241; 62.66% female; M(SD) = 65.60(7.61)
years) were randomly assigned to the intervention (51%) or the active control condition. WOMAC scores (24 months)
were the primary -, accelerometer-assessed MVPA (12 months) the key secondary outcomes. The PrevOP-PAP was a
12-month intervention with computer-assisted face-to-face and phone-based sessions designed to increase HAPA-
proposed volitional precursors of MVPA change (up to 24 months; secondary outcomes). Intent-to-treat analyses
included multiple regression and manifest path models.
Results MVPA (12 months) did not mediate effects of the PrevOP-PAP on WOMAC scores (24 months). Compared
to the active control condition, WOMAC scores (24 months) were lower in the intervention condition, but this
effect did not remain stable in sensitivity analyses (b(SE) = -8.41(4.66), 95%-CI [-17.53; 0.71]). However, explora-
tory analyses revealed significantly stronger reductions in WOMAC-pain (24 months) in the intervention condition
(b(SE) = -2.99(1.18), 95%-CI [-5.36; -0.63]). Groups did not differ in MVPA at 12 months (b(SE) = -3.78(3.42), 95%-CI

†
Noemi Lorbeer and Nina Knoll shared first authorship.
*Correspondence:
Nina Knoll
[email protected]
Full list of author information is available at the end of the article

© The Author(s) 2023. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which
permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the
original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or
other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line
to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory
regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this
licence, visit http://​creat​iveco​mmons.​org/​licen​ses/​by/4.​0/. The Creative Commons Public Domain Dedication waiver (http://​creat​iveco​
mmons.​org/​publi​cdoma​in/​zero/1.​0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data.
Lorbeer et al. BMC Musculoskeletal Disorders (2023) 24:550 Page 2 of 18

[-10.80; 2.58]). Of the proposed precursors of MVPA change, action planning was higher in the intervention than in the
control condition (24 months; b(SE) = 0.64(0.26), 95%-CI [0.14; 1.15]).
Conclusions Compared to an active control condition, the PrevOP-PAP did not produce reliable effects on WOMAC
scores and none on preceding MVPA. Of the HAPA-proposed volitional precursors, only action planning was sustain-
ably increased. Future interventions should use m-health applications to digitally support long-term changes in
proposed volitional precursors of MVPA change.
Trial registration German Clinical Trials Register; https://​drks.​de/​search/​de/​trial/​DRKS0​00096​77; also available at
http://​apps.​who.​int/​trial​search/; registration number: DRKS00009677; date of registration: 26/01/2016.
Keywords Knee osteoarthritis, Osteoarthritis symptoms, Physical activity, Accelerometer, Planning, Action control,
Health action process approach, RCT​

Background proposed to predict intention formation towards behav-

Osteoarthritis of the knee (OAK) is a highly prevalent, ior change [14]. The subsequent volitional phase includes
progressive, age-related disease that causes pain and further predictors to bridge the intention-behavior gap
stiffness of the affected joint, causing reductions in indi- including action planning to determine when, where, and
viduals’ quality of life [1–4]. Recent estimations suggest how to perform the recommended behavior [14, 15, 20];
that worldwide over 650 million persons over the age of coping planning to identify barriers for regular PA and
40 suffer from OAK [2]. Conservative treatment of OAK to prepare adequate coping strategies to deal with them
includes use of nonsteroidal anti-inflammatory drugs and [21–23]; and action control, where individuals moni-
different forms of physical activity (PA) [4, 5]. Guidelines tor the progress and deviations from their PA goals and
recommend at least 150 min per week of moderate physi- engage in regulatory efforts if their current behavior does
cal activity (MPA) or 75 min per week of vigorous physi- not meet these goals [23, 24]. Two further volitional types
cal activity (VPA), or a combination of both (MVPA), of self-efficacy are proposed, i.e., maintenance self-effi-
with added muscle strength, flexibility, and balance cacy and recovery self-efficacy. Maintenance self-efficacy
training [6]. This randomized controlled trial (RCT), addresses the belief that one is competent to maintain
which was part of the “PREVenting the impairment of behavior change despite barriers. Recovery self-efficacy
primary Osteoarthritis by high-impact long-term Physi- addresses the belief in one’s competence to resume the
cal exercise regimen” (PrevOP) trial, was designed to behavior following lapses or phases of inaction [14, 25].
enhance patients’ adherence to regular MVPA in order The HAPA also considers contextual barriers (e.g., envi-
to reduce OAK symptoms (as measured by the Western ronmental conditions such as rainy weather) and facilita-
Ontario and McMaster Universities Osteoarthritis Index, tors to behavior change [15]. One such facilitating factor is
WOMAC [7]) [8]. the social network of the individual who wants to increase
Although patients with OAK report strong intentions PA [26]. Network members may assist in target persons’
to adapt their lifestyles to relieve severity or slow pro- behavior change via providing support to become active
gression of OAK symptoms [9–11], high levels of pain or engaging in PA together with them and thus provide
and other barriers challenge uptake and maintenance of an added social benefit or strengthened commitment to
recommended PA levels [12, 13]. The health action pro- the behavioral goal [27–31]. Intervention strategies that
cess approach (HAPA; [14–16]), a psychological model encourage the formation of collaborative implementa-
of health behavior change, proposes key cognitions and tion intentions (when, where, how, and how often are we
self-regulatory strategies to conquer challenges that per- going to be active together?) [32–34] were shown to help
sons with OAK face when attempting to perform regu- motivated persons to become more active together [34] or
lar PA. In this theory, behavior change is subdivided in reduce being inactive in their daily lives [33].
two phases: the motivational and volitional phases [14, There has been extensive research on PA in patients
17]. Key predictors of the motivational phase include with OAK, however, only few intervention programs
the following: Risk perception, i.e., perceived vulner- were theory-based [35–38]. Lack of theory-basis usu-
ability and severity of suffering from progression of a ally complicates the identification of active ingredients
disease if behavior is not changed [18]; outcome expec- in interventions and also impedes the accumulation of
tancies or pros and cons associated with the uptake and evidence on intervention efficacy. Consequently, inno-
maintenance of PA [19]; finally, task self-efficacy or the vative theory-based interventions and intervention
belief that one is competent to change a behavior, e.g., components are needed to facilitate the uptake and
regular PA [19]. These motivational precursors are then long-term maintenance of PA in patients with OAK.
Lorbeer et al. BMC Musculoskeletal Disorders (2023) 24:550 Page 3 of 18

In this trial, we used the HAPA as a theoretical frame- PA to reduce symptoms of OAK progression as part of
work [14–16]. Although HAPA-based interventions the PrevOP-PAP using a research design with multiple
have been designed and evaluated for different popu- assessments over an extended period of two years.
lations suffering from chronic diseases [15, 22, 39],
HAPA-based interventions for persons with OAK are Research question and hypotheses
still rare. As a notable exception, the ENHANCE trial In a population of individuals with moderate OAK, the
[38, 40] has evaluated a 12-week HAPA-based com- present study addresses the following primary research
bined counselling and exercise intervention in patients question: Is there an indirect effect of a HAPA-based
awaiting hip and knee arthroplasty, with the aim to psychological intervention (PrevOP-PAP intervention),
support patients’ uptake and maintenance of physi- consisting of a motivational intervention and a voli-
cal activity from pre- to up to 6 months post-surgery. tional intervention including network formation, on
Whereas both the intervention and usual care con- participants’ OAK symptoms (WOMAC) via MVPA
trol groups showed improvements in PA and OAK when compared to an active control condition receiving
symptoms in this time frame, authors did not find the motivational intervention only [8]? In the primary
between-group differences. Moreover, initial between- hypothesis we expected that compared to the PrevOP-
group differences in proposed psychological mediators PAP active control condition, participants receiving the
addressed by this intervention were not maintained at PrevOP-PAP intervention would report decreased OAK
6 months post-surgery [38]. This points out the need symptoms (WOMAC) at 24 months following entry
to extend the time of active intervention delivery by into the study and that this effect would be mediated
adding regular intervention boosters, a feature that by increased MVPA at the end of the active interven-
was implemented in the present PrevOP Psychologi- tion phase (12 months post study entry). By investigat-
cal Adherence Program (PrevOP-PAP) that delivered ing MVPA as a mediator, we thus aimed to elucidate the
several intervention booster sessions over the span of causal mechanism of a central active ingredient of the
12 months [8]. PrevOP-PAP intervention on the clinical outcome. In
Furthermore, most HAPA-based interventions so far additional exploratory follow-up analyses, we further
have addressed contextual facilitators or barriers indi- investigated intervention effects of the PrevOP-PAP inter-
rectly, for instance, as part of action or coping plan- vention via MVPA at 12 months on different domains of
ning strategies where participants identify (alone or OAK symptoms (WOMAC) at 24 months, i.e., WOMAC-
with others) good opportunities to act or barriers that functional limitations, WOMAC-pain, and WOMAC-
keep them from implementing the planned behavior stiffness. Secondary research questions and hypotheses
[15, 22, 32, 33, 39, 41]. Particularly for patients with addressed the predicted simple effects of the interven-
OAK, a direct and systematic intervention-aided setup tion on OAK symptoms (WOMAC) at 24 months, MVPA
of contextual facilitators, such as social network forma- at 12 months, as well as HAPA-proposed precursors
tion, a novel intervention component of the PrevOP- of change in MVPA at 24 months post study entry. Fol-
PAP, appears promising, especially at later points of lowing a brief motivational intervention delivered to all
the intervention-assisted behavior-change process participants, we assumed an overall increase in inten-
when intervention effects on self-regulation may start tion to engage in regular MVPA up to one week fol-
to decline [42]. To date, apart from general encourage- lowing the treatment. We further assumed differential
ment to seek social support if needed [38], an optional, long-term increases in HAPA-proposed volitional precur-
but systematic social network formation intervention sors of MVPA change that were directly addressed in the
that also encourages the formation of collaborative PrevOP-PAP intervention, that is higher 24-month levels
implementation intentions to become active together of action planning and coping planning, maintenance self-
with a chosen network member, has not been tested as efficacy and recovery self-efficacy, action control, and col-
part of HAPA-based intervention programs for patients laborative implementation intentions, as an indicator of
with OAK. social network formation, in participants of the PrevOP-
In addition, current RCTs with patients with OAK PAP intervention condition, when compared to those of
mainly focus on shorter-term effects [38, 43], whereas the PrevOP-PAP active control condition.
longer term assessment periods are needed to under-
stand causal mechanisms of complex interventions in
the long run. Consequently, we aimed to test the effec- Method
tiveness of HAPA-derived intervention strategies [22, 23, Procedure, randomization, and design
32–34, 44–46], including social network formation, in The PrevOP-PAP was an unblinded randomized con-
the context of the adoption and maintenance of regular trolled trial embedded in a parallel group design with
Lorbeer et al. BMC Musculoskeletal Disorders (2023) 24:550 Page 4 of 18

