A short-term radiological and clinical comparison between the bi-cruciate and cruciate retaining total knee arthroplasty A retrospective case controlled study

Abstract

Purpose

The aim of this study was to provide a short term comparison in radiological and clinical outcome between Bi-Cruciate Retaining (BCR)- and Cruciate Retaining (CR) Total Knee Arthroplasty (TKA).

Methods

The cohort consists of 122 patients undergoing a TKA with PSI, equally distributed over the BCR- and CR-TKA group. Perioperative conditions were observed and radiological images were analysed pre-, 6-weeks, and 1-year postoperative to quantify alignment differences between BCR- and CR-TKA. Preoperatively predicted templates were compared with the implanted size to determine predictive value. In addition mean range of motion and revision rates were determined in both groups.

Results

No significant difference was observed in amount of outliers in component alignment between BCR- and CR-TKA. Outliers of the Hip-Knee-Ankle-Axis (HKA-axis) occurred significantly more frequent (P = 0.009) in the BCR-group (37.7%) compared to CR-TKA (18.0%). No clinically relevant differences regarding the predictive sizing of implant components was obtained. No significant differences were observed in revision rates (P = 1.000) and ROM (p = 0.425) between the BCR-groep and CR-group at 2-years FU.

Conclusion

This study illustrates that although the HKA-axis was not fully restored, bi-cruciate retaining surgical technique for BCR-TKA is safe and effective with comparable radiological and clinical outcome as CR TKA. Randomized controlled trials with longer follow up on the HKA-axis alignment and clinical parameters are needed to confirm the presented results and should focus on possible cut off values concerning leg axis in order to define in what patients a BCR-TKA can safely be used.

Level of evidence IV

Retrospective Case Controlled Study.

List of abbrevations

HKA

Hip-knee-ankle angle

FFC

Frontal femoral component

FTC

Frontal tibial component

LFC

lateral femur component

LTC

Lateral tibial component

BCR-TKA

Bi-Cruciate Retaining Total Knee Arthroplasty

CR-TKA

Cruciate Retaining Total Knee Arthroplasty

ACL

Anterior Cruciate Ligament

PCL

Posterior Cruciate Ligament

BMI

Body Mass Index

KL

score Kellgren-Lawrence score

OA

Osteoarthritis

ADL

Activities of Daily Living

1. Introduction

Symptomatic knee osteoarthritis (OA) can be very incapacitating, especially in the younger and physically more active population.^1,3 For the treatment of more severe knee OA, the total knee arthroplasty (TKA) is one of the most effective orthopaedic procedures considering functional restoration and pain relief of the knee joint.³ In order to thrive for the most optimal result, changes in design and implementation of the knee prosthesis have taken place over the last decades.^4,5 Patients with severe OA who are not suitable for the unicompartmental knee arthroplasty tend to be operated with the posterior cruciate retaining total knee arthroplasty (CR-TKA) in order to reduce pain and increase range of motion. However, this technique suffers from reduced proprioception and loss of normal kinematics of the knee contributing to dissatisfaction rates up to 20%.1, 2, 3^,5,6 Trying to resolve the shortcomings of existing techniques, bi-cruciate retaining TKA (BCR-TKA) seems to offer a solution for active patients who are not eligible for a UKA due to bicompartimental degenerative cartilage but with preservation of both ACL and PCL.^7,8 In BCR-TKA increased stability and proprioception, might lead to a closer to anatomic movement pattern compared to CR-TKA1, 2, 3^,7 and thereby possibly leading to improvement of functional outcome and patient satisfaction.9, 10, 11^,26 On the other hand, previous literature pointed out that BCR-TKA suffers from higher revision rates which could be the result of malalignment of the TKA due to the more difficult surgical technique compared to CR-TKA.^10,12 Implementation of the BCR-TKA is tought to be challenging due to a different design of the tibial component that provides room for the anterior cruciate ligament (ACL).²³ The key elements in obtaining a better clinical outcome lays within accurate alignment of the prosthesis and restoration of a neutral HKA-axis). Furthermore, a mechanical axis within the range of ±3° varus/valgus is found to improve the longevity of TKA.¹³ Another possible downside of this technique and in contrast to what would be expected, is that patients operated with a BCR-TKA seem to suffer more frequent from postoperative stiffening leading up to an impaired range of motion.²⁵ Postoperative stiffening with limited motion can be induced by scar forming around the ACL during implantation.²⁹

The primary objective of this study was to assess radiological outcome (e.g. amount of outliers, preoperatively planned size prediction) between patients operated with the BCR-TKA and CR-TKA. It was hypothesized that there would be no difference in the number of outliers and accuracy of preoperatively predicting the implant size between the BCR-TKA and CR-TKA. Secondary objective of this study was to objectify differences in clinical outcomes (e.g. improvement of ROM and revision rates) between BCR TKA and CR TKA. It was hypothesized that there would be no difference in ROM and survival rate between BCR TKA and CR TKA.

