Table 1 Analyses examining trajectories of pain in arthritis and cancer using latent class growth analyses (LCGAs) or Growth mixture models (GMMs), the trajectories identified and clinical and socio-demographic predictors

 Collins, Katz, Dervan Losina [5]

OAI cohort (osteoarthritis): LCGA with up to 4th order polynomial

5 stable trajectories: severe pain (6%), high (17%) and low (32%) moderate pain, mild pain (35%), no pain (11%) (WOMAC)

Multivariate MLR, ref. no pain: female sex, depression, graded OA severity (moderate/severe +), obesity, non-white race (severe +), pre-college education (high mod./severe +)

 Barnabe et al. [9]

CATCH cohort (early rheumatoid arthritis): LCGA with cubic polynomial

5 trajectories differing in baseline and speed of remission: high to medium (10%), low (19%) or remission (20%), medium to low (30%) or remission (21%) (DAS28)

Univariate analyses, Bonferroni corrected: age 50+, non-white, non-college education, unemployment, < $50K income, comorbidities, (non-remission trajectories +, High > Med/Low ++), symptom duration (mod. at baseline/High > Med+), DAS28, tender/swollen joints, ESR, CRP, physician, patient and pain score (High +), HAQ (high baseline and High > Med +)

 Bastick et al. [6]

CHECK cohort (assumed early OA hip respondents only): LCGA with up to quadratic polynomial

4 trajectories: mild pain (42%), moderate decrease (17%), moderate progression (24%), severe pain (16%) (NRS)

Univariable and multivariable MLR, final model: pre-university education, use of pain transformation to cope, pain with internal hip rotation (progression/severe +), WOMAC physical function (all but mild +)

 Nicholls, Thomas, van der Windt, Croft and Peat [10]

CAS-K (knee OA risk group), replicated in OAI (OA cohort): LCGA (polynomial info. reported in appendix)

5 in CAS-K: mild non-progressive (35%), progressive (28%), moderate (22%), improving (12%), severe non-improving (3%). 4 in OAI: mild, non-progressive (41%), moderate A (24%), B (19%) and C (11%), severe, non-improving (5%) (WOMAC pain)

Baseline CAS-K: differences reported on age, gender, BMI, IMD, employment, manual job, self-reported health, HADS, widespread pain, knee pain, WOMAC function, radiography, health care use inc. knee replacement

 Holla et al. [7]

CHECK cohort (early symptomatic OA, knee OA only): LCGA (no info. on polynomials reported)

3 mostly stable trajectories differing in baseline and follow-up pain: good (47%), moderate (37%) and poor outcomes (16%) (WOMAC)

Uni/multivariate analyses, ref. good outcome: age, knee flexion range (poor –), BMI (mod. +), NRS, hip pain, comorbidity (mod./poor +), SF-36 vitality (mod./poor –), bony tenderness, osteophytosis, PCI resting (poor +)

 Verkleij et al. [8]

Previously reported RCT (hip OA study): LCGA, linear model only

5 trajectories; three stable, two changing: mild (31%) or moderate pain (14%), alwaysin pain (14%) and regular (22%) or rapidly (19%) progressing (VAS)

Univariate MLR, ref. mild: low education, (mod./always +), BMI, morning hip stiffness, hip flexion (all bar mild +), KL, hip pain > 3 years (always/rapid +), generalised OA (always/regular prog. +), hip internal rotation (always +), back pain (all bar mild and reg prog.), trochanteric pain (all bar mild and rapid prog. +)

 Bastick et al. [11]

CHECK cohort (early symptomatic OA, knee only): LCGA up to cubic polynomials

6 trajectories: constant mild (26%) or severe pain (10%), severe (5%) or moderate progression (24%), major (3%) or moderate regression (29%) (NRS)

Uni/multivariate MLR, ref. constant mild: BMI (mod. prog. and severe +), education (all bar ref. and severe –), comorbidity (severe cons. and prog. +), WOMAC physical (all bar sev., prog. and ref. +), knee joint space tenderness (prog. and mod. reg. +), painful knee flexion (maj. reg. –)

 Norton et al. [13]

ERAS cohort (early (< 2 years) RA), baseline, 6 months and annual to 10 years follow-up: GMM, MAR assumed

4 trajectories: low (6%), moderate (28%) and high stable (20%), and moderate increasing (46%) (HAQ)

Univariate analyses: age, female, educational/economic disadvantage, unemployment, DAS/VAS/Larsen, comorbidity, mortality (track severity)

 Wesseling et al. [4]

CHECK cohort (symptomatic knee OA), 5 years follow-up: LCGA, quadratic added

3 trajectories: marginal pain (31%), mild pain (42%) and moderate pain (26%) (with progression) (NRS, last week)

Univariate and multivariate LRs: BMI ≤ 25, subtertiary education, hip pain, comorbidities, PCI worrying and resting (marginal –)

 Norton et al. [12]

ERAS and NOAR cohorts (early RA and early inflammatory polyarthritis): LCGM, polynomials added

Both cohorts showed 4 J-shaped trajectories differing in baseline severity, low (21%), moderate (32%, 33%), high (30%, 26%) and severe (16%, 20%) (HAQ)

Age, % female gender, lower SES, DAS28 increase alongside severity

 Miaskowski et al. [14]

Sampled from breast care centres, post surgery: GMM, quadratic added, 0–6 months follow-up (NRS at shoulder/arm)

3 trajectories: no pain (42%), mild pain (24%), moderate pain (35%). All differ at baseline and remain stable

Age, white ethnicity (no pain +), education (mild pain longer time in educ.), income mild > mod. at >  $100K), BMI, depression, trait anxiety (mod. +), QoL (tracks pain severity)

 Miaskowski et al. [2]

Sampled from breast care centres, post surgery: GMM, quadratic added, 0–6 months follow-up (NRS at breast)

4 trajectories: no pain (32%), mild (43%), moderate (13%) and severe pain (12%). Moderate group shows progression, severe shows slight pain regression

Age (no +), trait anxiety, sleep disturbance, QoL (no –), non-white ethnicity, household income (severe –), BMI, comorbidities, QoL (severe +), CES-D depression (mod./sev +)