e-space.mmu.ac.uk Drew JOSPT.docx · Web viewXXXX is funded by a National Institute for Health Research (NIHR) Clinical Doctoral Research Fellowship (CDRF -2013-04-044); EMAH, XXXX

Data-driven diagnostic subgroups in patellofemoral pain

Development of data-driven diagnostic subgroups for people with patellofemoral pain

using modifiable clinical, biomechanical and imaging features

Authors: XXXX

XXXX

Ethical approval All procedures performed in the study involving human participants were in

accordance with ethical standards. Ethical approval for the study was obtained prior to

commencement of the study from XXXX Research Ethics Committee (14/NE/1131)

Funding: XXXX is funded by a National Institute for Health Research (NIHR) Clinical

Doctoral Research Fellowship (CDRF -2013-04-044); EMAH, XXXX and XXXX are

supported by the National Institute for Health Research (NIHR) infrastructure at XXXX. This

paper presents independent research funded by the National Institute for Health Research

(NIHR). XXXX is supported by the National Institute for Health Research (NIHR) XXXX

Biomedical Research Centre (BRC)The views expressed are those of the authors and not

necessarily those of the NHS, the NIHR or the Department of Health. This work was also

supported in part by funding from the XXXX Osteoarthritis Treatment Centre (Ref 20083)

and the XXXX Centre for Sport, Exercise and Osteoarthritis (Ref 20194).

Statement of financial disclosure and conflict of interest: None of the authors declare

any conflict of interest

1

1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25


Address for correspondence:

XXXX

Word Count: 3998

2

26

27

28

29

30

31

32


Acknowledgments: XXXX

3

33

34

35

36

37


Abstract

Background. Unfavourable treatment outcomes for people with patellofemoral pain (PFP)

have been attributed to the potential existence of subgroups that respond differently to

treatment.

Objectives: This study aimed to identify subgroups within PFP by combining modifiable

clinical, biomechanical and imaging features and exploring the prognosis of these

subgroups.

Methods. Longitudinal cohort with baseline cluster analyses. Baseline data were analysed

using a two-stage cluster analysis; 10 features were analysed within 4 health domains

before being combined at the second stage. Prognosis of the subgroups was assessed at

12-months with subgroup differences in the Global Rating of Change Scale analysed using

an exploratory logistic regression adjusted for known confounders.

Results. 70 participants were included (mean age 31 years; 43 (61%) female). Cluster

analysis revealed 4 subgroups: ‘Strong’, ‘Pronation & Malalignment’, ‘Weak’ and ‘Active &

Flexible’. Descriptively, compared to the Strong subgroup (55% favourable), the odds of a

favourable outcome were lower in the Weak subgroup (31% favourable; adjusted odds ratio

[OR] 0.30; 95% confidence intervals [CI] 0.07, 1.36) and Pronation & Malalignment subgroup

4

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61


(50%; OR 0.64; 95% CI 0.11, 3.66), and higher in the Active & Flexible subgroup (63%; OR

1.24 (95% CI 0.20, 7.51). After adjustment, compared to the Strong subgroup, differences

between some subgroups remained substantive but none were statistically significant.

Conclusion. In this relatively small cohort, 4 PFP subgroups were identified which show

potentially different outcomes at 12 months. Further research is required to determine

whether a stratified treatment approach using these subgroups would improve outcomes for

people with PFP.

Level of evidence: Diagnosis, level 2b.

Key words: knee; patellofemoral joint; biomechanics; MRI

5

62

63

64

65

66

67

68

69

70

71

72

73

74


Introduction

Patellofemoral pain (PFP) is widely considered a multifactorial condition characterised by a

gradual onset of pain related to changes in the patellofemoral joint (PFJ) and not associated

with any other knee condition [7]. One in 6 care-seeking adults with knee pain are diagnosed

with PFP [68] and, in adolescents, the population prevalence is 6% within the general

population [42]. There also remains a concern that PFP may be a precursor to future

osteoarthritis [11]. The multifactorial nature of PFP means its causes have been attributed to

various clinical, biomechanical and structural factors, or a combination of these [67]. Current

treatments that aim to address these factors appear suboptimal with 40% of people still

reporting unfavourable outcomes one year after treatment [8].

These unfavourable outcomes have been attributed to the belief that subgroups may exist

within the wider PFP population and respond differently to treatment [67]. Emerging evidence

has shown that stratifying patients may optimise treatment outcomes across a range of

conditions[25] , but there is a paucity of evidence for subgrouping and subsequent

stratification of care in PFP. A few randomised controlled trials [18, 41, 43] have stratified PFP

patients by matching treatment to a specific problem. These trials reported improved

outcomes in 79% of PFP participants following foot orthotics [41], 62% following hip

strengthening [18] and one extra successful outcome for every three patients treated with foot

exercises and orthoses [43]. These findings support the concept of targeted, stratified care in

PFP and best practice guidelines for PFP recommend a tailored treatment approach [3].

6

75

76

77

78

79

80

81

82

83

84

85

86

87

88

89

90

91

92

93

94

95

96

97

98


A proposed framework for subgrouping research [32] suggests that prior to testing the effect

of subgrouping, a hypothesis-setting stage is required which attempts to identify clinically

important subgroups and explore the prognostic effect attributed to subgroup membership

[32]. At this stage, data-driven diagnostic subgroups are advantageous as they can later be

studied against a range of treatments [34, 40], rather than groups being based on the response

to only one treatment (i.e. treatment effect modifiers). Diagnostic PFP subgroups have been

suggested by a number of studies based on single factors [14, 24, 48, 57]. However, only a few

studies [31, 55, 56] have identified diagnostic subgroups comprising of multiple factors from

multiple domains. Of these studies, only Selfe et al (2016) [55] derived subgroups from

rigorous statistical methods. The three diagnostic subgroups they identified are of high

clinical utility, requiring only six simple clinical tests. However, these subgroups do not

incorporate PFJ structure, which requires imaging [48, 57] and biomechanical function and

involves complex equipment and evaluation [14, 63]. Overall, their prognostic value remains

unknown [47].

The primary aim of this study was to combine modifiable clinical, biomechanical and imaging

features to identify potential data-driven diagnostic subgroups within a PFP cohort. Based on

data from a 12-month follow-up, the secondary aim was to explore the prognosis of these

data-driven subgroups.

Methods

Study design

7

99

100

101

102

103

104

105

106

107

108

109

110

111

112

113

114

115

116

117

118

119

120

121

122


This longitudinal cohort study comprised of a cross-sectional analysis of baseline

characteristics and a 12-month follow-up exploratory analysis. Ethical approval was obtained

(14/NE/1131) and all participants completed written informed consent prior to entering the

study. The sample size was based on the recommended rule of thumb for cluster analyses

of n =2k (whereby k is the number of variables) [21]. For our model, variables were analysed

within selected health domains. We allowed at most six variables (k), representing the

selected domains, requiring a minimum of 64 participants (26). To account for a potential

20% drop-out rate, we aimed to recruit 77 participants. The study was reported in

accordance with the Strengthening the Reporting of Observational studies in Epidemiology

(STROBE) guidelines [62]

Setting

All assessments were conducted at a UK teaching hospital from November 2014 to April

2016. Participants from the general population were recruited from a local National Health

Service (NHS) musculoskeletal service via clinician referral, posters in local sports clubs and

university alumni. Further electronic searches of the local NHS musculoskeletal database

were also made for patients previously diagnosed with either ‘anterior knee pain’ or

‘patellofemoral pain’.