the PrevOP-PAP crossed within the same sample of study condition (51%) and all participants allocated to
the PrevOP-Main Medical Trial (PrevOP-MMT; pre- the PrevOP-PAP active control group were collapsed in
registered at [47], see below and additional file (Addi- another (49%), PrevOP-MMT condition allocation was
tional Figure 1). Thus, PrevOP-PAP and PrevOP-MMT controlled for (see below).
shared participants, inclusion and exclusion criteria, All participants received a brief motivational interven-
and the primary outcome of this study, OAK symptoms tion following baseline assessment (M0) prior to randomi-
(WOMAC) [8]. Both PrevOP-PAP and PrevOP-MMT zation (see below and [8] for intervention content). In the
trials followed randomized, controlled, prospective, PrevOP-PAP intervention condition, intervention periods
longitudinal designs. lasted 53 weeks [8]. A main computer-assisted face-to-face
The PrevOP-MMT tested a high-impact long-term intervention delivered by trained study personnel was con-
physical exercise regimen with resistive vibration exer- ducted one week following baseline (M0) at the main study
cise (PrevOP-MMT high-impact exercise condition) center. Four computer-assisted phone-based booster inter-
against a low-impact long-term exercise regimen with ventions took place at 3, 27, 50, and 52 weeks following
walking exercise (PrevOP-MMT low-impact exercise M0 to ensure intensive intervention delivery at the start
condition) and an unstructured, non-monitored exercise of PrevOP-PAP and booster sessions every six months fol-
control group (PrevOP-MMT active control condition). lowing the respective data assessments. Additional paper–
The PrevOP-MMT high-impact exercise condition and pencil activity calendar phases took place between week 1
PrevOP-MMT low-impact exercise condition received and week 4, between week 25 and week 28, and between
structured and monitored training for 12 months, which week 50 and week 53 following baseline (see below and [8]
was followed up by a home-based-mobility maintenance for intervention content). For computer-assisted phone-
program (see trial registration [47]). based interventions, participants were called at a location
The PrevOP-PAP was crossed with the PrevOP-MMT of their preference by trained study personnel. Paper–pen-
(see below and additional file; Additional Figure 1), where cil activity calendars were completed daily by participants
a randomly assigned 51% of the total sample received the for three periods of four weeks each. Participants received
HAPA-based psychological intervention (PrevOP-PAP travel cost reimbursement of EUR 5 per study center visit
intervention, see below) and 49% served as the active (assessment or intervention sessions).
control group (PrevOP-PAP active control condition, see In addition to six medical study visits with physi-
below). cal examinations and radiographic imaging as part of
Individuals interested in study participation were the PrevOP-MMT protocol (at baseline, 3, 6, 9, 12, and
informed about the study and screened for inclusion and 24 months), data were assessed from all participants
exclusion criteria in an initial telephone interview and at baseline (M0), 6 months (month “M”6), 12 months
during a medical examination by PrevOP-MMT medical (month “M”12), 18 months (month “M”18), and
personnel at the study center at Charité – Universitäts- 24 months (month “M”24) via self-report measures and
medizin Berlin. Prior to the medical examination, partici- three one-week accelerometer assessments of daily PA
pants provided written informed consent. (M0, M12, M24). Self-report measures were assessed via
Randomization of participants took place following paper–pencil questionnaire booklets at the main study
baseline assessment (month “M”0) and was conducted center (Charité – Universitätsmedizin Berlin, Germany;
at the Institute for Clinical Epidemiology and Applied M0, M6; M12; M24) or at participants’ homes (M18)
Biometry at Tübingen University Medical Center, using and returned directly to study personnel or mailed to the
computer-generated random numbers, stratified by sex. health pychology lab at Freie Universität Berlin. All data
Participants were fully informed about randomization were collected between February 2016 (first assessment)
procedures and randomly allocated to one of a total of six and January 2021 (last assessment). The present report
intervention constellations (see additional file, Additional used data relevant for examining the primary research
Figure 1): PrevOP-MMT high-impact exercise interven- question with assessments at M0, M12, and M24 [8].
tion (1) with the PrevOP-PAP intervention or (2) as part The ethics committee of the Charité – Universitäts-
of the PrevOP-PAP active control condition; the PrevOP- medizin Berlin approved this study (EA4/027/15). The
MMT low-impact exercise intervention (3) with the present primary analysis report complies with CON-
PrevOP-PAP intervention or (4) as part of the PrevOP- SORT guidelines and TIDieR guidelines [48, 49].
PAP active control condition; or the PrevOP-MMT con-
trol condition (5) with the PrevOP-PAP intervention or Power, recruitment, and inclusion
(6) as part of the PrevOP-PAP active control condition. For the PrevOP-PAP, with an alpha level of 0.05 and a sta-
For the purpose of analysis, all participants allocated bility factor of 0.68 of the measure to assess the primary
to the PrevOP-PAP intervention were collapsed in one outcome (OAK symptoms as measured by the WOMAC
Lorbeer et al. BMC Musculoskeletal Disorders (2023) 24:550 Page 5 of 18

Fig. 1 Consolidated Standards of Reporting Trials (CONSORT) diagram depicting participant flow through the study
Lorbeer et al. BMC Musculoskeletal Disorders (2023) 24:550 Page 6 of 18

[7]), a minimum sample size of n = 122 was determined Table 1 Sample characteristics at baseline (M0)
which included 2 groups at 5 points in time up to the PrevOP-PAP PrevOP-PAP active
primary endpoint to detect a small effect (f = 0.1) of a intervention control condition
within by between subjects factors interaction with a condition (n = 123) (n = 118)
power of 0.95. To detect the proposed indirect effect of Age in ­yearsa 65.46 (7.95) 65.75 (7.28)
the PrevOP-PAP intervention condition on OAK symp- Sex: ­Femaleb 77 (63) 74 (63)
toms (WOMAC) via MVPA with small-to-medium path Family ­statusb
coefficients (α = 0.26 and β = 0.26) and a power of 0.80 Married 73 (60) 57 (48)
using bias-corrected bootstrapping with 2,000 resamples, Committed relationship 15 (12) 14 (12)
a minimum sample size of n = 148 was determined [50].
Divorced 14 (12)c 29 (25)c
With an expected drop-out rate of 20%, the required
Single 11 (9) 16 (14)
sample size increased to n = 153 or n = 185, respectively.
Widowed 14 (12) 6 (5)
For the PrevOP-MMT a sample size of N = 240 had been
­ iplomab
High school d 73 (59) 62 (54)
determined [8]. Reactive recruitment strategies were
University ­degreeb 58 (48) 50 (42)
implemented throughout the greater Berlin (Germany)
Employedb 47 (38) 40 (34)
metropolitan area and included flyers, posters, social
Income (per month)b
media, press-releases, and regional and national news-
  < €750 10 (9) 7 (6)
published interviews on OAK with calls for participation.
€750 to < €1250 20 (17) 25 (22)
Proactive recruitment strategies included mailings via
€1,250 to < €2,000 33 (28) 34 (30)
local registration offices in the Berlin area (Germany) as
> €2,000 53 (46) 47 (42)
well as recruitment of patients from an ambulatory clinic
Childrenb 101 (83) 86 (74)
at the study center (Charité – Universitätsmedizin Ber-
Body mass ­indexa 28.45 (4.27) 28.61 (5.45)
lin). Patients were recruited between February 2016 and
Kellgren-Lawrence grade
November 2018.
Grade ­2b 44 (36) 41 (35)
Inclusion and exclusion criteria were mostly relevant
Grade ­3b 79 (64) 77 (65)
for the medical PrevOP-MMT and are listed in the addi-
Disease duration in years 11.41 (10.52) 11.64 (9.92)
tional file (Additional Information 1) [8]. (knee osteoarthritis)a
Comorbidityb 98 (81) 101 (86)
Sample 229 ≤ n ≤ 241 participants due to missing values. Kellgren-Lawrence grade
Of N = 243 persons with OAK enrolled in the crossed ranges from 0 (no knee osteoarthritis) to 4 (most severe knee osteoarthritis) [51]
PrevOP-PAP and PrevOP-MMT trials, N = 241 persons a
Values are mean (standard deviation)
were randomly allocated to study conditions and n = 194 b
Values are n (valid %, rounded)
took part in at least 80% of the intervention sessions c
Frequencies with the subscript c differ at p < .05 between the PrevOP-PAP
intervention condition and the PrevOP-PAP active control condition
(i.e., at least 80% intervention fidelity, as assessed by the
trained study staff ) or were part of the control group. On
average, participants of the PrevOP-PAP intervention programs in primary and tertiary prevention settings
condition attended 3.71 intervention sessions. At M24, [22, 23, 32–34, 44, 45], and adapted to patients with
data from n = 172 (71%) were available (see Fig. 1). The OAK in close collaboration with medical experts from
intent-to-treat sample was thus N = 241. For participant the field. As part of a two-week piloting phase, ten
characteristics, see Table 1. patients with OAK of the outpatient clinic of Charité
– Universitätsmedizin Berlin tested the intervention
Masking materials and provided feedback, which was subse-
PrevOP-PAP intervention content could not be masked quently used to further refine the intervention materi-
for study staff or participants. Study staff were aware of als. For a more detailed description of all intervention
participants’ study group allocation at the beginning of components and materials used, see [8]. The interven-
the first intervention session of the PrevOP-PAP. Moreo- tion procedures were applied exactly as specified in the
ver, data analyses were conducted by N.L., N.K., and R.S. study protocol [8] with no modifications to the inter-
and were also unmasked. vention during the course of the study. The intervention
materials can be made available by the corresponding
Intervention content author upon request. Intervention contents were deliv-
All intervention contents were delivered in German, ered by trained study staff (i.e., 2 to 3 Bachelor’s and
derived from theory-based established intervention Master’s students of psychology; employed as student
Lorbeer et al. BMC Musculoskeletal Disorders (2023) 24:550 Page 7 of 18

