Study reference

Study characteristics

Patient characteristics

Candidate predictors

Model development, performance and evaluation

Outcome measures and results

Comments

Interpretation of model

Efe, 2011

Source of data¹ and date: cohort between 1984 and 2001

Setting/ number of centres and country:

Hospital, Germany

Funding and conflicts of interest:
Funding: not reported
Conflicts of interest: None

Recruitment method²:

Consecutive patients

Inclusion criteria:

- Lateral closing-wedge osteotomy performed between 1984 and 2001 within author’s institution

- Indication was symptomatic medial compartment osteoarthritis with varus malalignment

Exclusion criteria:

- not reported

Treatment received?

-All patients were underwent a lateral closing-wedge technique (4). Fibular transection was performed at the

junction of the middle and distal thirds (20) through a separate incision. A transverse incision with the patient

in supine position was performed for the tibial osteotomy. The peroneal nerve was exposed and protected.

The osteotomy was performed below the tibial tuberosity,

leaving the medial cortex intact. Bone wedge size was based on the preoperative calculations from the

long leg standing radiograph. A laterally-based wedge of bone was removed and the osteotomy was fixed with an AO-plate. Postoperative early active movement and physiotherapy

were allowed. No casts were applied. Partial

weight bearing using 2 crutches for 6 weeks was allowed until bony union was reached. Control standing X-rays

were taken 6 weeks, 6 months, and 1 year after surgery as well as at time of latest follow-up.

Participants:

N=199
(N=239 -->
17 (11%): too ill or refused to participate
13 (5.5%): lost to follow-up)

Mean age ± SD:
54 ± 8 years

Sex: % M / % F

55.3%: Male
44.7%: Female

Other important characteristics:

Mean BMI ± SD:
28 ± 48

Mean Varus angle ± SD: 6^◦ ± 3

Describe candidate predictors³ and method and timing of measurement:

Predictor 1: age
At the time of closing wedge osteotomy.

Predictor 2: gender
At the time of closing wedge osteotomy.

Predictor 3: BMI
At the time of closing wedge osteotomy.

Predictor 4: Kellgren-Lawrence grade of osteoarthritis
Preoperative radiographs included the full leg and standard

short anteroposterior views in the standing position with the knee in full extension, as well as lateral

and merchant views. The severity of osteoarthritis was scored according to the Kellgren-Lawrence system (0 normal, 4 severe).

Predictor 5: varus angle

Preoperative radiographs included the full leg and standard

short anteroposterior views in the standing position with the knee in full extension, as well as lateral

and merchant views.

Varus angle was measured

using mechanical axes of the femur and the tibia (hipknee-

ankle angle) obtained on full length, full weightbearing

standing radiographs.

Predictor 5: Partial arthroscopical menisectomy
prior to HTO

Number of participants with any missing value⁴?

N (%): not reported/excluded, see participants

How were missing data handled⁵?
not reported/excluded, see participants

Development

Modelling method⁶:

Multivariate Cox regression models

Performance

Calibration measures⁷ and 95%CI:

NR

Discrimination measures⁸ and 95%CI:

NR

Classification measures⁹:

NR

Evaluation

Method for testing model performance¹⁰: internal/external

NR

Type of outcome: single/combined?

High tibial osteotomy failure

Definition and method for measurement of outcome:

The need for conversion to total knee arthroplasty

Endpoint or duration of follow-up:

9.6 (1-18) years

Number of events/outcomes:

36 (16%)

RESULTS

Multivariable model¹¹:

Alternative presentation of final model¹²:
NR

KL-grade: Patients with a higher grade of osteoarthritis (>2) may

be better treated with arthroplasty

Conclusion: A significant risk factor for HTO failure was preoperative

Kellgren-Lawrence osteoarthritis grade >2.

Figueroa, 2017

Source of data¹ and date: retrospective cohort from December 2002 until November 2013 from clinical research database

Setting/ number of centres and country:

Sydney Orthopaedic Research Institute, Australia

Funding and conflicts of interest:
Funding: not reported
Conflicts of Interests: none

Recruitment method²:

Consecutive patients

Inclusion criteria:

- Patients with symptomatic medial compartment OA on X-ray that underwent medial opening wedge high tibial osteotomy

with internal fixation and a bone graft.