Participants

TABLE 1 shows the participant eligibility criteria. Participants were screened by the same

muscuskeletal clinician who based their PFP diagnosis on both imaging and clinical findings.

During the 12 months between baseline and follow-up, 14 participants were also involved in

8

123

124

125

126

127

128

129

130

131

132

133

134

135

136

137

138

139

140

141

142

143

144

145

146

147


a nested feasibility study and received a 6-week targeted hip strengthening intervention [18].

Further details on the intervention are provided elsewhere [18]. The other 56 participants were

advised to continue with their normal activities and were allowed to be treated if required.

Exercise based treatment was adjusted for in the analyses.

Variables

To capture the multifactorial nature of PFP, a range of features were considered to inform

the diagnostic subgrouping. Variables were derived from systematic reviews which identified

features associated with PFP [1, 17, 39]. Variables were selected if they satisfied all the following

criteria: i) demonstrated association with PFP from at least 2 or more studies; ii) published

thresholds and/or normative data that can be used to clinically interpret findings; iii)

considered clinically modifiable with conservative treatment.

Data sources

Figure 1 provides an overview of the selected variables and how they were collected.

Participants completed assessments in the following order: clinical, biomechanical and MRI.

A detailed description of the procedures for the assessment methods is available in the

APPENDIX 1. Additional variables including patient-related factors (shown in Table 2) and

supplementary clinical descriptors were collected and applied to each of the final subgroups.

Clinical assessment

9

148

149

150

151

152

153

154

155

156

157

158

159

160

161

162

163

164

165

166

167

168

169

170

171


Hamstrings (Fig.1d), quadriceps (Fig.1e) and gastrocnemius (Fig.1f) flexibility were

measured using a digital inclinometer in accordance with previously published methods [54].

Digital inclinometers have shown intraclass correlation coefficients (ICCs) of 0.53-0.98 for

the knee[53] and 0.91-0.97 for the ankle[60] Static foot posture was measured using the Foot

Posture Index (FPI), a 6-item clinical tool that quantifies foot posture with an excessive score

indicating pronation [51].

Biomechanical assessment

Three-dimensional kinematics were assessed during stair descent using a VICON, motion

capture system (Vicon Nexus Version 1.6; Vicon, Oxford, UK). Stair descent is recognised to

load the patellofemoral joint [22] and is reported to be symptomatic for people with PFP [10]

(Fig.1c). Data collected were analysed in Visual 3D (C-Motion, Rockville, Maryland). The

kinematics of interest were: i) peak hip internal rotation angle, defined as the thigh with

respect to the pelvis; ii) peak knee flexion angle, defined as the thigh with respect to the

shank.

A Biodex isokinetic system 4 (IRPS Mediquipe, UK) was used to assess muscle strength.

Data were collected by Biodex Advantage Software (IRPS Mediquipe, UK). The concentric

strength measures of interest were: i) peak hip abduction torque based on the maximum hip

abduction torque across five repetitions (Fig.1h); ii) peak knee extension torque based on

the maximum knee extension torque across five repetitions (Fig.1i). Biodex has shown test-

retest reliability of 0.95 ICC for knee extensor strength[20] and 0.81-0.95 for hip abduction

strength[5]. These strength measures were normalised to body weight (Nm/kg).

10

172

173

174

175

176

177

178

179

180

181

182

183

184

185

186

187

188

189

190

191

192

193

194

195

196


MRI assessment

In brief (full version see APPENDIX 1), the protocol comprised of sagittal, transverse and

coronal plane sequences acquired with a 3.0T scanner (Siemens Magnetom Verio, Siemens

Healthcare, Germany) while participants were supine with the knee in full extension and the

quadriceps relaxed. The two variables of interest were MRI bisect offset (BSO) and MRI

patella tilt angle (PTA) which measure the alignment of the patella and were originally

developed using x-ray. Figure 1a and 1b show how both BSO and PTA were calculated

respectively. The intra-rater reliability for both BSO and PTA was established for a single

reader (AJG) by re-scoring 10 participants scans. This showed intraclass correlation

coefficient (3, k) values of 0.94 (95% CI 0.74, 0.99) for BSO and ICC (3, k) 0.98 (95% CI

0.91, 0.99) for PTA. In addition to the selected variables, a MRI osteoarthritis knee semi-

quantitative score (MOAKS) [28] was calculated to quantify the degree of patellofemoral

osteoarthritis (PFOA). PFOA was defined [27] by the presence (>1) within the patella and/or

trochlear of i) a definitive osteophyte ii) partial or full thickness cartilage loss.

Statistical analysis

Statistical analysis was carried out in SPSS software, version 21.0 (Armonk, NY: IBM Corp).

The reliance on solely data-driven cluster analysis can lead to subgroups being difficult to

interpret and apply clinically [34]. As a result, a two-stage approach advocated by Kent et al

(2015) [34] was applied. This approach requires variables to be classified into health domains

(Fig. 2). Guidance for this classification process was based on a previous classification for

PFP [66] with domain names revised to reflect modern understanding. The 10 selected

variables were classified into the health domains by the authors (BD and JS),

11

197

198

199

200

201

202

203

204

205

206

207

208

209

210

211

212

213

214

215

216

217

218

219

220

221

222


physiotherapists with 5 and 30 years of specialist interest in PFP respectively (Fig. 2). The

cluster analysis method used for each stage of the subgroup identification was a TwoStep

cluster analysis (TwoStep CA).

First stage of clustering

The first stage of clustering was performed only within each health domain. Clustering was

conducted using the TwoStep CA analysis using a log-likelihood similarity measure. The

optimal number of cluster solutions was derived using the Schwartz Bayesian Information

Criterion (BIC). Prior to performing the cluster analysis, data variance, normality and outliers

were checked. After completing the first stage of clustering, cross tabulation (see APPENDIX

2) was done to observe how the variables were distributed across the clusters within each

domain. We compared variables between clusters to inform cluster interpretation, at a 2-

tailed significance (p<0.05). Independent samples t-tests (for two clusters), and ANOVA (for

greater than 2 clusters) with Tukey post-hoc tests were performed for continuous variables.

Chi-squared tests with pairwise multiple comparisons were calculated for categorical

variables. Labelling the clusters (hereafter groups) was guided predominantly by statistical

differences and normative means derived from published literature (Table 3). For both BSO

and PTA, previous thresholds have been established under a full-weight bearing protocol [48]

but only BSO was shown to differ under non-weight bearing [16]. To account for these

procedural differences, a discretionary extra 5% was added to the published thresholds for

BSO.