research assistants in the trial) who were provided PrevOP‑PAP intervention: computer‑assisted face‑to‑face
with training manuals and received training sessions intervention
on how to conduct the brief motivational intervention, The computer-assisted face-to-face intervention, again
the computer-assisted face-to-face, and the computer- delivered by trained study staff, consisted of an introduc-
assisted phone-based intervention sessions. PrevOP- tory section that reminded participants of the program’s
PAP study researchers monitored the trained student goals, four ensuing sections focussing on outcome expec-
research assistants’ intervention delivery throughout tancies, self-efficacy, goal setting, planning, and a feedback
the study. Computer-assisted face-to-face interven- section [8].
tion sessions were delivered at the main study center Outcome expectancies were addressed by providing
and computer-assisted phone-based interventions via participants with a calculated pros-cons difference score
phone. Intervention delivery was one-to-one, allowing of outcomes of regular PA. Participants first indicated
for interaction between study staff and participant. All how much they agreed with five positive (e.g., less joint
participants received the identical intervention with stiffness, good for overall health) and five negative (e.g.,
the optional network creation intervention as part pain during specific activities, too time-consuming) out-
of the third and fourth phone-based intervention (as come expectancy statements on 6-point Likert scales
described below). To prevent drop-out and maintain (not at all true to completely true). Then scores for pros,
intervention delivery, participants were reminded via cons, and a benefit expectation difference were fed back
mail or phone prior to their appointment (face-to-face to participants. In case of con scores being larger than
or phone-based intervention). If an appointment was pro scores, trained study staff reviewed concerns with
not kept by patients, they were contacted via phone to participants and asked them to think of activities asso-
reschedule the appointment. ciated with less cons (BCT: pros and cons (9.2); [54]).
Self-efficacy was addressed by asking for participants’ PA
Brief motivational intervention biographies and mastery experiences with PA through-
Before randomization, all participants received a brief out their life-span (BCTs: self-monitoring of behavior
motivational intervention delivered by trained study (2.3), identification of self as role model (13.1), identity
staff that consisted of a brochure that participants were associated with changed behavior (13.5); [54]). The goal
asked to read, followed by a brief quiz in form of a setting section started with reminders of OAK-specific
cross-word puzzle to test knowledge transfer. The bro- MVPA guidelines and joint-friendly activity examples [6,
chure introduced participants to different intensities 52, 53]. Then, testimonials were provided that depicted
of PA, providing examples of joint-friendly MVPA and a 61-year old man and a 68-year old woman describing
muscle-force strengthening exercises, and MVPA guide- their PA goal pursuits. Participants then recorded up to
lines for persons with OAK [6, 52, 53]. It also addressed five of their own PA goals (i.e., type of activity and dura-
all motivational HAPA constructs [14, 15]: (1) risk of tion), including new activities and those that they already
insufficient MVPA and evidence for consequences for performed (BCT: goal setting (behavior) (1.1); [54]).
OAK symptom progression (risk perception); (2) OAK- Participants’ PA goals were reiterated one-by-one dur-
specific and generic benefits of increasing MVPA along ing the planning sections and participants were asked
with commonly perceived negative outcomes (outcome to create action plans for their goals. Plans should be
expectancies); and (3) use of self-instruction, recall of phrased as “If/When…, then…” sentences with specific
prior mastery experiences of increasing MVPA, calls for cue-situations (If/When) connected to the PA goal-activ-
increasing MVPA in daily life, calls for thinking about ity (then). Participants then indicated on a 6-point scale
role models for MVPA in participants’ social networks (not at all true to completely true) their plan-execution
(PA-specific self-efficacy). In summary, intervention self-efficacy, named a start date and were asked to copy
strategies used in the brochure comprised the following their plans, as presented on the computer screen, into
behavior change techniques (BCTs): goal setting (behav- provided paper-pen activity calendars (see below). Sub-
ior) (1.1), social support (unspecified) (3.1), instruction sequently, each action plan was shown to participants
on how to perform the behavior (4.1), information about again and they were asked to generate a coping plan by
health consequences (5.1), information about social and identifying a potential barrier that may keep them from
environmental consequences (5.3), information about following through with their action plan (If/When-
emotional consequences (5.6), credible source (9.1), part of the coping plan) and specify how to manage this
social reward (10.4), focus on past success (15.3), self- barrier (Then-part of the coping plan; e.g., by perform-
talk (15.4), and vicarious consequences (16.3) [8, 54] (see ing a different activity or choosing another time/place)
[8] for a more detailed description of the motivational (BCTs: action planning (1.4); coping planning (1.2) [54]).
intervention). All plans were then shown to participants on summary
Lorbeer et al. BMC Musculoskeletal Disorders (2023) 24:550 Page 8 of 18

screens. A feedback section ended the computer-assisted were asked to use the self-regulatory strategies in their
face-to-face intervention. The computer-assisted face-to- daily lives, and were thanked.
face intervention session lasted 60 min (see [8] for a more The final component of the PrevOP-PAP intervention
detailed description of this part of the intervention). were paper–pencil activity calendars to promote action
control and maintenance self-efficacy as well as recovery
PrevOP‑PAP intervention: computer‑assisted phone‑based self-efficacy using BCTs self-monitoring of behavior (2.3),
intervention and activity calendars self-monitoring of outcome(s) of behavior (2.4), and feed-
Trained staff (i.e., trained Bachelor’s and Master’s stu- back on outcome(s) of behavior (2.7) [54]. Activity cal-
dents of psychology employed as students research assis- endars consisted of tables with columns for 7 days, with
tants in the trial) followed a computer-based structured each column sectioned to indicate morning, noon, and
intervention [8], designed to boost participants’ plan- evening times. During the guided intervention sessions
ning, self-efficacy, and action control concerning regular (face-to-face and phone-based), participants were asked
PA and recorded participants’ responses in the program to fill in the day and date (headers) and their current PA-
that provided the intervention contents produced by par- specific action plans (i.e., cues and behavior). The final
ticipants during the last session. two activity calendars also asked participants to indicate
To increase maintenance self-efficacy and recovery with whom they planned to pursue an activity. Calendars
self-efficacy, participants were first asked to review their were completed by participants daily at the end of each
PA-plan pursuit and indicate a success rate of implement- day throughout weeks 1 to 4, weeks 25 to 28, and weeks
ing their PA-specific plan enactment in percent. To do 50 to 53 following M0. For the activity-calendar peri-
so, participants used their completed activity calendars ods, participants were asked to put a checkmark next to
of the two weeks prior to the phone-based intervention. each action plan they had implemented as planned on a
Participants were then asked to recall positive experi- given day. At the bottom of the calendar-columns, par-
ences with implementing their PA plans during the past ticipants could enter additional and/or alternative activi-
two weeks. Following this, participants were given the ties pursued during a given day. Completed sheets were
opportunity to revise PA goals and associated action and sent back to the study center (see [8] for a more detailed
coping plans or add new ones, up to a maximum of five. description of these intervention components).
This was done in the same manner as in the computer-
assisted face-to-face intervention with interventionists Measures
recording and reading out the intervention content to The present article used data relevant for the analysis of
participants. If plans were kept, participants were asked the pre-registered primary research question under study
to rate their plan-execution self-efficacy anew. At the [8], these include M0, one week following M0, M12, and
end of this section, interventionists repeated each kept, M24.
altered, or new action plan aloud and asked participants
to fill them into a new set of activity calendars. A sum- OAK symptoms
mary print-out of all action and coping plans generated The primary endpoint was self-reported OAK symptoms
during the phone-based intervention session was also at M24 assessed with the WOMAC in its version for
sent to participants’ homes. OAK administered in German [7]. The WOMAC is a vali-
Phone-based interventions 3 and 4 had an additional dated and internationally used questionnaire which com-
optional component of network creation, when partici- prises 24 items with three subscales of OAK symptoms,
pants were encouraged to identify a sports companion, i.e., OAK functional limitations (17 items), OAK pain (5
contact them (phone-based intervention 3) and include items), and OAK stiffness (2 items), to which participants
these companions (i.e., their initials) into their action responded on 11-point scales ranging from 0 to 10. Item
plans, creating collaborative implementation intentions missings were imputed with item means and responses
(phone-based intervention 4) (BCTs: action planning were summed ranging from 0 to 240 (WOMAC total
(1.4), social support (practical) (3.2); [54]). If participants score), 0 to 170 (WOMAC-functional limitations), 0 to
preferred to be active without a companion, these sec- 50 (WOMAC-pain), and 0 to 20 (WOMAC-stiffness)
tions were skipped. [38]. Higher values indicated higher-levels of OAK symp-
Computer-assisted phone-based interventions lasted toms. M0 and M24 indicators were used in the present
between 20 and 60 min. At the end of each intervention analyses. Internal consistencies were medium to high,
session (face-to-face or phone-based) participants rated with Cronbach’s alphas α = 0.95 (M0) and α = 0.96 (M24)
the quality of the session, were asked if they had any for the overall score of OAK symptoms, α = 0.94 (M0)
questions, were reminded of the next study appointment, and α = 0.95 (M24) for WOMAC-functional limitations,
Lorbeer et al. BMC Musculoskeletal Disorders (2023) 24:550 Page 9 of 18