- Operated by one of the three surgeons (DP, MC, BF) at our centre between December 2002 and November 2013.

- Patients who underwent any concomitant procedures such

as ligament reconstruction (anterior cruciate ligament (ACL), posterior cruciate ligament (PCL)), meniscus surgery (menisectomies) or cartilage surgery (microfracture,

chondroplasty, Matrix-Induced Autologous Chondrocyte

Implantatio (MACI)) were included.

Exclusion criteria:

- Osteotomies around the knee other than medial opening

wedge osteotomies, for example, closing wedge osteotomies, distal femoral osteotomies, tibial tubercle osteotomies or

fibular osteotomies.

- Patients operated in the public hospital.
- Patients who underwent multiligament reconstruction along

with a high tibial osteotomy
- Patients who received external fixation of the osteotomy

Treatment received?

Medial opening wedge high tibial osteotomies were performed

under tourniquet control, as previously described at two

private hospitals. A combination of flexible osteotomes and

laminar spreaders were used to enlarge the osteotomy for the

required correction. Femoral head allograft, iliac crest allograft

or bone substitute was used to fill in the gap and the osteotomy was internally fixed with a plate and screws. Prior to 2009, X-rays and alignment rods were used to monitor correction. The line of weight bearing was shifted to the Fujisawa’s point

 in these cases. In the remaining cases, alignment was

verified ntraoperatively using imageless navigation (OrthoMap

Precision Navigation Software (Stryker, Mahwah, New Jersey,

USA) and ORTHOsoft Zimmer Computer Assisted Solutions

(Zimmer Biomet, Warsaw, Indiana, USA)) in 61% of cases. The

osteotomy gap was filled with femoral head allograft alone in

the majority (89.5%) of cases, with the remainder comprising

allograft and bone substitutes (0.95%), autograft (4.8%) or bone substitute alone (2.4%). The median plate tooth size (in plates with tooth) was 11.25mm (95%CI 11 to 12.5mm), with the

majority of osteotomies internally fixed with VERSUS Osteotomy (Zimmer Biomet) (46.2%) or Puddu plates (Arthrex, Naples, Florida, USA) (30%), with the remainder (23.8%) comprising ContourLock HTO Plate (Arthrex) Tomofix (Depuy Synthes,

West Chester, Pennsylvania, USA), ELIX (Biotech, Wiesbaden,

Germany) iBalance (Arthrex) and Targon (Aesculap, Tuttlingen,

Germany). Concurrent surgeries were performed with the HTO to reconstruct the ACL (5.7%), as well as treat cartilage lesions

using chondroplasty (9.5%), microfracture (22.5%) or MACI

(11%) or partial meniscectomy (17.7%). The indication to make

a chondral resurfacing technique was related to the presence

of focal lesions. Lesions smaller than 2cm2

 were treated with

microfracture and larger lesions were treated using MACI

Participants:

N=210

Mean age ± SD:

51 (95%CI: 42-57)

Sex: % M / % F

84% M

Other important characteristics:

BMI
28.9 kg/m² (95%?CI: 23-36)

Describe candidate predictors³ and method and timing of measurement:

Predictor 1: bilateral
unilateral/bilateral surgery
extracted from database

Predictor 2: gender
extracted from database

Predictor 3: age

extracted from database

Predictor 4: ever smoked
extracted from database

Predictor 5: ligament reconstruction

Concomitant surgery
extracted from database

Predictor 6: cartilage treatment
Concomitant surgery
extracted from database

Predictor 7: meniscectomy

Concomitant surgery
extracted from database

Number of participants with any missing value⁴?
not reported

N (%): not reported/excluded, see participants
not reported

How were missing data handled⁵?
not reported/excluded, see participants
not reported

Development

Modelling method⁶:

Multivariable cox proportional hazards model

Performance

Calibration measures⁷ and 95%CI:

NR

Discrimination measures⁸ and 95%CI:
NR

Classification measures⁹:

NR

Evaluation

Method for testing model performance¹⁰: internal/external

NR

Type of outcome: single/combined?