Second stage of clustering

12

223

224

225

226

227

228

229

230

231

232

233

234

235

236

237

238

239

240

241

242

243

244

245

246

247


Using the first stage domains, a second stage clustering was performed to identify groups

across all domains. TwoStep CA is also known to be sensitive to the order of cases [33] so a

random number generated order was computed and kappa coefficients (k) calculated to

compare with the original order. Similar to the first stage, the groups were cross tabulated

and compared clinically and statistically between groups as above. The stability of cluster

solutions from both stages was examined against a hierarchical cluster analysis (HCA)

performed using Wards methods, with a squared Euclidean distance similarity measure and

standardized to Z scores. Kappa coefficients were calculated to quantify the stability

between TwoStep CA and HCA methods and interpreted using recognised criteria [36].

Subgroup prognosis

An exploratory analysis was used to determine the prognosis of the eventual subgroups.

Logistic regression was applied with a 11-point Global Rating of Change Scale (GROC) as

an outcome ( anchor points ‘very much worse’ [-5] to ‘completely recovered’ [+5])

dichotomised into favourable outcome (≥ 2 points) and unfavourable outcome (<2 points) [30]

with p < 0.1 used as a criterion for potential differences. The model was adjusted for factors

known to influence prognosis in PFP [8, 37] which included duration of symptoms (categorised

into 3-12 months; greater than 12 months) and baseline Anterior Knee Pain Score (AKPs)[35]

(continuous outcome). Treatment attendance (categorised into Yes or No) was also adjusted

for as some participants will have received their own treatment or were involved within the

nested feasibility study [18]. For those participants lost to follow-up without GROC and

treatment attendance data, outcomes were estimated using multiple imputations on the

assumption that data was missing at random. Other data missing at follow up did not impact

on the analysis. Twenty imputed datasets were created [23] using monotone regression . In

addition to the final analysis model variables, the imputation model included previous

treatment at baseline and worst numerical rating score (NRS) as predictive auxiliary

13

248

249

250

251

252

253

254

255

256

257

258

259

260

261

262

263

264

265

266

267

268

269

270

271

272

273

274


variables. A sensitivity analysis was conducted examining the differences between the

original and imputed datasets.

Results

Participants

In total, 148 participants were invited to participate in the study. Twenty-four of these

declined to take part and 47 were excluded following eligibility screening. Seventy-seven

participants were consented to the study. Based on previous procedures for PFP recruitment

[9], the MRI reports for all participants were checked for competing diagnoses e.g. patellar

tendinopathy and seven patients were excluded at this stage. Seventy participants were

included in the cluster analyses at baseline; Table 2 shows their characteristics. Of these 70

participants, 58 completed outcomes at 12 months (dropout rate 17%). There were no

substantive baseline differences between the participants who dropped out (n=12) and those

that completed the outcomes at 12 months (n=58) (APPENDIX 2). Of the 58 participants at

12 months, 50% (29/58) reported having treatment which included those in the nested

feasibility study [18].

Figure 2 shows the results at each stage of clustering. The cluster patterns of each domain

are shown in APPENDIX 3. During the second cluster stage, 4 groups were identified from

the cluster patterns. This cluster solution was compared to the random case ordering

showing a substantial agreement [36] between orders, k = 0.68 ( p< 0.001, 95% Confidence

Intervals [CI] 0.55, 0.81). Comparison with the output from the HCA cluster solution showed

14

275

276

277

278

279

280

281

282

283

284

285

286

287

288

289

290

291

292

293

294

295

296

297

298


a moderate agreement, k = 0.59 (p < 0.001, 95% CI 0.46, 0.73) (APPENDIX 4). Profiling of

the clusters identified the following groups (Table 4):

Strong subgroup

The Strong subgroup showed a significantly greater hip abductor (1.8 Nm/kg) and knee

extensor strength (2.1 Nm/kg) compared to each of the other subgroups. Their

gastrocnemius flexibility was also significantly greater compared to the Weak subgroup (40.5

vs 35.0). Based on the clinical thresholds, this subgroup demonstrated all variables within

normal limits including the strength measures. This subgroup had significantly more males

(59.3% vs 9.1%) and significantly less functional disability (i.e. higher AKP) compared to the

Weak subgroup (82.4 vs 73 points).

Pronation & Malalignment subgroup

The Pronation & Malalignment subgroup showed the largest BSO (73.6%) and FPI (8.0)

which was statistically significant compared to each of the other groups. Based on the

clinical thresholds, this subgroup showed a mean BSO and FPI value that exceeded the

defined thresholds suggesting increased patellar malalignment and foot pronation. In

addition, clinically they exceeded the defined normative threshold for PTA (10.3) and

demonstrated marked weakness in hip abductor strength (< 2 SD) and moderate weakness

in knee extensor strength (< 1 SD). This subgroup showed the highest BMI (29.4 kg/m2) and

duration of symptoms (73.9 months) but neither were statistically significantly different

across groups. Within this subgroup, 30% and 40% of the group had MRI defined

osteophytes and PFOA respectively. Compared to all other subgroups, both these variables

were considered statistically significance (p = 0.02).

15

299

300

301

302

303

304

305

306

307

308

309

310

311

312

313

314

315

316

317

318

319

320

321

322

323

324


Weak subgroup

The Weak subgroup demonstrated the least hip abductor strength (1.1 Nm/kg) and knee

extensor strength (1.0 Nm/kg) but with only the hip abductor strength showing statistically

significant differences between the Active & Flexible and Strong subgroups. This subgroup

also demonstrated the least gastrocnemius flexibility which was significantly lower compared

to the Strong subgroup. Based on the clinical thresholds, this subgroup demonstrated

marked weakness for both hip abductor (< 2 SD) and knee extensor strength (<2 SD) with all

other variables within normal limits. This group had significantly more females (90.9%), with

the lowest AKP. They also demonstrated significantly the lowest physical activity compared

to the Flexible subgroup (1.7 vs 4.9 hours/week).

Active & Flexible subgroup

The Active & Flexible subgroup demonstrated the greatest quadriceps (135.7) and

gastrocnemius (43.4) with quadriceps being statistically different to each of the other groups

and gastrocnemius statistically different to the Weak subgroup. Based on the clinical

thresholds, this subgroup demonstrated a moderate increase in flexibility for gastrocnemius

(> 1 SD) and hamstrings (> 1 SD) in addition to a moderate weakness in hip abductor

strength (< 1 SD) and knee extensor strength (< 1 SD). All other variables were within

normal limits including quadriceps flexibility. This group was significantly more physically

active compared to the Weak subgroup.

16

325

326

327

328

329

330

331

332

333

334

335

336

337

338

339

340

341

342

343

344

345

346

347

348


Prognosis of subgroups

No clinically meaningful differences were noted between the original and imputed results

(see APPENDIX 5) thus the imputed dataset is presented. The results of the exploratory

logistic regression (Table 5), using the largest subgroup (Strong: 55% [15/27] favourable

outcome) as the reference, showed that there were no statistically significant differences

between the groups in the odds of a favourable outcome. Overall the grouping variable was

not statistically significant for predicting a favourable outcome (p=0.26). Descriptively the

Weak (31% [7/22]; Odds Ratio [OR] 0.30; 95% CI 0.07, 1.36) and the Pronation &

Malalignment (50% [5/10]; OR 0.64, 95% CI 0.11,3.66) subgroups were less likely to report

a favourable outcome at 12 months. However, the Active & Flexible subgroup (63% [7/11];

OR 1.24, 95% CI 0.20,7.51) were more likely to report a favourable outcome. None of the

subgroups met the p< 0.1 criterion. An unadjusted sensitivity analysis (APPENDIX 6 ),

removing the Anterior Knee Pain score, showed that the Weak subgroup had statistically

significantly lower odds of a favourable outcome (p=0.04). However, the adjusted analysis is

reported due to the known prognostic effect of baseline pain on PFP[38].