α = 0.79 (M0) and α = 0.88 (M24) for WOMAC-pain, and index (BMI; objectively assessed at M0), sex, and age
α = 0.80 (M0) and α = 0.86 (M24) for WOMAC-stiffness. which together with all other socio-demographic vari-
ables (Table 1) were assessed via self-report at M0. Unless
Moderate‑to‑vigorous physical activity (MVPA) 1-item assessments were used, items were averaged to a
Daily MVPA (in minutes) averaged over one week as total mean score ranging from 1 “does not apply at all/
assessed with tri-axial accelerometer devices (ActiGraph very unlikely” to 6 “applies exactly/highly likely”.
GT3X, Pensacola, Fl) at M0 and M12 were used in the
present analyses. Participants were instructed to wear Statistical analyses
the devices at their right hip during waking hours for one Analyses were conducted using R Statistical Software
week at each assessment period. Using an algorithm by (v4.2.2 [59]). Randomization checks and drop-out analy-
Sasaki et al. [55], minutes of MVPA were calculated for ses with all self-report and PA M0-assessments (see [8])
participants who had worn their accelerometers on at were done using analyses of variance for continuous vari-
least 4 days for at least 10 h a day. Univariate outliers of ables and chi-square tests for nominal and ordinal-scale
MVPA (z >|3.29|) were substituted by values one unit data. In case of several, potentially inter-related, rand-
higher/lower compared to the next most extreme value in omization or drop-out mechanisms, these were followed
the distribution [56]. up with logistic regression analyses predicting interven-
tion group membership (coded 1; active control group
HAPA‑proposed volitional precursors of MVPA change membership coded 0) or drop-out status, respectively, to
All HAPA-defined volitional precursors of MVPA change determine unique associations.
addressed in the PrevOP-PAP intervention were included To benefit from full-information maximum likelihood
in the manipulation checks, including assessments at M0 procedures to retain all available data in models and per-
and M24. They were adapted from prior research [22, form analyses with an intent-to-treat approach [60], all
23, 29, 41] and assessed specifically for the PA domain. other analyses were conducted as manifest models using
Participants responded on 6-point scales ranging from 1 the lavaan R package (v0.6–12 [61]). To examine inten-
“does not apply at all/very unlikely” to 6 “applies exactly/ tion change following the brief motivational treatment,
highly likely”. Action planning was assessed with 4 items a simple latent change score model, mimicking a paired
(M0 α = 0.97; M24 α = 0.98) and coping planning was samples t-test was conducted [62]. For manipulation
measured with 5 items (M0 α = 0.94; M24 α = 0.94). checks of the HAPA-proposed volitional precursors of
Maintenance self-efficacy (M0 α = 0.84; M24 α = 0.79) behavior change and to test simple effects of the PrevOP-
and recovery self-efficacy (M0 α = 0.93; M24 α = 0.92) PAP intervention group membership (coded 1; active
were measured with 3 items each. Action control was control group membership coded 0) on OAK symptoms
assessed with 6 items at M0 (α = 0.91) and M24 (α = 0.90). (WOMAC) at M24 as well as on MVPA at M12, mani-
Finally, collaborative implementation intentions with a fest regression analyses were fit, regressing the respective
training partner (adapted from [29, 41]) were assessed M24 (or M12) indicator on the intervention condition
with 4 items (M0 α = 0.98; M24 α = 0.99). For each as well as on its M0 counterpart. To test the primary
HAPA-defined construct, we computed mean scores hypothesis, a manifest path model was fit with interven-
ranging from 1 to 6. tion condition (PrevOP-PAP intervention group coded
1; active control group coded 0), M0 indicators of OAK
Behavioral intentions and covariates symptoms (WOMAC) and MVPA, and M0 covariates as
Behavioral intentions as assessed at M0 and one week predictors, MVPA at M12 as a proposed mediator, and
after the motivational treatment received by all par- OAK symptoms (WOMAC) at M24 as the outcome. To
ticipants were measured with 4 items (M0 α = 0.81; one test the predicted indirect effect, we used bias-corrected
week after M0 α = 0.76; [22, 23]). Covariates included bootstrapping with 5,000 resamples [63].
baseline variables (M0) for which randomization failed or Sensitivity analyses for the primary hypothesis test
those that were associated with dropout. These included included several groups of covariates: BMI (range in this
positive outcome expectancies (assessed with 6 items; sample: 19.16 to 45.79 kg/m2), sex (0 = female, 1 = male)
α = 0.82; [22, 23]), negative affect as a source of self-effi- and age in years; randomization failures (VAS-pain in the
cacy (assessed with 2 items; α = 0.92; e.g., “Just before I knee today; divorced: 0/1); dropout mechanisms (posi-
start physical activities, I feel tired” [8, 57]), a visual ana- tive outcome expectancies; negative affect as a source
logue scale of pain (VAS-pain, 10 cm) in the knee on of self-efficacy); and dummy-coded medical PrevOP-
the M0 day (ranging from 0: “no pain” to 10: “strongest MMT-conditions (PrevOP-MMT high-impact exercise
conceivable pain”) [58], and being divorced (one item), condition, coded 1, PrevOP-MMT low-impact exercise
as assessed at M0. Additional covariates were body mass condition, coded 1, with PrevOP-MMT active control
Lorbeer et al. BMC Musculoskeletal Disorders (2023) 24:550 Page 10 of 18

condition being the reference group, coded 0). Continu- recovery self-efficacy, action control, and collaborative
ous covariates were grand-mean centered. implementation intentions, emerged. As for collaborative
As indicated in the study protocol [8], in preliminary implementation intentions as an indicator of network
analyses, we ascertained that dummy-coded PrevOP- formation, which was optional, only few participants
MMT-conditions did not moderate the association (27% of the PrevOP-PAP intervention condition) actually
between the proposed mediator (MVPA at M12) and pri- chose to consider being physically active together with a
mary outcome (OAK symptoms (WOMAC) at M24). No network member. Sensitivity analyses including all fur-
interactions emerged, hence PrevOP-MMT-conditions ther covariates revealed the same pattern of results (addi-
were included as simple-effect covariates in all sensitiv- tional file: Additional Table 1).
ity analyses [8]. As models including all covariates did
not converge when using bias-corrected bootstrapping,
the Sobel test was used to test predicted indirect effects Indirect effects of the PrevOP‑PAP intervention on OAK
in sensitivity analyses [64]. All manifest path models symptoms via MVPA
were fully saturated; thus, no model fit indices could be Results of manifest path models predicting participants’
determined. OAK symptoms (WOMAC) at 24 months with MVPA at
12 months as a mediator are displayed in Table 3. Con-
Results trolling for M0 OAK symptoms (WOMAC) and MVPA,
Attrition analyses and randomization check no indirect effect of the PrevOP-PAP intervention on
Participants dropping out before M24 (n = 69) and con- participants’ OAK symptoms (WOMAC, at 24 months)
tinuing participants (n = 172) were similar in most of via MVPA at 12 months emerged, when compared to the
the variables under study. However, participants drop- PrevOP-PAP active control condition. At the end of the
ping out before M24 reported higher levels of positive active intervention phase (12 months post study entry)
outcome expectancies, t(238) = 2.31, p = 0.022, d = 0.33, and controlling for M0, participants in the PrevOP-PAP
and higher levels of negative affect as a source of self- intervention condition did not differ from those in the
efficacy, t(99.73) = 2.29, p = 0.024, d = 0.35, at baseline. PrevOP-PAP active control condition with regard to
Randomization checks indicated no significant baseline MVPA (Table 3; for multiple regression models testing
differences between the PrevOP-PAP intervention con- simple effects see additional file (Additional Table 1)).
dition and PrevOP-PAP active control condition, except At 24 months, participants in the PrevOP-PAP inter-
for levels of pain in the knee (VAS-pain) on the M0 day vention condition reported decreased levels of OAK
being lower in the PrevOP-PAP intervention condition, symptoms (WOMAC) when compared to participants
t(234) = -2.30, p = 0.022, d = 0.30, and more participants in the PrevOP-PAP active control condition (b = -9.81,
being divorced in the PrevOP-PAP active control condi- SE = 4.77, 95% CI [-19.51; -0.66], p = 0.040). However, in
tion than in the PrevOP-PAP intervention condition, χ2 sensitivity analyses this effect did not remain statistically
(1) = 7.00, p = 0.008, V = 0.17. significant (p = 0.071; see Model 2, Table 3; for multiple
regression models testing simple effects, see additional
Manipulation checks file, Additional Table 1).
Statistics of the central study variables and between- Exploratory follow-up analyses with different domains
group differences are displayed in Table 2. For fully con- of OAK symptoms (WOMAC-pain, WOMAC-func-
trolled multiple regression models testing simple effects tional limitations, WOMAC-stiffness) as outcomes and
of the PrevOP-PAP intervention condition (vs. PrevOP- controlling for M0 levels revealed that participants in the
PAP active control condition) see also additional file PrevOP-PAP intervention condition reported lower levels
(Additional Table 1). Manipulation checks revealed sig- of WOMAC-pain at 24 months post study entry when
nificant increases in participants’ intentions to engage compared to the PrevOP-PAP active control condition
in regular MVPA at one week following the brief moti- (additional file: Additional Table 1). Sensitivity analy-
vational intervention (b = 0.17, SE = 0.07, 95% CI [0.04; ses revealed the same pattern of results. With regard
0.30], p = 0.009). Controlling for M0, participants in the to WOMAC-functional limitations and WOMAC-
PrevOP-PAP intervention condition reported higher lev- stiffness, no intervention effects emerged. Again, and
els of action planning at M24 when compared to those consistent with results regarding overall OAK symp-
in the PrevOP-PAP active control condition (b = 0.64, toms (WOMAC), no indirect effects of PrevOP-PAP
SE = 0.26, 95% CI [0.14; 1.15], p = 0.013). However, no via MVPA at 12 months on WOMAC-pain, -functional
further group differences in long-term increases in limitations, or -stiffness were found (additional file: Addi-
HAPA-proposed volitional precursors of MVPA change, tional Tables 2, 3, and 4). All manifest path models were
including coping planning, maintenance self-efficacy and fully saturated.
Lorbeer et al. BMC Musculoskeletal Disorders (2023) 24:550 Page 11 of 18

Table 2 Descriptive statistics of central variables

PrevOP-PAP intervention condition PrevOP-PAP active control condition Difference between PrevOP-PAP
(n = 123) (n = 118) intervention and PrevOP-PAP active
control condition

Variable [scale’s range] M (SD) M (SD) Est (SE)