Failure

Definition and method for measurement of outcome:

Failures were identified by detailed review of the clinical database

and clinical notes, with failure defined as a conversion to a TKR. The Australian National Joint Replacement Registry was queried to account for conversions in any cases lost to follow-up.

Endpoint or duration of follow-up:

NR

Number of events/outcomes:

13 (6.3%)

RESULTS

Multivariable model¹¹:

Alternative presentation of final model¹²:
NR

T.a.v. ΔKOOS (pain relief), wordt een partial leas squares regression analysis gedaan:

Partial least squares regression with up to 10 components

was performed to associate patient demographics, arthritis

severity and surgical options with change in KOOS-pain (pain

relief) between preoperative and postoperative measurements.

Due to missing data in some patients, two other models (models 2 and 3) were constructed for this case using a partial

least squares regression model (table 1), with model 3 including

preoperative alignment measures. Previous studies have investigated the relationship between these factors and procedure

survival, but not pain relief. In the case of alignment, the

reason to include only preoperative alignment in the analysis

was mainly due to the fact that postoperative alignment is not

a useful predictive measure, as it is not a fixed value and it

will change during follow-up due to progression of the osteoarthritic disease. The efficacy of the models was assessed by

comparing the predicted values to actual values and examining

the model SE and predicted R2

. Individual predictor variables

were assessed for influence with the dependent variable by the

magnitude of the standardised coefficient

Model 2: (n=91)
bilateral, gender, age at surgery, bmi, KOOS-pre, MFC ICRS Grade, Ever smoked, Ligament reconstruction, Cartilage treatment, Menisectomy, KOOS follow-up, and interactions

pain relief was significant (P<0.01)

with a predicted R2

 (adjusted) of 17% and SE of 15.5 points

Model 2 identified two components (PC1 and PC2)

as predictors of pain relief, with PC1 dominated by cartilage

treatment, and the interaction between preoperative KOOSpain and cartilage treatment. The second component (PC2)

was influenced by ICRS grade of the medial femoral condyle

cartilage and lifetime smoking status.

Model 3: (n=41)
bilateral, gender, age at surgery, bmi, KOOS-pre, MFC ICRS Grade, Ever smoked, Ligament reconstruction, Cartilage treatment, Menisectomy, KOOS follow-up, Preop alignment, and interactions

Generated a significant (P<0.01) fit to pain relief. The model explained 50% of the variance in the response (R2, with an SE 11.8 KOOS points

Model 3 also identified two principal components, with cartilage treatment dominating PC1, together with the interaction

between age at surgery and cartilage treatment. PC2 was influence by preoperative KOOS-pain score, as well as the interaction

between preoperative KOOS-pain and age at surgery.

Keenan, 2019

Source of data¹ and date: retrospective cohort from January 1, 1994 until December 31, 2011 from a prospectively compiled database

Setting/ number of centres and country:

Department of Orthopaedics and Trauma, Royal Infirmary of Edinburgh,

UK

Funding and conflicts of interest:

None

Recruitment method²:

Consecutive patients

Inclusion criteria:

- indication for undertaking the procedure was isolated medial compartment osteoarthritis in a varus knee

Exclusion criteria:

- patients undergoing HTO for other indications, such as cartilage resurfacing, meniscal transplantation and ligament reconstruction

Treatment received?

Under tourniquet, an eight to 10 cm incision was made parallel to the anterior border of the medial collateral ligament, adjacent to the anteromedial aspect of the proximal tibia. The medial

collateral ligament was stripped from the tibia posteriorly to expose the proximal medial surface of the tibia. Under the guidance

of fluoroscopy, a guide wire was advanced medially from one to two centimetres distal to the level of the joint up to the lateral cortex, and an osteotomy apparatus was mounted on the guide wire. A second Kirschner wire was then introduced at an appropriate angle up to the lateral cortex. Medial, anterior and posterior cortices were cut immediately under the guide wire up to one centimetre to the lateral cortex. The osteotomy was opened until the lower limb mechanical axis passed across the Fujisawa point. The osteotomy was fixed with a plate (non-locking and when locking plates became more mainstream

practice was changed) and the osteotomy site was grafted with an allogenic bone graft (femoral head). After the closure of the

layers, the knee was placed into a hinged immobilizer. Prophylactic intravenous antibiotics were administered before application

of the tourniquet. Patients were allowed to walk non-weight-bearing (using two crutches) for six weeks. Full weight-bearing was

permitted six weeks after surgery, when strengthening exercises were initiated.