Discussion

This study has demonstrated that 4 subgroups within a PFP cohort can be identified using

clinical, biomechanical and imaging features that are potentially modifiable and therefore

amenable to treatment. This is the first time that prognostic outcomes of cluster-analysis

derived PFP subgroups have been explored. There were no statistically significant

differences between the subgroups in the odds of a favourable outcome; however,

17

349

350

351

352

353

354

355

356

357

358

359

360

361

362

363

364

365

366

367

368

369

370

371

372


descriptively the Weak subgroup were the least, and the Flexible subgroup the most, likely

to report a favourable outcome at 12 months.

The 4 subgroups identified in the current study are comparable to 4 empirical subgroups

identified previously but these were not derived using a statistical clustering approach [31, 56].

Selfe et al [55] also used cluster analysis, within a cross-sectional study, identifying three

subgroups: Weak & Tighter group, Strong group and Weak & Pronated group. The clinical

tests they used included the same measures of quadriceps flexibility, gastrocnemius

flexibility and FPI as used in the current study. Knee extensor and hip abductor strength was

also used but was measured isometrically. Despite identifying only three groups, the groups

identified by Selfe et al [55] showed many similarities to those of the current study. Both

strong groups demonstrated high strength measures and were comprised predominantly of

men. The Weak subgroup show similarities with the Selfe et al [55] Weak & Tighter group in

terms of lowest strength, physical activity and poor functional disability. The Pronation &

Malalignment and Active & Flexible subgroups in the current study show some similarities

with the Selfe et al [55] Weak & Pronated group in terms of high FPI and the greatest

gastrocnemius flexibility. The lack of agreement between studies is likely the result of slight

variations in statistical methodology and the fact that the current study incorporated imaging

and biomechanical features. Furthermore, the Selfe et al [55] study had a slightly lower mean

age (26 years) and a higher overall proportion of females (84%) which may have also

contributed to the different categories. Nevertheless, these findings combined provide further

support for the existence of PFP subgroups, thereby providing a basis for a stratified

rehabilitation management approach.

This is the first study to investigate the long-term outcome of statistically derived PFP

subgroups using clinical, biomechanical and imaging features. The Weak subgroup were

18

373

374

375

376

377

378

379

380

381

382

383

384

385

386

387

388

389

390

391

392

393

394

395

396

397

398


substantively more likely to report an unfavourable outcome at 12 months, however, on the

basis of our exploratory results, the difference in the Weak subgroup did not meet the p <0.1

criterion (p= 0.12). The result of the unadjusted sensitivity analysis did show a statistical

difference for the Weak subgroup, however, there is a need to show usefulness above and

beyond baseline pain, otherwise basline pain could just be measured with no need to collect

the other features. Furthermore, the current analyses only observe the natural history of the

subgroups. It should still be possible to design tailored interventions for these modifiable

features. The Weak subgroup were found to have the weakest hip abductor and knee

extensor strength compared the other groups and report the least physical activity and worst

AKP scores. Less knee strength [44, 50] and poor baseline function [37] have previously been

shown to lead to a poor long-term response to treatment in PFP. The Flexible subgroup

which showed the greatest flexibility in quadriceps and gastrocnemius were also the most

physically active and were the most likely to report a favourable outcome. This group may

represent people who, due to their increased physical activity, are transiently exceeding joint

loading which has been linked to an increase in PFP symptoms [26]. Simple activity

modification may have explained the improvement in this group, however this is difficult

elucidate from the data available. Only one other study [31] has investigated the long-term

follow up of PFP subgroups, which were empirically derived. In a three-year follow up, Keays

et al (2015) [31] reported no improvement in pain for any of their four subgroups: hypermobile

stance group; hypomobile group; faulty movement pattern group; and PFOA group. The fact

that groups were derived empirically rather than statistically and from a wider age group of

participants (13-82 years) does restrict a direct comparison with the current study.

The identification of diagnostic subgroups provides the opportunity for a range of

interventions to be matched accordingly [32] and largely confirms the subgroups identified by

Selfe et al [55] . With the exception of the Strong subgroup, all the other groups were

19

399

400

401

402

403

404

405

406

407

408

409

410

411

412

413

414

415

416

417

418

419

420

421

422

423

424

425


considered clinically weak and so provides a rationale for continuing to prescribe standard

knee and hip strengthening based exercises associated with current practice [58]. The Strong

subgroup showed normal strength levels and thus are unlikely to gain any further benefit

from additional routine strengthening exercises [18]. This subgroup may instead be better

targeted with movement retraining based interventions which have been shown to be

effective in runners with PFP [45, 46, 64]. The Pronation & Malalignment subgroup demonstrated

excessive structural features (largest BSO and FPI) and therefore might benefit from passive

interventions such as knee braces and foot orthotics which have been showed to reduce

BSO [15] and FPI [2] respectively. Importantly, the MRI BSO only uses a single axial slice

which was originally developed using radiographs and could be easily measured from

standard x-rays. From a clinical service provision viewpoint, these prognostic findings

highlight who might be unlikely to benefit from additional treatment [25, 29]. Our findings

suggest that the most active, Active & Flexible subgroup may represent a self-limiting form of

PFP which may require simple advice on load management [19] and limited follow up. In

contrast, the Weak subgroup may require increased service provision with more

physiotherapy input. The prognosis of PFP subgroups remains a research priority and

further evaluation of other datasets is required before these results can be applied within

clinical practice.

Our findings are based on a relatively small cohort, however, the use of a rule of thumb [21]

for cluster analysis was intended to minimise the over fitting of data. Furthermore, to handle

this problem the two-stage cluster approach used is a dimensionality reduction technique,

allowing more features to be analysed within shared domains whilst also aiding the cluster

interpretation [34]. Sample size for logistic regression met a bare minimum requirement of 5

events per variable [61], but for a confirmatory trial we would wish to increase this to at least

10 i.e. doubling the required sample size. Strict criteria were used for the selection of

20

426

427

428

429

430

431

432

433

434

435

436

437

438

439

440

441

442

443

444

445

446

447

448

449

450

451

452


variables; however, it is known that other kinematic variables would have satisfied the

selection criteria based on different assessment tasks such as running. Stair descent was

selected as its achievable for both active and sedentary individuals thus identified subgroups

are likely to represent the wider population. In terms of treatment, some of the participants

were enrolled into a nested feasibility study focused on targeted hip strengthening within the

12 months. To account for this difference, treatment was adjusted for in our regression

model but there is still the potential for some residual confounding.

Further research is required to see if these subgroups can be replicated in larger PFP data

sets and the matching of treatment to the respective subgroup features needs investigating.

Despite the potential clinical implications, our study design does not allow us to conclude

that a stratified treatment approach would be effective at this stage; this would need to be

established in future research.

