Table 2 Description of the papers

Attur 2011 [18]

Genetic/gene expression

Cohort (prosp)

complete-linkage hierarchical clustering

1: 41^a

2: 36^a

3: 86^a

1: 25^a

2: 0^a

3: 12^a

-

1: 16/41 = 39 %.

2: 8/36 = 22 %,

3: 33/86 = 38 %

-

-

-

-

Bae 2010 [19]

Imaging (photography)

Cross sectional

K-means cluster analysis

127

-

-

-

-

-

20 %^b

-

Berry 2010a [20]

Biomarker

Cohort (prosp)

Mann–Whitney u, χ², Multiple regression analysis

117

-

-

-

-

Prevalence not reported

-

-

Berry 2010b [21]

Biomarker

Cohort (prosp)

Mann–Whitney u, Multiple regression and logistic regression analysis

117

-

-

-

-

-

-

Prevalence not reported

Blumnenfeld 2013 [22]

Biomarker

Cohort (prosp)

Binary logistic regression analysis

Different in different analysis

Different in different analysis

-

-

-

Prevalence not reported

-

-

Cruz-Almeida 2013 [23]

Lab experimental (non-biomech)

Cross-sectional

Hierarchical cluster analysys

194

-

32/194 = 16 %

-

-

-

-

-

Doss 2007 [24]

Biomarker

Cross-sectional

Mann–Whitney

49

-

-

8/49 = 16 %

-

-

-

-

Egsgaard 2015 [25]

Biomarker

Case control

Principal component analysis/Hierarchical cluster analysis

216

64

41/212 = 19 %

-

-

-

-

-

Fernández-Tajes 2014 [26]

Genetics

Case control

Cluster analysys (unsupervised)

23

18

-

7/23 = 30 %

-

-

-

-

Holla 2013 [27]

Epidemiology

Cohort (prosp)

Latent class growth analysis

697

-

-

-

-

-

-

330/697 = 47 %

Jenkins 2015 [28]

Epidemiology

Secondary data analysis

Hierarchical and k -means cluster analysis

75

-

-

-

-

-

-

Prevalence not reported

Kerkhof 2008 [29]

Genetics

Cross sectional

χ², OR, ANCOVA, meta-analysis of existing cohorts

4993

-

-

-

-

-

-

-

Kinds 2013 [9]

Imaging

Cohort (prosp)

Hierarchical cluster analysys

336

-

-

-

-

-

-

108/417 = 26 %

King 2013 [30]

Lab experimental (non-biomech)

Case control

ANCOVA

209

107

Subgroups splitted using mean value of womac (percentage not reliable)

-

-

-

-

-

Knoop 2011 [7]

Epidemiology

Secondary data analysis

K-means luster analysis

842

-

83/841 = 10 % (only depression)

-

168/841 = 22 % (only obese)

-

189/841 = 22 %

140/841 = 17 %

Murphy 2011 [31]

Epidemiology

Cross-sectional

Hierarchical cluster analysis

129

-

45/125 = 36 %

-

-

-

-

-

Otterness 2000 [32]

Biomarker

Case control

Principal component analysis

39

21

-

Prevalence not reported

-

Prevalence not reported

-

-

Pereira 2013 [33]

Epidemiology

Cross-sectional

T-test, OR, logistic regression

663

-

Prevalence not reported

-

-

-

-

-

Roemer 2012 [34]

Imaging

Cross sectional

OR

1248

-

-

-

-

1248 subjects/0,2 % hypertrophic-1.3 % atrophic

-

-

Sowers 2002 [35]

Biomarker

Cohort

ANOVA, χ²

1025

-

-

-

11 %^b

-

-

-

Van der Esch 2015 [36]

Epidemiology

Secondary data anlysis

K-means cluster analysis

551

-

86/551 = 15.6 % (only depression)

-

81/551 = 15 % (only obese)

-

114/551 = 20.6 %

154/551 = 28 %

Van Spil 2012 [37]

Biomarker

Cohort (prosp)

Principal component analysis, multiple linear regression (interaction terms)

1002

-

-

Prevalence not reported

-

Prevalence not reported

-

-

Waarsing 2015 [8]

Epidemiology

Secondary data analysis

Latent class cluster analysis

518

-

-

-

27 % (group with hypertension and higher BMI)

-

15 % (lateral degeneration) 12 %(previous injuries)

47 %^b

Iijima 2015 [38]

Epidemiology

Cross sectional

Multiple Logistic regression Analysis

266

-

-

-

-

-

26/266 = 9.7 % (static + dinamic malalignment)

-

Kittelson 2015 [40]

Epidemiology

Secondary data analysis

Latent class analysis

3494

-

337/3494 = 9.6 %

-

-

-

-

-