2. Materials and methods

2.1. Participants

This case controlled study includes 61 consecutive patients who underwent a bi-cruciate-retaining total knee arthroplasty (BCR-TKA) for end stage knee OA between 2014 and 2016 in the Zuyderland Medical Centre (Sittard-Geleen, the Netherland). Generally accepted contraindications for cruciate retaining TKA were also applicable in this patient cohort. In addition, to be eligible for BCR-TKA, a fully functional ACL had to be present (negative Lachmann and anterior drawer test); patients with an extension deficit were considered inappropriate candidates for BCR-TKA. To obtain informed conscent, patients were informed preoperatively that no longterm data on the BCR-TKA was available. Peroperative conversion to CR-TKA was made in case of a degenerative ACL and these patients were excluded from this analysis. Assesment of the state of the ACL was at the discretion of the operating surgeon. In order to reduce potential confounding, the patients were matched on baseline characteristics (age, gender, BMI and follow up date) with 61 patients who underwent cruciate retaining TKA (CR-TKA), contributing to a total cohort of 122 patients. Baseline demographics are shown in Table 1.

Table 1.

Lachman test graded by amount of anterior displcament of tibio-femoral joint.

Grade	Amount anterior displacement of tibio-femoral jointa
Grade 0	≤ 2 mm
Grade I	3–5 mm
Grade II	6–9 mm
Grade III	≥ 10 mm

^aSide-to-side difference in anterior displacement of tibio-femoral joint (eg. Anterior translation of tibia relative to femur).

2.2. Procedures

All the patients were operated by one experienced surgeon. Both groups were operated by similar surgical technique using the para-patellar approach and instrumentation sets.^9,18 In CR-TKA the ACL was routinely sacrificied for optimal exposure of cutting planes.^19,20 All the patients were operated using patient-specific instruments (Signature, Zimmer-Biomet, Warsaw, IND) for TKA alignment in both groups.^5,19,20 This PSI technique uses the principle of bony referenced alignment. Prior to surgery a MRI scanning of the knee was performed according to the standard Signature scanning protocol. Software (Mimics, Materialise NV, Leuven, Belgium) was used to create virtual three-dimensional models of femur and tibia. Position of the prosthesis was calculated to retrive a neutral position of tibial and femoral components relative to the mechanical axes of femur and tibia in the frontal plane and to achieve a neutral mechanical axis. In the sagittal plane, flexion of the femoral component and posterior slope of the tibial component were calculated at 3°. Femoral rotation was set parallel to the transepicondylar axis in the coronal plane. Rotation of the tibial component could not be calculated preoperatively using software. Based on these models appropriate guides for implant positioning were determined for each patient individually.¹⁹ Pre-operative digital plans were reviewed by the operating surgeon and adjustments in sizing or implant position were made when deemed necessary. Operative procedure using these guides was described in detail by Boonen et al.²⁰ In respect of the mechanical axis of femur and tibia, a 90° angle was handled to perform the distal femur cut and horizontal tibia cut. Significant releasing of major soft tissue (medial/lateral collateral ligaments) was avoided. Iliotibial band release and/or popliteus release occurred as necessary.9, 23 Ligamentous balancing was performed in the case of significant fixed flexion deformity, limited flexion or significant ligament contracture. The flexion and extension gaps were measured manually using a modified dogbone with a notch confirming the area of the saved eminentia.¹⁷ Minimal releasing to the medial or lateral soft tissues in rare instances were performed if needed to achieve symmetric medial and lateral flexion and extension gaps. The BCR technique strives towards a 2 mm joint space opening. Differential bearing thicknesses could be deployed based on surgeon discretion to achieve optimum soft tissue balance.