Conclusions

This study suggests that using modifiable clinical, biomechanical and imaging features, four

PFP subgroups can be indentified. The results show preliminary evidence that outcomes at

12 months may differ between subgroups, thus justifying progress to a larger-scale

confirmatory trial. These PFP subgroups were consistent with and extend the subgroupings

proposed by Selfe et al [55] The identification of subgroups provides the opportunity for a

range of interventions to be matched accordingly, and further research is now warranted to

determine whether a stratified treatment approach using these subgroups would be

efficacious. Furthermore, this is the first time that prognostic outcomes of statistically derived

21

453

454

455

456

457

458

459

460

461

462

463

464

465

466

467

468

469

470

471

472

473

474

475

476


PFP subgroups have been described which improves our understanding of potentially

differing mechanisms underpinning the overall PFP presentation.

Key points

Findings : By combining modifiable clinical, biomechanical and imaging features using

cluster analysis, four PFP subgroups have been identified: ‘Strong’, ‘Pronation &

Malalignment’, ‘Weak’ and ‘Active & Flexible’. No statistically significant differences were

found between subgroups at 12 months outcomes, however, based on the exploratory

analyses potentially different outcomes may exist with the Weak subgroup the least and the

Active & Flexible subgroup most likely to report a favourable outcome at 12 months.

Implications: Clinical service provision and research priorities could be tailored to account

for these observed differences in outcome.

Caution: Our findings are based on a relatively small cohort and therefore further research

is warranted to see if these subgroups can be validated in larger PFP data sets.

22

477

478

479

480

481

482

483

484

485

486

487

488

489

490

491

492

493

494

495


References

1. Barton CJ, Levinger P, Menz HB, and Webster KE. Kinematic gait characteristics associated with patellofemoral pain syndrome: a systematic review. Gait & posture. 2009; 30(4). 405-16.

2. Barton CJ, Levinger P, Crossley KM, Webster KE, and Menz HB. Relationships between the Foot Posture Index and foot kinematics during gait in individuals with and without patellofemoral pain syndrome. J Foot Ankle Res. 2011; 4(1). 10.

3. Barton CJ, Lack S, Hemmings S, Tufail S, and Morrissey D. The ‘Best Practice Guide to Conservative Management of Patellofemoral Pain’: incorporating level 1 evidence with expert clinical reasoning. Br J Sports Med. 2015. bjsports-2014-093637.

4. Bennett MI, Smith BH, Torrance N, and Potter J. The S-LANSS score for identifying pain of predominantly neuropathic origin: validation for use in clinical and postal research. The Journal of Pain. 2005; 6(3). 149-58.

5. Boling MC, Padua DA, and Alexander Creighton R. Concentric and eccentric torque of the hip musculature in individuals with and without patellofemoral pain. J Athl Train. 2009; 44(1). 7-13.

6. Bovi G, Rabuffetti M, Mazzoleni P, and Ferrarin M. A multiple-task gait analysis approach: kinematic, kinetic and EMG reference data for healthy young and adult subjects. Gait & posture. 2011; 33(1). 6-13.

7. Callaghan MJ and Selfe J. Patellar taping for patellofemoral pain syndrome in adults. Cochrane Database Syst Rev. 2012; 4.

8. Collins NJ, Bierma-Zeinstra SM, Crossley KM, van Linschoten RL, Vicenzino B, and van Middelkoop M. Prognostic factors for patellofemoral pain: a multicentre observational analysis. Br J Sports Med. 2013; 47(4). 227-33.

9. Cook C, Hegedus E, Hawkins R, Scovell F, and Wyland D. Diagnostic accuracy and association to disability of clinical test findings associated with patellofemoral pain syndrome. Physiother Can. 2010; 62(1). 17-24.

10. Crossley KM, Cowan SM, Bennell KL, and McConnell J. Knee flexion during stair ambulation is altered in individuals with patellofemoral pain. Journal of Orthopaedic Research. 2004; 22(2). 267-74.

11. Crossley KM. Is patellofemoral osteoarthritis a common sequela of patellofemoral pain? Br J Sports Med. 2014; 48(6). 409-10.

12. Crossley KM, Stefanik JJ, Selfe J, et al. 2016 Patellofemoral pain consensus statement from the 4th International Patellofemoral Pain Research Retreat, Manchester. Part 1: Terminology, definitions, clinical examination, natural history, patellofemoral osteoarthritis and patient-reported outcome measures. Br J Sports Med. 2016.

13. Danneskiold‐Samsøe B, Bartels E, Bülow P, et al. Isokinetic and isometric muscle strength in a healthy population with special reference to age and gender. Acta Physiol (Oxf). 2009; 197(s673). 1-68.

23

496

497498499

500501502

503504505

506507508

509510511

512513514

515516

517518519

520521522

523524525

526527

528529530531532

533534535


14. Dierks TA, Manal KT, Hamill J, and Davis I. Lower extremity kinematics in runners with patellofemoral pain during a prolonged run. Med Sci Sports Exerc. 2011; 43(4). 693-700.

15. Draper CE, Besier TF, Santos JM, et al. Using real-time MRI to quantify altered joint kinematics in subjects with patellofemoral pain and to evaluate the effects of a patellar brace or sleeve on joint motion. J Orthop Res. 2009; 27(5). 571-7.

16. Draper CE, Besier TF, Fredericson M, et al. Differences in patellofemoral kinematics between weight‐bearing and non‐weight‐bearing conditions in patients with patellofemoral pain. Journal of Orthopaedic Research. 2011; 29(3). 312-17.

17. Drew BT, Redmond AC, Smith TO, Penny F, and Conaghan PG. Which patellofemoral joint imaging features are associated with patellofemoral pain? Systematic review and meta-analysis. Osteoarthritis Cartilage. 2015.

18. Drew BT, Conaghan PG, Smith TO, Selfe J, and Redmond AC. The effect of targeted treatment on people with patellofemoral pain: a pragmatic, randomised controlled feasibility study. BMC Musculoskelet Disord. 2017; 18(1). 338.

19. Esculier J-F, Bouyer LJ, Dubois B, et al. Is combining gait retraining or an exercise programme with education better than education alone in treating runners with patellofemoral pain? A randomised clinical trial. Br J Sports Med. 2017. bjsports-2016-096988.

20. Feiring DC, Ellenbecker TS, and Derscheid GL. Test-retest reliability of the Biodex isokinetic dynamometer. J Orthop Sports Phys Ther. 1990; 11(7). 298-300.

21. Formann AK, Die latent-class-analyse: Einführung in Theorie und Anwendung. 1984: Beltz.

22. Goudakos IG, Konig C, Schottle PB, et al. Regulation of the patellofemoral contact area: an essential mechanism in patellofemoral joint mechanics? J Biomech. 2010; 43(16). 3237-9.

23. Graham JW, Olchowski AE, and Gilreath TD. How many imputations are really needed? Some practical clarifications of multiple imputation theory. Prevention science. 2007; 8(3). 206-13.

24. Harbaugh CM, Wilson NA, and Sheehan FT. Correlating femoral shape with patellar kinematics in patients with patellofemoral pain. J Orthop Res. 2010; 28(7). 865-72.