95% CI
Maintenance self-efficacy [1-6]
M0 4.28 (1.26) 4.55 (1.03)
M24 4.19 (1.16) 3.98 (1.07) 0.24 (0.17)
[-0.08; 0.57]
ΔR2 = .01
Recovery self-efficacy [1-6]
M0 4.88 (1.09) 5.13 (0.89)
M24 4.75 (1.12) 4.54 (0.97) 0.25 (0.16)
[-0.06; 0.56]
ΔR2 = .01
Action planning [1-6]
M0 3.96 (1.77) 3.99 (1.76)
M24 4.27 (1.73) 3.65 (1.87) 0.64 (0.26)*
[0.14; 1.15]
ΔR2 = .03
Coping planning [1-6]
M0 2.97 (1.58) 3.04 (1.59)
M24 3.10 (1.52) 3.20 (1.47) -0.10 (0.20)
[-0.50; 0.31]
ΔR2 = .00
Action control [1-6]
M0 3.22 (1.45) 3.38 (1.39)
M24 3.52 (1.34) 3.36 (1.33) 0.17 (0.18)
[-0.20; 0.53]
ΔR2 = .00
Collaborative implementation intentions [1-6]a
M0 3.19 (2.12) 3.29 (2.08)
M24 3.69 (2.10) 3.35 (2.12) 0.26 (0.44)
[-0.60; 1.12]
ΔR2 = .00
Moderate-to-vigorous physical activity (MVPA)
M0 46.44 (29.11) 48.42 (28.21)
M12 40.69 (25.83) 49.64 (29.25) -3.45 (3.11)
[-9.55; 2.65]
ΔR2 = .00
OAK Symptoms (WOMAC) [0–240]
M0 71.76 (33.35) 75.87 (39.28)
M24 43.23 (32.21) 54.96 (37.43) -9.30 (4.74)†
[-18.58; -0.02]
ΔR2 = .02
WOMAC-functional limitations [0–170]
M0 47.98 (24.79) 50.77 (29.66)
M24 29.10 (22.12) 37.01 (26.76) -6.32 (3.30)†
[-12.79; 0.15]
ΔR2 = .01
WOMAC-pain [0–50]
M0 15.79 (7.91) 16.52 (8.38)
M24 8.94 (7.87) 12.30 (9.49) -2.77 (1.17)*
[-5.05; -0.49]
ΔR2 = .02
Lorbeer et al. BMC Musculoskeletal Disorders (2023) 24:550 Page 12 of 18

Table 2 (continued)
PrevOP-PAP intervention condition PrevOP-PAP active control condition Difference between PrevOP-PAP
(n = 123) (n = 118) intervention and PrevOP-PAP active
control condition

WOMAC-stiffness [0–20]
M0 7.98 (4.50) 8.58 (4.53)
M24 5.19 (4.03) 5.66 (3.89) -0.28 (0.56)
[-1.39; 0.82]
ΔR2 = .00
147 ≤ n ≤ 241 due to missing values unless otherwise noted
M Mean, SD Standard deviation, Est. Estimate, SE Standard error, CI Confidence interval, OAK Osteoarthritis of the knee, M0 Baseline, M12 12-months follow-up, M24
24-months follow-up, PrevOP-PAP PrevOP-Psychological Adherence Program, WOMAC Western Ontario and McMaster Universities Osteoarthritis Index [7]
a
Based on n = 128 (M0) and n = 88 (M24) as many participants decided not to participate in network formation intervention and thus did not provide data. Between-
condition differences as indicated by manifest regression analyses regressing the respective M24 (or M12) indicator on its M0 counterpart and intervention condition
(PrevOP-PAP intervention condition, coded 1, vs. PrevOP-PAP active control condition, coded 0); coefficients are unstandardized. ΔR2 = Incremental variance explained
in the respective M24 (or M12) outcome, controlled for its M0 counterpart, when entering the intervention condition (PrevOP-PAP intervention condition, coded 1, vs.
PrevOP-PAP active control condition, coded 0) in the manifest regression model
†
p < .10; * p < .05

Discussion functional limitations and stiffness at 24 months following

This primary analysis report evaluated outcomes of the study entry. These findings resemble meta-analytic evidence
psychological adherence program PrevOP-PAP that was on self-management education programs for osteoarthritis
designed to enhance PA and reduce OAK symptoms suggesting small – mostly shorter-term – improvements
(WOMAC) among patients with moderate OAK. The in pain, but no beneficial effects on physical function-
intervention program PrevOP-PAP adopted motiva- ing when compared to control groups [38, 65]. However,
tional, volitional, and networking intervention strategies underlying intervention processes with regard to the effect
based on the HAPA to support OAK patients’ uptake and on WOMAC-pain remain unclear. As our findings indi-
maintenance of regular MVPA and reduce OAK symp- cated null effects of the PrevOP-PAP intervention on levels
toms (WOMAC). Intervention effects were contrasted of MVPA at the end of the intervention period, MVPA did
with the PrevOP-PAP active control condition, in which not serve as the proposed mediator. Similar findings with
participants only received the motivational intervention. positive effects of a motivational interviewing-based inter-
As the primary hypothesis, it was assumed that partici- vention on OAK symptoms, but null effects on MVPA have
pants of the PrevOP-PAP intervention condition (com- been reported elsewhere [66]. Possibly, participants of the
pared with participants of the PrevOP-PAP active control PrevOP-PAP intervention condition learned over time
condition) would engage in more MVPA at the end of the how to better accept their levels of pain, which has
active intervention phase which would then translate to been shown to be associated with lower levels of pain
lower levels of OAK symptoms (WOMAC) at the end intensities [67].
of the study period. Present findings did not confirm the
proposed intervention effects on overall OAK symptoms Effects of the PrevOP‑PAP intervention on MVPA and its
(WOMAC) or MVPA. Moreover, MVPA did not mediate HAPA‑proposed volitional precursors
the association between the intervention and OAK symp- Importantly, the question arises why participants
toms (WOMAC). did not increase their MVPA during the intervention
period. On the one hand, increases in intentions to
Indirect effects of the PrevOP‑PAP intervention on OAK engage in regular MVPA one week after the motiva-
symptoms via MVPA tional intervention indicated a successful motivational
Compared to the control group, intervention effects on over- manipulation for all participants. However, manipula-
all OAK symptoms (WOMAC) trended towards a decrease tion checks of the PrevOP-PAP intervention yielded
at the end of the study period. Contrary to our hypoth- only one effect on the HAPA-proposed volitional pre-
esis, this effect did no longer reach statistical significance cursors of MVPA change at the end of the study period.
in sensitivity analyses. Still, exploratory follow-up analy- At 24 months following study entry, only action plan-
ses with different domains of OAK symptoms (WOMAC) ning showed a significant increase in the PrevOP-PAP
as outcomes (i.e., WOMAC-pain, WOMAC-functional intervention condition (vs. PrevOP-PAP active control
limitations, and WOMAC-stiffness) revealed effects of condition). The finding on intervention effects for action
the PrevOP-PAP intervention on pain, but null effects on planning is in line with prior findings in the context of
Lorbeer et al. BMC Musculoskeletal Disorders (2023) 24:550 Page 13 of 18

Table 3 Manifest path models predicting participants‘ symptoms of Osteoarthritis of the Knee (OAK symptoms)
Model 1 Model 2
Dependent variable: OAK symptoms (WOMAC) at M24 Dependent variable: OAK symptoms (WOMAC) at M24

Direct Effect Indirect Effect Direct Effect Indirect Effect

MVPA at M12 OAK symptoms EV → M → DV MVPA at M12 OAK symptoms EV → M → DV

(WOMAC) at M24 (WOMAC) at M24
Est (SE) Est (SE) Est (SE) Est (SE) Est (SE) Est (SE)
95% CI 95% CI 95% CI 95% CI 95% CI 95% CI

PrevOP-PAP intervention -3.78 (3.42) -9.81 (4.83)* 0.43 (0.89) -3.29 (3.05) -8.41 (4.66)† 0.32 (0.55)
condition (vs. PrevOP-PAP [-10.80; 2.58] [-19.18; -0.40] [-0.44; 3.80] [-9.27; 2.69] [-17.53; 0.71] [-0.77; 1.40]
active control condition)
Mediator:
MVPA at M12 -0.11 (0.15) -0.10 (0.14)
[-0.42; 0.19] [-0.38; 0.19]
Covariates:
OAK symptoms -0.06 (0.05) 0.46 (0.09)*** 0.01 (0.01) -0.05 (0.06) 0.31 (0.09)*** 0.00 (0.01)
(WOMAC) at M0 [-0.16; 0.02] [0.29; 0.64] [-0.01; 0.05] [-0.16; 0.07] [0.14; 0.48] [-0.01; 0.02]
MVPA at M0 0.72 (0.08)*** 0.07 (0.17) -0.08 (0.11) 0.71 (0.06)*** 0.03 (0.14) -0.07 (0.10)
[0.57; 0.86] [-0.25; 0.40] [-0.31; 0.13] [0.60; 0.82] [-0.23; 0.30] [-0.27; 0.13]
PrevOP-MMT high- 0.01 (3.74) 0.31 (5.66) -0.00 (0.36)
impact exercise condi- [-7.32; 7.35] [-10.78; 11.40] [-0.71; 0.70]
tion (vs. PrevOP-MMT
active control condition)
PrevOP-MMT low- 0.37 (3.71) -1.39 (5.74) -0.04 (0.36)
impact exercise condi- [-6.90; 7.64] [-12.65; 9.86] [-0.75; 0.67]
tion (vs. PrevOP-MMT
active control condition)
Sex (male vs. female) 0.86 (3.13) 0.04 (4.79) -0.08 (0.33)
[-5.27; 6.99] [-9.35; 9.43] [-0.73; 0.56]
Age -0.25 (0.21) -0.16 (0.34) 0.02 (0.04)
[-0.67; 0.16] [-0.83; 0.51] [-0.06; 0.11]
Body mass index -0.48 (0.35) 1.30 (0.51)* 0.05 (0.08)
[-1.16; 0.20] [0.29; 2.31] [-0.11; 0.20]
Divorced (vs. not 8.01 (3.96)* -5.28 (6.35) -0.77 (1.23)
divorced) [0.24; 15.78] [-17.72; 7.15] [-3.18; 1.64]
VAS-pain 0.21 (1.04) 3.41 (1.57)* -0.02 (0.11)
[-1.83; 2.25] [0.34; 6.48] [-0.23; 0.19]
Positive outcome -2.06 (1.75) -0.66 (2.70) 0.20 (0.35)
expectancies [-5.49; 1.37] [-5.95; 4.63] [-0.48; 0.88]
Source of self-efficacy: 0.55 (1.50) -1.34 (2.31) -0.05 (0.17)
negative affect [-2.38; 3.49] [-5.86; 3.18] [-0.38; 0.27]
R2 MVPA at M12: R2 = 0.57; OAK symptoms at M24: R2 = 0.26 MVPA at M12: R2 = 0.60; OAK symptoms at M24: R2 = 0.32
N = 241 participants. Because manifest path-models were fully saturated, no model fit could be determined. Unstandardized coefficients. Dichotomous covariates
(coded 1/0): PrevOP-PAP intervention condition (coded 1, vs. PrevOP-PAP active control condition, coded 0), PrevOP-MMT high- impact exercise condition (coded 1,
vs. PrevOP-MMT active control condition, coded 0), PrevOP-MMT low-impact exercise condition (coded 1, vs. PrevOP-MMT active control condition, coded 0), sex male
(coded 1, vs. female, coded 0), divorced (coded 1, vs. not divorced, coded 0). Continuous covariates (grand-mean centered, per one point increase): OAK symptoms
(WOMAC) at M0, MVPA at M0, age in years, body mass index, VAS-pain, positive outcome expectancies, source of self-efficacy: negative affect
Est. Estimate, SE Standard error, CI Confidence interval, EV Exogenous variable, M Mediator, DV Dependent variable, MVPA Moderate to Vigorous Physical Activity, M0
Baseline, M12 12-months follow-up, M24 24-months follow-up, PrevOP-PAP PrevOP-Psychological Adherence Program, PrevOP-MMT PrevOP-Main Medical Trial, VAS
Visual analogue scale, WOMAC Western Ontario and McMaster Universities Osteoarthritis Index [7]
†
p < .10; * p < .05; ** p < .01; *** p < .001

cardiac and orthopedic rehabilitation [45] and osteo- In this context, coping planning and action control
arthritis albeit for shorter time frames [38]. However, have been highlighted as important additional key
prior findings also suggest that action planning as a intervention components to enhance the effects of
stand-alone volitional strategy may not be sufficient to action planning [39]. However, in PrevOP-PAP, manip-
facilitate the uptake and maintenance of MVPA in the ulation checks revealed no significant between-group
long-term [68]. differences in coping planning and action control at
Lorbeer et al. BMC Musculoskeletal Disorders (2023) 24:550 Page 14 of 18