Participants:

N=111

Mean age ± SD:

45 years (range 18-68)

Sex: % M / % F

83.8% M
16.2% F

Other important characteristics:

NR

Describe candidate predictors³ and method and timing of measurement:

Predictor 1: age
from medical notes

Predictor 2: gender
from medical notes

Predictor 3: economic status

Scottish index of multiple deprivation (SIMD) at the level of patient postcode

Number of participants with any missing value⁴?

N (%): NR

How were missing data handled⁵?
NR

Development

Modelling method⁶:

Cox proportional hazards model

Performance

Calibration measures⁷ and 95%CI:

NR

Discrimination measures⁸ and 95%CI:
NR

Classification measures⁹:

sensitivity and specificity, AUC for cut-off point age

Evaluation

Method for testing model performance¹⁰: internal/external

NR

Type of outcome: single/combined?

Failure

Definition and method for measurement of outcome:

Early (Before 12 years) conversion to total knee arthroplasty

Endpoint or duration of follow-up:

12 years (6-21)

Number of events/outcomes:

40 (36%)

RESULTS

Multivariable model¹¹:

Alternative presentation of final model¹²:
age as binary variable (cut of 47 years old), adjusted for covariates:
HR: 2.49 (95%CI: 1.26-4.91, p=0.01)

Age sensitivity and specificity: 60%
AUC: 0.72 (95%CI 0.62-0.81, p<0.001)

Khosbin, 2017

Source of data¹ and date: retrospective cohort from July 1, 1994 until June, 30, 2010 from administrative healthcare database

Setting/ number of centres and country:

Ontario, Canada

Funding and conflicts of interest:
Funding: the studye was supported by funds from the arthritis research program of the university health network. The study was performed at the institute for clinical evaluative sciences, a non-profit research institute funded by the Ontario ministry of health and long-term care
Conflicts of interest: none

Recruitment method²:

Consecutive patients

Inclusion criteria:

- resident of Ontario
- tibial osteotomy
- prior diagnosis of osteoarthritis

Exclusion criteria:
- age under 18 years
- non-Ontario residents
- previous tibia or femur fracture, physeal surgery
- previous knee arthroplasty

- prior episode of patellar instability

- known inflammatory arhropaty or cancer

Treatment received?
osteotomy, not further described

Participants:

N=2671

Median age (IQR):

46 (39-53)

Sex: % M / % F

62% M

Other important characteristics:

NR

Describe candidate predictors³ and method and timing of measurement:

Predictor 1: age

Predictor 2: gender

Predictor 3: residence

Urban versus rural

Predictor 4: income quintile

Predictor 5: CADG
medical co-morbiditiy scores were estimated for the cohort using the Collapsed Aggregated Diagnosis Groups (CADG)

Predictor 6: Meniscual repair performed within 3 years prior of HTO

Predictor 7: Meniscal debridement performed within 3 years prior of HTO

Predictor 8: Condral surgery performed within 3 years prior of HTO

Predictor 9: Concurrent bone grafting performed with HTO

Predictor 10: Concurrent ligament reconstruction performed with HTO

Predictor 11: HTO performed at a teaching hospital

Predictor 12:Surgeon volume of HTO performed in the previous calendar year

Predictor 13: Surgeon years in practice

Number of participants with any missing value⁴?

N (%):
NR

How were missing data handled⁵?
1) NR

Development

Modelling method⁶:

Cox proportional hazards model

Performance

Calibration measures⁷ and 95%CI:

NR

Discrimination measures⁸ and 95%CI:
NR

Classification measures⁹:

NR

Evaluation

Method for testing model performance¹⁰: internal/external

NR

Type of outcome: single/combined?