25. Hill JC, Whitehurst DG, Lewis M, et al. Comparison of stratified primary care management for low back pain with current best practice (STarT Back): a randomised controlled trial. The Lancet. 2011; 378(9802). 1560-71.

26. Ho KY, Hu HH, Colletti PM, and Powers CM. Recreational runners with patellofemoral pain exhibit elevated patella water content. Magn Reson Imaging. 2014; 32(7). 965-8.

27. Hunter D, Arden N, Conaghan P, et al. Definition of osteoarthritis on MRI: results of a Delphi exercise. Osteoarthritis Cartilage. 2011; 19(8). 963-69.

28. Hunter DJ, Guermazi A, Lo GH, et al. Evolution of semi-quantitative whole joint assessment of knee OA: MOAKS (MRI Osteoarthritis Knee Score). Osteoarthritis Cartilage. 2011; 19(8). 990-1002.

24

536537538

539540541

542543544

545546547

548549550

551552553554

555556

557558

559560561

562563564

565566

567568569

570571572

573574

575576577


29. Industry AotBP, Stratified medicine in the NHS: An assessment of the current landscape and implementation challenges for non-cancer applications, 2014: United Kingdom.

30. Kamper SJ, Maher CG, and Mackay G. Global rating of change scales: a review of strengths and weaknesses and considerations for design. J Man Manip Ther. 2009; 17(3). 163-70.

31. Keays SL, Mason M, and Newcombe PA. Individualized physiotherapy in the treatment of patellofemoral pain. Physiotherapy Research International. 2015; 20(1). 22-36.

32. Kent P, Keating JL, and Leboeuf-Yde C. Research methods for subgrouping low back pain. BMC Med Res Methodol. 2010; 10(1). 62.

33. Kent P, Jensen RK, and Kongsted A. A comparison of three clustering methods for finding subgroups in MRI, SMS or clinical data: SPSS TwoStep Cluster analysis, Latent Gold and SNOB. BMC Med Res Methodol. 2014; 14(1). 113.

34. Kent P, Stochkendahl MJ, Christensen HW, and Kongsted A. Could the clinical interpretability of subgroups detected using clustering methods be improved by using a novel two-stage approach? Chiropractic & manual therapies. 2015; 23(1). 1.

35. Kujala UM, Jaakkola LH, Koskinen SK, Taimela S, Hurme M, and Nelimarkka O. Scoring of patellofemoral disorders. Arthroscopy. 1993; 9(2). 159-63.

36. Landis JR and Koch GG. The measurement of observer agreement for categorical data. Biometrics. 1977. 159-74.

37. Lankhorst N, van Middelkoop M, Crossley K, et al. Factors that predict a poor outcome 5–8 years after the diagnosis of patellofemoral pain: a multicentre observational analysis. Br J Sports Med. 2015. bjsports-2015-094664.

38. Lankhorst N, van Middelkoop M, Crossley K, et al. Factors that predict a poor outcome 5–8 years after the diagnosis of patellofemoral pain: a multicentre observational analysis. Br J Sports Med. 2015. bjsports-2015-094664.

39. Lankhorst NE, Bierma-Zeinstra SM, and van Middelkoop M. Factors associated with patellofemoral pain syndrome: a systematic review. Br J Sports Med. 2012. bjsports-2011-090369.

40. Matthews M, Rathleff M, Claus A, et al. Can we predict the outcome for people with patellofemoral pain? A systematic review on prognostic factors and treatment effect modifiers. Br J Sports Med. 2016. bjsports-2016-096545.

41. Mills K, Blanch P, Dev P, Martin M, and Vicenzino B. A randomised control trial of short term efficacy of in-shoe foot orthoses compared with a wait and see policy for anterior knee pain and the role of foot mobility. Br J Sports Med. 2012; 46(4). 247-52.

42. Mølgaard C, Rathleff MS, and Simonsen O. Patellofemoral pain syndrome and its association with hip, ankle, and foot function in 16-to 18-year-old high school students: a single-blind case-control study. J Am Podiatr Med Assoc. 2011; 101(3). 215-22.

43. Mølgaard CM, Rathleff MS, Andreasen J, et al. Foot exercises and foot orthoses are more effective than knee focused exercises in individuals with patellofemoral pain.

25

578579580

581582583

584585586

587588

589590591

592593594

595596

597598

599600601

602603604

605606607

608609610

611612613

614615616617

618619


Journal of Science and Medicine in Sport. 2017.

44. Natri A, Kannus P, and Järvinen M. Which factors predict the long-term outcome in chronic patellofemoral pain syndrome? A 7-yr prospective follow-up study. Medicine & Science in Sports & Exercise. 1998.

45. Neal BS, Barton CJ, Gallie R, O’Halloran P, and Morrissey D. Runners with patellofemoral pain have altered biomechanics which targeted interventions can modify: A systematic review and meta-analysis. Gait & posture. 2016; 45. 69-82.

46. Noehren B, Scholz J, and Davis I. The effect of real-time gait retraining on hip kinematics, pain and function in subjects with patellofemoral pain syndrome. Br J Sports Med. 2010. bjsports69112.

47. O'sullivan K, O'sullivan P, Fersum KV, and Kent P, Better targeting care for individuals with low back pain: opportunities and obstacles, 2016, BMJ Publishing Group Ltd and British Association of Sport and Exercise Medicine.

48. Pal S, Draper CE, Fredericson M, et al. Patellar maltracking correlates with vastus medialis activation delay in patellofemoral pain patients. The American journal of sports medicine. 2011; 39(3). 590-98.

49. Pal S, Besier TF, Beaupre GS, Fredericson M, Delp SL, and Gold GE. Patellar maltracking is prevalent among patellofemoral pain subjects with patella alta: an upright, weightbearing MRI study. Journal of Orthopaedic Research. 2013; 31(3). 448-57.

50. Payton C and Bartlett R, Biomechanical evaluation of movement in sport and exercise: the British Association of Sport and Exercise Sciences guide. 2007: Routledge.

51. Redmond AC, Crane YZ, and Menz HB. Normative values for the foot posture index. J Foot Ankle Res. 2008; 1(1). 1.

52. Reilly MC, Zbrozek AS, and Dukes EM. The validity and reproducibility of a work productivity and activity impairment instrument. Pharmacoeconomics. 1993; 4(5). 353-65.

53. Santos CMd, Ferreira G, Malacco PL, Sabino GS, Moraes GFdS, and Felício DC. Intra and inter examiner reliability and measurement error of goniometer and digital inclinometer use. Revista Brasileira de Medicina do Esporte. 2012; 18(1). 38-41.

54. Selfe J, Callaghan M, Witvrouw E, et al. Targeted interventions for patellofemoral pain syndrome (TIPPS): classification of clinical subgroups. BMJ open. 2013; 3(9). e003795.

55. Selfe J, Janssen J, Callaghan M, et al. Are there three main subgroups within the patellofemoral pain population? A detailed characterisation study of 127 patients to help develop targeted intervention (TIPPs). Br J Sports Med. 2016. bjsports-2015-094792.

56. Selhorst M, Rice W, Degenhart T, Jackowski M, and Tatman M. Evaluation of a treatment algorithm for patients with patellofemoral pain syndrome: a pilot study. Int J Sports Phys Ther. 2015; 10(2). 178.