2.3. Instruments

During their preoperative, 6-weeks and 12-months appointment all patients received standard radiograph of the knee consisting of AP and lateral views and a standing long-leg axis view. These radiographs were scored by two observers. Frontal view radiograph was used to evaluate varus/valgus positioning of each component (femur: frontal femoral component (FFC) and tibia: frontal tibial component (FTC)) whilst flexion/extension of the femur (lateral femoral component (LFC)) and degree of slope of the tibia (lateral tibial component (LTC)) were obtained on lateral view radiographs.^19,20 Preoperatively, the deviation on the leg-axis and the degree of OA using the Kellgren-Lawrence scale was measured. Three-months postoperatively fit and alignment of the signature implant is observed in the frontal and sagittal plane.^19,20 A difference in alignment of > 3-degrees compared to the preoperative calculated position of each component (FFC, FTC, LFC and LTC), was considered to be an outliers.^16,19 Additionally, radiographs were checked for possible early loosening at 2-years FU. Radiographic measurements were made using a calibrated protocol on digital images using SECTRA PACS medical viewer described by Schotanus et al.¹⁶ and Boonen et al.¹⁹ Differences in approved size by the surgeon of both components compared to the actual implanted sizes were noted. Postoperative AP-laxity and ROM by clinical assessment with a goniometer were obtained on their 3-moths, 12-months and 24-months follow up. Postoperative AP-laxity was measured using the Lachman test and graded by amount of anterior displcament of tibio-femoral joint (Table 1). The study was approved by the local hospital and the METC (METC number: Z2019005).

2.4. Statistical analysis

The analyses were performed with the use of SPSS version 17.0 software (SPSS Inc., Chicago, USA). Descriptive statistics were used to test differences of proportions. For all analyses, a p value was considered to be statistically significant at p < 0.05. In the case of a significant difference in one of the primary outcome measures, the odds ratio (OR) with 95% confidence interval (CI) was calculated to determine possible risk factors for malalignment of the TKA. Median values were used in order to create cut off points. Intra- and interobserver agreement was analysed using intraclass correlation coefficients. ICC’s were expressed as very good, good, moderate, and poor with a corresponding value of 0.81–1, 0.61 to 0.80, 0.41 to 0.60, and <0.40 respectively.²¹ Continuous variables were described using means (±SD) and categorical variables were tabulated with absolute frequencies (%).

3. Results

3.1. Clinical findings

A total cohort of 122 patients consisting of 76 females and 46 males were equally distributed in the BCR- and CR group. No significant differences in baseline characteristics were observed between the BCR- and CR group (Table 2). Mean age at surgery was 65.4- and 65.2-years in the CR- and BCR-group respectively. No conversions to conventional instrumentation were observed.

Table 2.

Baseline demographics, Mean (SD) and Proportion (%), compared between the cruciate retaining and the bi-cruciate retaining group.

	CR-TKA (n = 61)	Icc	BCR-TKA (n = 61)	95% CI	P-value
		Intra/inter
Age at surgery, year	65.4 (7.2)		65.2 (7.1)	[-2.7 – 2.4]	0.903
BMI, Kg/m³	28.0 (3.8)		27.5 (7.1)	[-2.6 – 1.5]	0.591
Gender [Female]	38 (62.3)		38 (62.3)		1.000
KL-score	2.8 (0.7)		2.7 (0.7)	[-3.7 – 0.1]	0.371
KL-score, n = 1/2/3/4	2/15/38/6		6/12/39/4		0.432
HKA-axis preoperative					0.346
Varus	41 (67.2)		48 (79.7)
Valgus	17 (27.9)		12 (19.7)
Neutral	3 (4.9)		1 (1.6)
Anglea	5.6 (3.7)	0.981/0.981	6.2 (3.2)	[-0.8 – 1.5]	0.535
Preoperative HKA outliers, > 3°	45 (73.8)		51 (83.6)		0.185
Preoperative HKA outliers, > 6°	26 (42.6)		25 (41.0)		0.854

BMI: Body Mass Index; KL-score: Kellgren-Lawrence score; HKA-axis: Hip-Knee-Ankle axis.

^aDeviation from neutral HKA-axis of 180°.

3.2. Radiological findings

A significant difference is reported in percentage of HKA-axis outliers (P < 0.009) with an accompanying significant difference in mean HKA deviation angle with respect to a neutral HKA-axis (p = 0.003) in favour of the CR-TKA group (11 outliers and mean deviation of 2.1°) compared to the BCR-TKA group (24 outliers with mean deviation of 3.1°). In regard to the individual components, there was no significant difference in amount of outliers and mean deviation angle between both groups for the femoral and tibial component (Table 3, Table 4). Median preoperative HKA-axis deviation was calculated at 6°. When handling this as a cut off value, preoperative HKA-axis of >6° with BCR-TKA is a risk factor for a deviant HKA-axis postoperative of more than 3° (OR 4.030, 95% CI 1.201–13.526). A different insert was used lateral versus medial to prevent laxity of the knee joint in 7 patients. Intra- and inter-observer agreement varied from good to very good (Table 4).