24 months following study entry. As a possible expla- around 47 min per day) when compared to previously
nation for the null effect regarding coping planning, reported levels of MVPA among patients with OAK [72].
action plans formed during the PrevOP-PAP interven- Thus, participants’ capacity for additional increases in
tion might already have been of high enough quality, MVPA throughout the study period might have been lim-
as patients also planned increases in physical activities, ited. Moreover, intervention programs using self-regula-
they were already familiar with. This may have resulted tory strategies such as action planning were shown to be
in less need for additional coping plans, and led to less effective among individuals who were already physi-
maintenance rather than increases in levels of coping cally active when compared to sedentary populations
planning. Moreover, participants of the PrevOP-PAP [73]. At the same time, higher levels of MVPA at baseline
intervention condition reported decreased levels of might also reflect some reactivity to the measurement
pain as an OAK-specific barrier to physical activity over via accelerometery [74]. Future studies should consider
time. This reduction in barriers to become more active extended periods of baseline measurement with acceler-
may have also contributed to a maintenance in levels of ometry in order to prevent measurement reactivity.
coping planning, rather than their expected increase.
Regarding the null effect on action control, three Strengths and limitations
one-month paper–pencil activity calendar phases deliv- This study has several strengths. First, this RCT included
ered throughout the PrevOP-PAP intervention with a long follow-up period up to 24 months post study entry
extended periods of no intervention delivery might (i.e., 12 months following the intervention period) to
not have been sufficient to foster action control in the elucidate causal mechanisms of the PrevOP-PAP inter-
long run. Future research could implement m-health vention in the long term. The intervention program
self-monitoring applications which are permanently PrevOP-PAP was based on theoretically-derived health
available to facilitate continuous and long-term action behavior change techniques to allow for testing underly-
control. ing processes of behavior change [14, 54]. The computer-
Furthermore, the overall moderate OAK severity and assisted intervention facilitated a standardized delivery of
prolonged disease duration in our sample as well as a the PrevOP-PAP intervention program. Moreover, in this
decrease in pain in the intervention condition might fur- RCT, MVPA was objectively assessed using accelerom-
ther explain null effects of the PrevOP-PAP intervention eters. This may have reduced problems often associated
on changes in self-efficacy. With decreasing salience of with MVPA self-reports such as recall biases or mere
barriers or decreasing barriers, such as pain, over time, measurement effects due to repeated active assessments
increases in behavior-specific self-efficacy become less [75, 76].
likely, because self-efficacy is always measured up against However, some limitations must be acknowledged.
perceived difficulties or barriers to act. Also, with regard Despite the advantages of mediation analyses to under-
to the social network formation indicator, i.e., collabora- stand the causal mechanisms of this complex interven-
tive implementation intentions, the PrevOP-PAP inter- tion over time, this modelling approach also comes with
vention did not yield significant intervention effects and drawbacks. Given the complexity of OAK symptoms, it
many participants of the intervention condition decided seems likely that further non-hypothesized factors may
not to participate in the optional network creation inter- have explained changes in OAK symptoms over time
vention. As vicarious experiences and positive affective which, however, were not captured using this theory-
states experienced in social networks serve as sources of guided approach. Future research may apply data-driven
self-efficacy, this might also have affected levels of self- approaches such as Bayesian Networks to further eluci-
efficacy in our sample [10, 69]. Again, the development date intervention mechanisms of the PrevOP-PAP inter-
of m-health applications with features to digitally create vention [77].
social networks may be beneficial to foster network for- Second, intervention delivery and data analyses in this
mation in the long run. Moreover, future interventions RCT were unmasked due to ethical and practical reasons.
could focus on perceived enjoyment with PA, done alone Participants were informed that they would be randomly
or with others, which has been shown to be a strong cor- allocated to either the PrevOP-PAP intervention condi-
relate of PA among patients with osteoarthritis [10] and tion or the PrevOP-PAP active control condition and
might also strengthen the network formation interven- study personnel were aware of delivering the intervention
tion [70, 71]. treatment. Masking of statistical analysis is desirable for
On the other hand, when interpreting null effects on future RCT evaluation.
MVPA, it must be noted that participants demonstrated Third, attrition rates were elevated across two years of
relatively high baseline levels of MVPA (i.e., on average the study period (29%) with the majority of participants
Lorbeer et al. BMC Musculoskeletal Disorders (2023) 24:550 Page 15 of 18

dropping out during the intervention period within Abbreviations

BCT Behavior Change Technique
the first year. This may be explained by a high par- BMI Body Mass Index
ticipant burden due to intensive intervention delivery CONSORT Consolidated Standards of Reporting Trials
and repeated measurements in crossed psychological HAPA Health Action Process Approach
M “Month”
PrevOP-PAP and medical PrevOP-MMT trials. How- MVPA Moderate-to-Vigorous Physical Activity
ever, it must be noted that the sample size of completers PA Physical Activity
in PrevOP-PAP was sufficient to detect the proposed PAP Psychological Adherence Program
PrevOP PREVenting the impairment of primary Osteoarthritis
mediation effect with a power of 0.80 as the overall sam- by high-impact long-term Physical exercise regimen
ple size estimation was higher for the PrevOP-MMT. PrevOP-MMT PrevOP-Main Medical Trial
Moreover, we aimed to reduce potential selection effects PrevOP-PAP PrevOP-Psychological Adherence Program
OAK OsteoArthritis of the Knee
due to attrition by conducting intent-to-treat analyses RCT​ Randomized Controlled Trial
with full maximum likelihood estimation and consider- TIDieR Template for Intervention Description and Replication
ing potential dropout mechanisms as covariates in our WOMAC Western Ontario and McMaster Universities Osteoar-
thritis Index
sensitivity analyses [60].
Fourth, despite numerous advantages of the objective
assessment of PA, this approach may also have draw- Supplementary Information
The online version contains supplementary material available at https://​doi.​
backs. Whereas participants may have increased their org/​10.​1186/​s12891-​023-​06661-x.
engagement in joint-friendly MVPA that are particularly
recommended in the context of OAK such as swimming Additional file 1: Additional Information 1. Inclusion and Exclusion
or riding a bike [6], hip-worn accelerometers cannot cap- Criteria. Additional Figure 1. Conditions of the PrevOP-Main Medical Trial
(PrevOP-MMT) nested in the PrevOP-Psychological Adherence Program
ture these specific types of PA. Thus, participants’ lev- (PrevOP-PAP) conditions. Additional Table 1. Manifest Regression
els of MVPA may have been underestimated. Lastly, the Analyses Predicting Central Variables. Additional Table 2. Manifest Path
active control group and the crossed study design with Models Predicting Participants‘ Functional Limitations Associated with
Osteoarthritis of the Knee. Additional Table 3. Manifest Path Models
structured exercise conditions may have caused over- Predicting Participants‘ Pain Associated with Osteoarthritis of the Knee.
all higher levels of MVPA throughout the study period Additional Table 4. Manifest Path Models Predicting Participants‘ Stiffness
both in the PrevOP-PAP intervention condition and the Associated with Osteoarthritis of the Knee.
PrevOP-PAP active control condition. This might have
further limited the variance in levels of MVPA. Acknowledgements
In grateful memory of Dieter Felsenberg.
Authors also wish to thank the teams of PrevOP-PAP and PrevOP-MMT for their
Conclusions dedicated work and contributions to the projects: Daniela Lange, Diana Hilda
This psychological adherence program was based on Hohl, Susannah Motter, Luisa Wirth, Lisa Bosch, Theresa Reschke, Annekathrin
HAPA-derived behavior change techniques and specifi- Teichmann, Eva Marie Keinert, Patrick Klaiber, Ulrike Panse, Nadine Christen,
Tim Felsenberg, Martina Kratzsch, Felix Müller, and Frank Touby.
cally designed for patients with moderate OAK to facili-
tate the uptake and maintenance of physical activity. Protocol version and trial status
Whereas levels of action planning significantly increased This is the primary analyses report as registered with the German Clinical Trials
Register on 26 January 2016. No trial registry modifications were undertaken.
following the intervention, primary analyses did not yield With regard to the study protocol [8], modifications are as follows: Rand-
beneficial effects of the PrevOP-PAP intervention on omization checks were conducted using univariate analyses of variance. For
physical activity and limited effects on OAK symptoms manipulation checks, manifest regression analyses were fit. Additional explora-
tory follow-up analyses with WOMAC-functional limitations, WOMAC-pain,
(WOMAC), i.e., only a decrease in WOMAC-pain at the and WOMAC-stiffness as outcomes were conducted.
end of the study period emerged in exploratory follow-up
analyses. Resembling meta-analytic findings on self-man- Authors’ contributions
N.K. and R.S. (principal investigators of PrevOP-PAP) and N.L. (PrevOP-PAP
agement programs for OAK, the PrevOP-PAP interven- study researcher): statistical analyses and first draft of the manuscript. J.K., A.D.,
tion might thus appear promising for improved disease and S.D.M.: (PrevOP-PAP study researchers): coordination of PrevOP-PAP. G.A.
management (e.g., coping with pain). However, as physi- and H.B. (PrevOP-MMT study researchers): coordination of PrevOP-MMT. P.M.
(principal trial statistician of PrevOP-PAP and PrevOP-MMT): randomization,
cal activity did not serve as a mediator of this exploratory power analyses PrevOP-MMT, support in statistical analyses. W.E. (principal
finding, underlying mechanisms of improvements in pain investigator of PrevOP-MMT): recruitment of participants, inclusion, medical
still remain unclear. Future research should further inves- assessments. All authors contributed to the writing of this manuscript and
approved the final version.
tigate which intervention components of the PrevOP-
PAP specifically targeted the patients’ pain management. Funding
Subsequently, the intervention program could be refined Open Access funding enabled and organized by Projekt DEAL. This work is
part of the overarching OVERLOAD-PrevOP consortium (https://​overl​oad-​
and provided as an m-health application on a large scale prevop.​chari​te.​de/) and was supported by two subproject-grants from the
for patients with OAK. German Federal Ministry of Education and Research (Bundesministerium für
Lorbeer et al. BMC Musculoskeletal Disorders (2023) 24:550 Page 16 of 18