26

620

621622623

624625626

627628629

630631632

633634635

636637638639

640641642

643644

645646647

648649650

651652653

654655656657

658659660


57. Sheehan FT, Derasari A, Brindle TJ, and Alter KE. Understanding patellofemoral pain with maltracking in the presence of joint laxity: complete 3D in vivo patellofemoral and tibiofemoral kinematics. J Orthop Res. 2009; 27(5). 561-70.

58. Smith BE, Hendrick P, Bateman M, et al. Current management strategies for patellofemoral pain: an online survey of 99 practising UK physiotherapists. BMC Musculoskelet Disord. 2017; 18(1). 181.

59. Smits-Engelsman B, Klerks M, and Kirby A. Beighton score: a valid measure for generalized hypermobility in children. The Journal of pediatrics. 2011; 158(1). 119-23. e4.

60. Venturni C, André A, Aguilar BP, and Giacomelli B. Reliability of two evaluation methods of active range of motion in the ankle of healthy individuals. Acta Fisiátrica. 2006; 13(1). 39-43.

61. Vittinghoff E and McCulloch CE. Relaxing the rule of ten events per variable in logistic and Cox regression. Am J Epidemiol. 2007; 165(6). 710-18.

62. Von Elm E, Altman DG, Egger M, et al. The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) Statement: guidelines for reporting observational studies. International Journal of Surgery. 2014; 12(12). 1495-99.

63. Watari R, Kobsar D, Phinyomark A, Osis S, and Ferber R. Determination of patellofemoral pain sub-groups and development of a method for predicting treatment outcome using running gait kinematics. Clinical Biomechanics. 2016; 38. 13-21.

64. Willy RW and Davis IS. Varied response to mirror gait retraining of gluteus medius control, hip kinematics, pain, and function in 2 female runners with patellofemoral pain. J Orthop Sports Phys Ther. 2013; 43(12). 864-74.

65. Witvrouw E, Lysens R, Bellemans J, Cambier D, and Vanderstraeten G. Intrinsic risk factors for the development of anterior knee pain in an athletic population a two-year prospective study. The American journal of sports medicine. 2000; 28(4). 480-89.

66. Witvrouw E, Werner S, Mikkelsen C, Van Tiggelen D, Berghe LV, and Cerulli G. Clinical classification of patellofemoral pain syndrome: guidelines for non-operative treatment. Knee Surgery, Sports Traumatology, Arthroscopy. 2005; 13(2). 122-30.

67. Witvrouw E, Callaghan MJ, Stefanik JJ, et al. Patellofemoral pain: consensus statement from the 3rd International Patellofemoral Pain Research Retreat held in Vancouver, September 2013. Br J Sports Med. 2014; 48(6). 411-4.

68. Wood L, Muller S, and Peat G. The epidemiology of patellofemoral disorders in adulthood: a review of routine general practice morbidity recording. Prim Health Care Res Dev. 2011; 12(2). 157-64.

69. Youdas JW, Krause DA, Hollman JH, Harmsen WS, and Laskowski E. The influence of gender and age on hamstring muscle length in healthy adults. J Orthop Sports Phys Ther. 2005; 35(4). 246-52.

27

661662663

664665666

667668669

670671672

673674

675676677

678679680681

682683684

685686687

688689690

691692693

694695696

697698699

700


28

701


Figure captions

FIGURE 1: Assessment procedures for the selected variables

a) MRI patella tilt: angle formed by a line between AB and EF; b) MRI Bisect offset: length of AC / length of BC x 100%; c) Stair descent: each participant completed a minimum 5 successful stair descents at a self-selected speed; d) Hamstring flexibility: supine with knee and hip at 90°. The knee is passively extended with an average of 3 digital inclinometer readings on the tibia; e) Quadriceps flexibility: prone with the contralateral limb on the floor at 90° hip flexion. The knee is passively flexed with an average of 3 digital inclinometer readings on the tibia; f) Gastrocnemius flexibility: 0.6m from the wall with both toes pointing forwards. The index limb behind extended is actively flexed at the ankle keeping the heel on the floor with an average of 3 digital inclinometer readings on the tibia; g) Foot Posture Index: a 6-item clinical tool that was measured in double stance; h) Biodex knee extension: performed at 60°/s with participants having 3 submaximal (50%) practice trials before commenced 5 trials of maximal repetitions; i) Biodex hip abduction: performed at 30°/s with participants having 3 submaximal (50%) practice trials before commenced 5 trials of maximal repetitions.

FIGURE 2: Two stage cluster approach

29

702

703

704

705

706707708709710711712713714715716717718719

720

721

722

723

724

725

726


TABLE 1: Participant eligibility criteria

Inclusion

1 Aged 18-40 years

2 Insidious onset of anterior or retropatellar knee pain

3 Pain on 2 or more of the following activities: prolonged sitting, kneeling, squatting, running, patellar palpation, hopping, stair walking, stepping down or isometric quadriceps contraction [12]

4 Pain for greater than 6 weeks duration

Exclusion

4 Clinical examination showed another cause of knee pain such as, but not restricted to: meniscal pathologies, quadriceps tendon injuries, patella tendinopathy, tibial tubercle apophysitis; bursitis

5 History of significant knee surgery

6 Competing pathology identified on the MRI report [9]

7 Contraindication to MRI

8 Physiotherapy or podiatric treatment within the last 3 months

30

727

728

729

730

731


TABLE 2: Participant characteristics and descriptors. Values are means (SD) unless stated otherwise

Characteristics Baseline cohort (n=70)

Age (years) 31.03 (5.32)

Females, n (%) 43 (61.4)

BMI (kg/m2) 26.25 (5.52)

Height (m) 1.71 (0.09)

Weight (kg) 76.65 (18.57)

Physical activity level (hours/week)* 3.12 (2.59)

Median (interquartile range) duration of knee pain (months) 35.50 (18.0-73.5)

Received previous treatment, n (%) 53 (75.7)

Bilateral knee pain, n (%) 36 (51.4)

Beightons score (/9)[59] 2.75 (2.48)

Anterior Knee Pain Scale 77.19 (11.73)

Worst pain 4.59 (2.28)

Average pain 2.96 (1.83)

Joint crepitus, n (%) 40 (57.1)

% Impact on work productivity (WPAIQ subscale) [52] 15.48 (23.02)

S-LANSS [4] 5.16 (5.41)

WPAIQ: Work Productivity and Activity Impairment Questionnaire; S-LANSS: Self completed Leeds Assessment of Neuropathic Symptoms and

Signs Pain Scale; NRS: Numerical Rating Scale

* Based on a subscale of the Global Physical Activity Questionnaire) – How much time (hours) do you spend doing moderate intensity ( causes at

least a small increase in breathing or heart rate) sport, fitness or recreational activities on a typical week?