Time to knee arthroplasty

Definition and method for measurement of outcome:

Difference between service dates of the index event and main outcome measured in months

Endpoint or duration of follow-up:

30 June 2012, allowing for a minimum 2-year follow-up

Number of events/outcomes:

32.3%

RESULTS

Multivariable model¹¹:

Alternative presentation of final model¹²:
NR

Van Wulfften Palthe, 2018

Source of data¹ and date: retrospective cohort from January 1992 until July 2002 from patient database

Setting/ number of centres and country:

Department of, Noordwest Ziekenhuisgroep, The Netherlands

Funding and conflicts of interest:

None

Recruitment method²:

Consecutive patients

Inclusion criteria:

- undergoing a High tibial osteotomy for the treatment of symptomatic medial compartment osteoarthritis

Exclusion criteria:

- patients with post-traumatic malalignment of the knee
- age under 18 years

Treatment received?
Open wedge osteotomy or closed wedge osteotomy or dome osteotomy or hemicallotasis or chevron transfixation

Participants:

N=223

Mean age ± SD:

54 (range 24-80)

Sex: % M / % F

55.2% M
44.8% F

Other important characteristics:

BMI (SD)
27.2 (3.9)

Describe candidate predictors³ and method and timing of measurement:

Predictor 1: age
<50 years, ≥50 years
from patient notes

Predictor 2: gender
from patient notes

Predictor 3: surgeon

from patient notes

Predictor 4: technique
Chevron, Callus, Dome, Closing, Opening
from patient notes

Number of participants with any missing value⁴?

N (%):
NR

How were missing data handled⁵?
1) NR

Development

Modelling method⁶:

Cox regression analysis was used to identify independent predictors

of survival, entering all variables that were signifcant or

demonstrated a trend towards on univariate analysis.

Performance

Calibration measures⁷ and 95%CI:

NR

Discrimination measures⁸ and 95%CI:
NR

Classification measures⁹:

NR

Evaluation

Method for testing model performance¹⁰: internal/external

NR

Type of outcome: single/combined?

Failure

Defnition and method for measurement of outcome:

conversion to an arthroplasty of the

treated knee, either unicompartmental or total knee arthroplasty

Endpoint or duration of follow-up:

NR

Number of events/outcomes:

98 (44%)

RESULTS

Multivariable model¹¹:

Alternative presentation of final model¹²:
NR

Study reference

(first author, year of publication)

Study participation¹

Study sample represents the population of interest on key characteristics?

(high/moderate/low risk of selection bias)

Study Attrition²

Loss to follow-up not associated with key characteristics (i.e., the study data adequately represent the sample)?

(high/moderate/low risk of attrition bias)

Prognostic factor measurement³

Was the PF of interest defined and adequately measured?

(high/moderate/low risk of measurement bias related to PF)

Outcome measurement³

Was the outcome of interest defined and adequately measured?

(high/moderate/low risk of measurement bias related to outcome)

Study confounding⁴

Important potential confounders are appropriately accounted for?

(high/moderate/low risk of bias due to confounding)

Statistical Analysis and Reporting⁵

Statistical analysis appropriate for the design of the study?

(high/moderate/low risk of bias due to statistical analysis)

Efe, 2011

Low

Unknown (lost to follow-up patients are not described with key characteristics)

Low

Low

Low

Moderate (36 events, 5 factors)

Figueroa, 2017

High (only patients with concomitant surgeries included, but adjustment included in analyses)

Unknown (lost to follow-up not reported)

Low

Low

Low

High (13 events, 7 factors)

Keenan, 2019

Low

Unknown (lost to follow-up not reported)

Low

Low

Low

Moderate (40 events, 6 factors)

Khosbin, 2017

Moderate (patients with concomitant surgeries included, but adjustment included in analyses and unknown whether patients suffered from unicompartmental osteoarthritis))

Unknown (lost to follow-up not reported)

Low

Low

Low

Low (32.3% events in 2671 patients, 17 factors)

Van Wulfften Palthe, 2018

Low

Unknown (lost to follow-up not reported)

Low

Low

Low

Low (98 events, 7 factors)