Bildung und Forschung, BMBF, Heinemannstr. 2 & 6, 53175 Bonn, Germany; 6. Centers for Disease Control and Prevention: Physical activity for arthri-
01EC1408H, SPP7 to N.K. and R.S.; 01EC1408L, SPP6 to Dieter Felsenberg, W.E., tis. https://​www.​cdc.​gov/​arthr​itis/​basics/​physi​cal-​activ​ity-​overv​iew.​
and P.M.). Authors also gratefully acknowledge the support from the Focus html (2018). Accessed 11 Feb 2023.
Area DynAge (Freie Universität Berlin; Charité Universitätsmedizin Berlin) 7. Bellamy N. WOMAC osteoarthritis index user guide, version V. Brisbane; 2002.
during the preparation of the grant application (DynAge 1_4 to W.E. and Petra https://​womac.​com/​womac/​womac_​userg​uide.​php.
Knaus) and the support by the Open Access Publication Fund of the Freie 8. Knoll N, Hohl DH, Motter S, Keller J, Lange D, Felsenberg D, et al. Facili-
Universität Berlin. tating physical activity and reducing symptoms in patients with knee
osteoarthritis: study protocol of a randomized controlled trial to test a
Availability of data and materials theory-based PrevOP-psychological adherence program (PrevOP-PAP).
The datasets generated and analyzed during the current study are available BMC Musculoskelet Disord. 2018;19:221.
from the corresponding author upon request. 9. Krauss I, Katzmarek U, Rieger MA, Sudeck G. Motives for physical exer-
cise participation as a basis for the development of patient-oriented
exercise interventions in osteoarthritis: a cross-sectional study. Eur J
Declarations Phys Rehabil Med. 2017;53(4):590–602.
10. Berry A, McCabe CS, Halls S, Muir S, Walsh N. Beliefs, motives and gains
Ethics approval and consent to participate associated with physical activity in people with osteoarthritis. Muscu-
The ethics committee of the Charité – Universitätsmedizin Berlin approved loskeletal Care. 2021;19(1):52–8.
this study (EA4/027/15). All procedures were carried out in compliance with 11. Dobson F, Bennell KL, French SD, Nicolson PJ, Klaasman RN, Holden
the Helsinki Declaration. Before inclusion into the study by PrevOP-MMT medi- MA, et al. Barriers and facilitators to exercise participation in people
cal personnel, written informed consent was obtained from each participant with hip and/or knee osteoarthritis: synthesis of the literature using
for participation in all study parts (PrevOP-MMT and PrevOP-PAP). Amongst behavior change theory. Am J Phys Med Rehabil. 2016;95(5):372–89.
other information, participants were informed: (1) that participation in the 12. Coste N, Guiguet-Auclair C, Gerbaud L, Pereira B, Berland P, Gay
study is completely voluntary, (2) that they have the right to withdraw from C, et al. Perceived barriers to and facilitators of physical activity in
the trial whenever they desire and that they do not have to state a reason for people with knee osteoarthritis: development of the evaluation of the
their decision, (3) that refusal to participate or discontinuation of participation perception of physical activity questionnaire. Ann Phys Rehabil Med.
will not have any consequences for the usual care they receive, (4) that their 2020;63(3):202–8.
identifying information will be kept strictly confidential (and apart from the 13. Kanavaki AM, Rushton A, Efstathiou N, Alrushud A, Klocke R, Abhishek
remainder of their data), their data being made anonymous by assignment A, et al. Barriers and facilitators of physical activity in knee and hip
of a pseudonym (i.e., a participant ID-number), (5) that their data will be osteoarthritis: a systematic review of qualitative evidence. BMJ Open.
stored, analyzed, and published in an anonymous form by collaborating study 2017;7(12):e017042.
researchers. 14. Schwarzer R. Modeling health behavior change: how to predict and
modify the adoption and maintenance of health behaviors. Appl
Consent for publication Psychol. 2008;57(1):1–29.
Not applicable. 15. Schwarzer R, Lippke S, Luszczynska A. Mechanisms of health behavior
change in persons with chronic illness or disability: the Health Action
Competing interests Process Approach (HAPA). Rehabil Psychol. 2011;56(3):161–70.
The authors declare no competing interests. 16. Zhang C-Q, Zhang R, Schwarzer R, Hagger MS. A meta-analysis of the
health action process approach. Health Psychol. 2019;38(7):623–37.
Author details 17. Heckhausen H. Motivation und Handeln. 2nd ed. Heidelberg: Springer
1
Department of Education and Psychology, Health Psychology Division, Berlin; 1989.
Freie Universität Berlin, Habelschwerdter Allee 45, Berlin 14195, Germany. 18. Rosenstock IM. Historical origins of the health belief model. Health
2
Centre for Muscle‑ and Bone Research, Department of Radiology, Charité Educ Monogr. 1974;2(4):328–35.
– Universitätsmedizin Berlin, Hindenburgdamm 30, Berlin 12200, Germany. 19. Bandura A. Health promotion by social cognitive means. Health Educ
3
Institute for Clinical Epidemiology and Applied Biometry, Universitätsklinikum Behav. 2004;31(2):143–64.
Tübingen, Silcherstr. 5, Tübingen 72076, Germany. 4 Department of Trau- 20. Gollwitzer PM. Implementation intentions: strong effects of simple
matology and Reconstructive Surgery, Charité – Universitätsmedizin Berlin, plans. Am Psychol. 1999;54(7):493–503.
Hindenburgdamm 30, Berlin 12200, Germany. 5 CARE‑BEH Center for Applied 21. Kwasnicka D, Presseau J, White M, Sniehotta FF. Does planning how
Research on Health Behavior and Health, SWPS University, ul. Ostrowskiego to cope with anticipated barriers facilitate health-related behaviour
30b, Wrocław 53‑238, Poland. change? A systematic review. Health Psychol Rev. 2013;7(2):129–45.
22. Sniehotta FF, Scholz U, Schwarzer R. Action plans and coping plans for
Received: 17 February 2023 Accepted: 23 June 2023 physical exercise: a longitudinal intervention study in cardiac rehabili-
tation. Br J Health Psychol. 2006;11(1):23–37.
23. Sniehotta FF, Scholz U, Schwarzer R. Bridging the intention–behaviour
gap: planning, self-efficacy, and action control in the adoption and
maintenance of physical exercise. Psychol Health. 2005;20(2):143–60.
References 24. Carver CS, Scheier MF. Control processes and self-organization as
1. Hunter DJ, March L, Chew M. Osteoarthritis in 2020 and beyond: a Lancet complementary principles underlying behavior. Pers Soc Psychol Rev.
Commission. Lancet. 2020;396(10264):1711–2. 2002;6(4):304–15.
2. Cui A, Li H, Wang D, Zhong J, Chen Y, Lu H. Global, regional prevalence, 25. Burkert S, Knoll N, Scholz U, Roigas J, Gralla O. Self-regulation follow-
incidence and risk factors of knee osteoarthritis in population-based ing prostatectomy: phase-specific self-efficacy beliefs for pelvic-floor
studies. EClinicalMedicine. 2020;29–30:100587. exercise. Br J Health Psychol. 2012;17(2):273–93.
3. Vitaloni M, Botto-van Bemden A, Sciortino Contreras RM, Scotton D, 26. Jackson SE, Steptoe A, Wardle J. The influence of partner’s behavior
Bibas M, Quintero M, et al. Global management of patients with knee on health behavior change: the English longitudinal study of ageing.
osteoarthritis begins with quality of life assessment: a systematic review. JAMA Intern Med. 2015;175(3):385–92.
BMC Musculoskelet Disord. 2019;20:493. 27. Pauly T, Keller J, Knoll N, Michalowski VI, Hohl DH, Ashe MC, et al. Moving
4. Katz JN, Arant KR, Loeser RF. Diagnosis and treatment of hip and knee in sync: hourly physical activity and sedentary behavior are synchronized
osteoarthritis: a review. JAMA. 2021;325(6):568–78. in couples. Ann Behav Med. 2020;54(1):10–21.
5. Kolasinski SL, Neogi T, Hochberg MC, Oatis C, Guyatt G, Block J, et al. 28. Keller J, Hohl DH, Hosoya G, Heuse S, Scholz U, Luszczynska A, et al. Long-
2019 American College of Rheumatology/Arthritis Foundation guide- term effects of a dyadic planning intervention with couples motivated to
line for the management of osteoarthritis of the hand, hip, and knee. increase physical activity. Psychol Sport Exerc. 2020;49:101710.
Arthritis Rheumatol. 2020;72(2):220–33.
Lorbeer et al. BMC Musculoskeletal Disorders (2023) 24:550 Page 17 of 18