31

732733

734


TABLE 3: Subgrouping variable mean (SD) and normative data or defined thresholds

Subgrouping Variable Mean (SD)Normative data

- 2 SD - 1 SD Mean + 1 SD

Peak hip abductor strength (Nm/kg) [13] 1.5 (0.4) 1.4 1.6 2.1 2.5

Peak knee extensor strength (Nm/kg) [13 ] 1.5 (0.6) 1.2 1.9 2.4 2.9

Peak angle hip internal rotation (°) [6] - 8.8 (5.6) -27.6 - 16.8 -5.8 4.8

Peak knee flexion angle (°) [6] 74.9 (10.0) 54.8 64.1 73.4 82.7

Quadriceps flexibility (°) [65] 125.1 (10.1) 99.4 115.8 132.2 148.6

Gastrocnemius flexibility (°) [65] 38.8 (6.8) 22.0 28.6 35.2 41.8

Hamstring flexibility (°)[69] * 154.0 (10.1) 127.1 136.9 146.7 156.5

Defined Threshold

Foot posture index [51] 4.3 (2.9) 6

MRI bisect offset (%) [16, 49] * 57.2 (7.4) 68.3% (65 +5%)

MRI patella tilt (°)[16]* 8.7(4.5) 9

* Gender specific thresholds combined

32

735


TABLE 4: Mean values (SD) across the four subgroups

Variables

2nd stage subgroups

ANOVAStrong

(n=27)

Mean (SD)

Pronation & Malalignment

(n=10)

Mean (SD)

Weak

(n=22)

Mean (SD)

Active & Flexible

(n=11)

Mean (SD)

Peak hip abductor strength(Nm/kg) 1.8 (0.3) † 1.3 (0.5) 1.1 (0.3) * 1.4 (0.2) * F =19.67 p <0.001

Peak knee extensor strength (Nm/kg) 2.1 (0.5) † 1.4 (0.6) 1.0 (0.3) 1.3 (0.1) F = 24.502 p <0.001

Peak angle hip internal rotation () -9.1 (5.4) -10.8 (6.3) -7.0 (5.1) -9.9 (5.9) F = 1.448 p = 0.24

Peak knee flexion angle () 73.5 (10.5) 73.6 (13.4) 75 (9.4) 79.6 (11.0) F = 1.032 p = 0.38

Quadriceps flexibility () 125.0 (10.3) 122.1 (11.3) 121.4 (6.1) 135.7 (8.0) † F = 6.75 p < 0.001

Gastrocnemius flexibility () 40.5 (6.8) § 37.5 (7.5) 35.0 (4.6) *§ 43.4 (6.4) * F = 5.53 p = 0.002

Hamstring flexibility () 155.4 (11.8) 154.0 (10.3) 150.3 (6.9) 158.0 (9.3) F = 1.81 p = 0.15

Foot posture index 3.7 (2.3) 8.0(2.1) † 3.6 (2.8) 3.8 (3.0) F = 8.17 p< 0.001

33

736

737

738

739


MRI bisect offset (%) 55.4 (6.3) 70.3 (4.7) † 54.1 (4.4) 55.6 (3.1) F = 25.49 p < 0.001

MRI patella tilt () 8.1 (5.2) 10.3 (4.9) 8 (4.2) 9.8 (2.8) F =0.96 p = 0.41

Patient related factors

Age (years) 30.7 (5.13) 30.6 (5.3) 30.1 (6.2) 34 (2.9) F =1.45, p=0.24

Gender (male %) 16 (59.3) * 5 (50.0) 2 (9.1) * 4 (36.4) x2 = 13.52, p =0.004

BMI (kg/m2) 24.9 (4.6) 29.4 (7.8) 27.3 (5.7) 24.8 (3.3) F =2.27, p =0.09

Physical activity (hours/week) 3.5 (2.4) 3.1 (2.6) 1.7 (1.6) * 4.9 (3.3) * F=4.713, p=0.005

Duration of pain (months) 52.3 (58.9) 73.9 (72) 59.4 (68.8) 57.7 (82.7) F=0.25, p=0.86

Previous treatment (%) 17 (63.0%) 8 (80.0%) 18 (81.8%) 10 (90.9%) x2 = 4.31, p =0.23

Bilaterality (%) 15 (55.6) 5 (50) 12 (54.6) 4 (51.4) x2 = 1.28, p =0.74

Beightons score (/9)[59] 2.1 (2.3) 3.8 (2.6) 2.8 (2.3) 3.3 (3.0) F=1.35, p=0.27

Anterior Knee Pain Score 82.4 (9.7) * 75.1 (12.4) 73 (11.5) * 74.5 (12.9) F=3.29, p=0.03

Average NRS 2.4 (1.6) 3.3 (2.7) 3.2 (1.6) 3.5 (1.6) F=1.38, p=0.26

Worst NRS 4.0 (1.9) 4.2 (3.0) 5.0 (2.3) 5.5 (2.2) F=1.42, p=0.24

S-LANSS [4] 4.1 (5.4) 5.1 (6.6) 6.9 (5.4) 4.2 (3.9) F=1.21, p=0.31

WPAIQ subscale - % impact on work productivity [52]

8.5 % (15.4) 21.25% (26.9) 26.67% (25.9) 9% (25.1) F=2.90, p =0.04

34


Supplementary clinical descriptors

Total patella mobility (mm) 12.8 (4.6) 14.8 (5.0) 12.1 (4.4) 13.5 (4.0) F =0.82, p=0.49

MRI cartilage loss (≥1) 9 (33.3) 6 (60%) 9 (40.9%) 3 (27.3%) x2 = 2.894, p =0.41

MRI osteophyte (≥1) 7 (25.9%) 4 (40%) 0 (0%) 0 (0%) x2 = 12.73, p = 0.005

PFOA (OA present %) 4 (14%) 3 (30%) 0 (0%) 0 (0%) x2 =8.81, p=0.03

Contact area (mm2) 154.1 (39.2) 109.5 (44.8) 118.0 (55.8) 127.7 (55.8) F=3.24, p=0.03

Insall-Salvati (ratio) 1.2 (0.1) * 1.4 (0.1) * 1.3 (0.2) 1.2 (0.2) F =4.19, p=0.009

† Different from each of the other three groups (p<0.05)

* Subgroup pairs different (p<0.05)

§ Subgroup pairs different (p<0.05)

WPAIQ: Work Productivity and Activity Impairment Questionnaire; S-LANSS: Self completed Leeds Assessment of Neuropathic Symptoms and Signs Pain Scale; NRS: Numerical Rating Scale

PFOA: patellofemoral osteoarthritis

35

740


TABLE 5: Multivariable binary logistic regression exploring the association between subgroups and likelihood of a favourable outcome at 12 months

SubgroupMultivariable *

OR (95% CI) † P- valuePronation & Malalignment group 0.64 (0.11,3.66) 0.62Weak group 0.30 (0.07, 1.36) 0.12Active & Flexible group 1.24 (0.20, 7.51) 0.82Duration of symptoms (<12 months) 0.08 (0.01, 0.77) 0.03Baseline AKP 1.03 (0.98, 1.09) 0.28Treatment (no treatment) 0.51 (0.15, 1.70) 0.27

* Adjusted for duration of symptoms (< 12 months); baseline AKP; treatment (no treatment)

† Reference group: Strong group

36

741


37

742

Documents

e-space.mmu.ac.uk Drew JOSPT.docx · Web viewXXXX is funded by a National Institute for Health Research (NIHR) Clinical Doctoral Research Fellowship (CDRF -2013-04-044); EMAH, XXXX