29. Knoll N, Hohl DH, Keller J, Schuez N, Luszczynska A, Burkert S. Effects of Vibrationstrainingsgerät (Galileo) https://​drks.​de/​search/​de/​trial/​DRKS0​
dyadic planning on physical activity in couples: a randomized controlled 00096​77. Accessed 11 Feb 2023.
trial. Health Psychol. 2017;36(1):8–20. 48. Schulz KF, Altman DG, Moher D. CONSORT 2010 statement: updated
30. Rackow P, Scholz U, Hornung R. Effects of a new sports companion on guidelines for reporting parallel group randomised trials. J Pharmacol
received social support and physical exercise: an intervention study. Appl Pharmacother. 2010;1(2):100–7.
Psychol Health Well Being. 2014;6(3):300–17. 49. Hoffmann TC, Glasziou PP, Boutron I, Milne R, Perera R, Moher D, et al.
31. Scholz U, Berli C, Lüscher J, Knoll N. Dyadic interventions to promote Better reporting of interventions: template for intervention description
behavior change in couples. In: Hagger MS, Cameron L, Hamilton K, and replication (TIDieR) checklist and guide. Br Med J. 2014;348:g1687.
Hankonen N, Lintunen T, editors. The Handbook of Behavior Change 50. Fritz MS, MacKinnon DP. Required sample size to detect the mediated
Cambridge. UK: Cambridge University Press; 2020. p. 632–48. effect. Psychol Sci. 2007;18(3):233–9.
32. Kulis E, Szczuka Z, Keller J, Banik A, Boberska M, Kruk M, et al. Collabora- 51. Kellgren JH, Lawrence JS. Radiological assessment of osteo-arthrosis.
tive, dyadic, and individual planning and physical activity: a dyadic Ann Rheum Dis. 1957;16(4):494–502.
randomized controlled trial. Health Psychol. 2022;41(2):134–44. 52. Haskell WL, Lee I-M, Pate RR, Powell KE, Blair SN, Franklin BA, et al. Physi-
33. Szczuka Z, Kulis E, Boberska M, Banik A, Kruk M, Keller J, et al. Can indi- cal activity and public health: updated recommendation for adults
vidual, dyadic, or collaborative planning reduce sedentary behavior? A from the American College of Sports Medicine and the American Heart
randomized controlled trial. Soc Sci Med. 2021;287:114336. Association. Circulation. 2007;116(9):1081–93.
34. Prestwich A, Conner MT, Lawton RJ, Ward JK, Ayres K, McEachan RRC. 53. Robert Koch-Institut. Arthrose. Gesundheitsberichterstattung des
Randomized controlled trial of collaborative implementation inten- Bundes. Heft 54. Berlin: Robert Koch-Institut; 2013.
tions targeting working adults’ physical activity. Health Psychol. 54. Michie S, Richardson M, Johnston M, Abraham C, Francis J, Hardeman W,
2012;31(4):486–95. et al. The behavior change technique taxonomy (v1) of 93 hierarchi-
35. Bartholdy C, Juhl C, Christensen R, Lund H, Zhang W, Henriksen M. The cally clustered techniques: building an international consensus for
role of muscle strengthening in exercise therapy for knee osteoarthritis: the reporting of behavior change interventions. Ann Behav Med.
a systematic review and meta-regression analysis of randomized trials. 2013;46(1):81–95.
Semin Arthritis Rheum. 2017;47(1):9–21. 55. Sasaki JE, John D, Freedson PS. Validation and comparison of ActiGraph
36. Fransen M, McConnell S, Harmer AR, Van der Esch M, Simic M, Bennell KL. activity monitors. J Sci Med Sport. 2011;14(5):411–6.
Exercise for osteoarthritis of the knee: a cochrane systematic review. Br J 56. Tabachnick BG, Fidell LS. Using multivariate statistics. New York: Allyn &
Sports Med. 2015;49(24):1554–7. Bacon/Pearson Education; 2007.
37. Gay C, Chabaud A, Guilley E, Coudeyre E. Educating patients about 57. Warner LM, Schüz B, Wolff JK, Parschau L, Wurm S, Schwarzer
the benefits of physical activity and exercise for their hip and knee R. Sources of self-efficacy for physical activity. Health Psychol.
osteoarthritis. Systematic literature review. Ann Phys Rehabil Med. 2014;33(11):1298–308.
2016;59(3):174–83. 58. Huskisson EC. Measurement of pain. Lancet. 1974;304(7889):1127–31.
38. Williams AD, O’Brien J, Mulford J, Mathew R, Thapa DK, Hamilton K, et al. 59. R Core Team. R: A language and environment for statistical computing.
Effect of combined exercise training and behaviour change counsel- Vienna: Foundation for Statistical Computing; 2022.
ling versus usual care on physical activity in patients awaiting hip and 60. Graham JW. Missing data analysis: making it work in the real world. Annu
knee arthroplasty: a randomised controlled trial. Osteoarthr Cartil Open. Rev Psychol. 2009;60(1):549–76.
2022;4(4):100308. 61. Rosseel Y. lavaan: an R package for structural equation modeling. J Stat
39. Sniehotta FF, Scholz U, Schwarzer R, Fuhrmann B, Kiwus U, Völler H. Long- Softw. 2012;48(2):1–36.
term effects of two psychological interventions on physical exercise 62. Coman EN, Picho K, McArdle JJ, Villagra V, Dierker L, Iordache E. The paired
and self-regulation following coronary rehabilitation. Int J Behav Med. t-test as a simple latent change score model. Front Psychol. 2013;4:738.
2005;12:244–55. 63. MacKinnon DP, Lockwood CM, Williams J. Confidence limits for the
40. O’Brien J, Hamilton K, Williams AD, Fell J, Mulford J, Cheney M, et al. indirect effect: distribution of the product and resampling methods.
Improving physical activity, pain and function in patients waiting for Multivariate Behav Res. 2004;39(1):99–128.
hip and knee arthroplasty by combining targeted exercise training with 64. Sobel ME. Asymptotic intervals for indirect effects in structural equations
behaviour change counselling: study protocol for a randomised con- models. Sociol Methodol. 1982;13:290–312.
trolled trial. Trials. 2018;19:425. 65. Kroon FPB, van der Burg LRA, Buchbinder R, Osborne RH, Johnston
41. Burkert S, Scholz U, Gralla O, Roigas J, Knoll N. Dyadic planning of health- RV, Pitt V. Self-management education programmes for osteoarthritis.
behavior change after prostatectomy: a randomized-controlled planning Cochrane Database Syst Rev. 2014;1:CD008963.
intervention. Soc Sci Med. 2011;73(5):783–92. 66. Gilbert AL, Lee J, Ehrlich-Jones L, Semanik PA, Song J, Pellegrini CA, et al. A
42. Bodenmann G, Randall AK, Falconier MK. Coping in couples: The systemic randomized trial of a motivational interviewing intervention to increase
transactional model (STM). In: Falconier MK, Randall AK, Bodenmann G, lifestyle physical activity and improve self-reported function in adults
Falconier MK, Randall AK, Bodenmann G, editors. Couples coping with with arthritis. Semin Arthritis Rheum. 2018;47(5):732–40.
stress: a cross-cultural perspective. New York: Routledge/Taylor & Francis 67. McCracken L. Learning to live with the pain: acceptance of pain predicts
Group; 2016. p. 5–22. adjustment in persons with chronic pain. Pain. 1998;74(1):21–7.
43. Lima YL, Lee H, Klyne DM, Dobson FL, Hinman RS, Bennell KL, et al. How 68. Sniehotta FF, Schwarzer R, Scholz U, Schüz B. Action planning and coping
do nonsurgical interventions improve pain and physical function in peo- planning for long-term lifestyle change: theory and assessment. Eur J Soc
ple with osteoarthritis? A scoping review of mediation analysis studies. Psychol. 2005;35:565–76.
Arthritis Care Res (Hoboken). 2023;75(3):467–81. 69. Bandura A. Self-efficacy: the exercise of control. New York: Freeman; 1997.
44. Evers A, Klusmann V, Ziegelmann JP, Schwarzer R, Heuser I. Long-term 70. Ziegelmann JP, Knoll N. Future directions in the study of health behavior
adherence to a physical activity intervention: the role of telephone- among older adults. Gerontology. 2015;61:469–76.
assisted vs. self-administered coping plans and strategy use. Psychol 71. Hoppmann CA, Gerstorf D. Biobehavioral pathways underlying spousal
Health. 2012;27(7):784–97. health dynamics: its nature, correlates, and consequences. Gerontology.
45. Fleig L, Pomp S, Schwarzer R, Lippke S. Promoting exercise maintenance: 2014;60(5):458.
how interventions with booster sessions improve long-term rehabilita- 72. Farr JN, Going SB, Lohman TG, Rankin L, Kasle S, Cornett M, et al. Physical
tion outcomes. Rehabil Psychol. 2013;58(4):323–33. activity levels in patients with early knee osteoarthritis measured by
46. Keller J, Fleig L, Hohl DH, Wiedemann AU, Burkert S, Luszczynska A, accelerometry. Arthritis Care Res (Hoboken). 2008;59(9):1229–36.
et al. Which characteristics of planning matter? Individual and dyadic 73. Carraro N, Gaudreau P. Spontaneous and experimentally induced action
physical activity plans and their effects on plan enactment. Soc Sci planning and coping planning for physical activity: a meta-analysis.
Med. 2017;189:53–62. Psychol Sport Exerc. 2013;14:228–48.
47. German Clinical Trials Register: Prävention des Fortschreitens einer 74. Baumann S, Groß S, Voigt L, Ullrich A, Weymar F, Schwaneberg T, et al.
Gonarthrose Grad 2–3 durch mechanische Stimulation mit einem Pitfalls in accelerometer-based measurement of physical activity: the
Lorbeer et al. BMC Musculoskeletal Disorders (2023) 24:550 Page 18 of 18

presence of reactivity in an adult population. Scand J Med Sci Sports.

2017;28(3):1056–63.
75. Godin G, Bélanger-Gravel A, Amireault S, Vohl M-C, Pérusse L. The effect
of mere-measurement of cognitions on physical activity behavior: a
randomized controlled trial among overweight and obese individuals. Int
J Behav Nutr Phys Act. 2011;8:2.
76. Prince SA, Adamo KB, Hamel ME, Hardt J, Connor Gorber S, Tremblay M.
A comparison of direct versus self-report measures for assessing physical
activity in adults: a systematic review. Int J Behav Nutr Phys Act. 2008;5:56.
77. Pearl J, Mackenzie D. The ladder of causation. In: The book of why the
new science of cause and effect. New York: Allen Lane; 2018.

Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in pub-
lished maps and institutional affiliations.

Ready to submit your research ? Choose BMC and benefit from:

• fast, convenient online submission

• thorough peer review by experienced researchers in your field
• rapid publication on acceptance
• support for research data, including large and complex data types
• gold Open Access which fosters wider collaboration and increased citations
• maximum visibility for your research: over 100M website views per year

At BMC, research is always in progress.

Learn more biomedcentral.com/submissions