Table 3.

Number of patients (%) with outliers of >3° deviation of the planned and postoperative alignment compared between the CR- and BCR-TKA.

	CR-TKA (n = 61)	BCR-TKA (n = 61)	P-value
Postoperative HKA outliers	11 (18.0)	24 (39.3)	0.009
FFC outliers	7 (11.5)	9 (14.8)	0.592
FTC outliers	3 (11.5)	4 (6.6)	0.697
LFC outliers	14 (23.0)	18 (29.5)	0.410
LTC outliers	10 (16.4)	8 (13.1)	0.610
Different inserts medial vs. lateral	Not applicable	7 (11.5)	Not applicable

HKA: Hip-knee-ankle angle; FFC: Frontal femoral component; FTC: Frontal tibial component; LFC: Flexion/extension of the femur component; LTC: Lateral tibial component.

Table 4.

Mean (SD) deviation values of the planned alignment and postoperative alignment compared between the CR- and BCR-TKA.

	CR-TKA (n = 61)	icc	BCR-TKA (n = 61)	95% CI	P-value
		Inter/intra
Postoperative HKA deviationa	2.1 (1.6)	0.970/0.969	3.1 (2.1)	[0.4–1.8]	0.003
FFC deviationb	1.8 (1.5)	0.719/0.712	2.1 (1.6)	[-0.3 – 0.8]	0.380
FTC deviationb	1.3 (1.3)	0.702/0.985	1.5 (1.2)	[-0.3 – 0.6]	0.513
LFC deviationb	2.6 (2.1)	0.906/0.970	2.7 (2.1)	[-0.6 – 0.9]	0.761
LTC deviationb	2.4 (1.5)	0.784/0.961	2.1 (1.5)	[-0.8 – 0.3]	0.302

HKA: Hip-knee-ankle angle; FFC: Frontal femoral component; FTC: Frontal tibial component; LFC: lateral femur component; LTC: Lateral tibial component.

^aDeviation from neutral HKA-axis of 180°.

^bDeviation from neutral axis of 90°.

3.3. Preoperative sizing

The results and the proportions of oversized, undersized and correctly predicted templates planned by the technician and surgeon are summarized in Table 5. Neither the technician or the surgeon showed a significant difference in proportions of correctly predicted templates between BCR-TKA and CR-TKA for both components. On the other hand, significantly more adjustments in the technicians’ predicted templates of the tibial size occurred in the BCR-TKA (P = 0.000).

Table 5.

Amount (%) of identical planning by the technician (default) and surgeon (approved) in relation with the implanted size of tibial and femoral components compared between BCR-TKA and CR-TKA groups.

	CR-TKA (n = 61)	BCR-TKA (n = 61)	P-value
Femur size Default vs. OR	54 (89)	48 (79)	0.263
Femur size Approved vs. OR	60 (98)	61 (100)	0.315
Tibia size Default vs. OR	40 (66)	32 (52)	0.000
Tibia size Approved vs OR	51 (84)	54 (89)	0.710

3.4. Clinical outcome

One revision occurred in the CR-TKA group (2-months FU) and one in the BCR-TKA group (at 4-months FU), resulting in a survival rate of 98.4% for both groups. Both Revisions were insert changes due to persistent pain in the CR-TKA group and valgus thrust in the BCR-TKA group. There were no noticeable radiolucent lines on the 2-year FU radiographs in neither of both groups. The ROM was comparable in both groups. At 2-years FU, 3-patients in the CR-TKA group were found to have mild hyperextension (2 patients with 5° and 1 patient with 10°) compared to none of the patients in the BCR-TKA group. No significant differences in cases of increased AP laxity were found. Detailed results of the ROM and postoperative laxity at 3-months, 1- and 2-years, as well as early revision rates and presence of exceeded AP laxity are presented in Table 6.

Table 6.

Clinical outcomes: mean values of postoperative total ROM, number of patients (%) with revisions and radiolucent lines on radiograph and presence of exceeded AP-laxity between the cruciate retaining and the bi-cruciate retaining group.

	CR-TKA (n = 61)	BCR-TKA (n = 61)	95% CI	P-value
ROM, degrees
3-months postoperative	108.4	109.8	[-5.50 – 2.68]	0.497
1-year postoperative	112.5	115.1	[-6.24 – 1.22]	0.185
2-years postoperative	115.5	117.6	[-7.29 – 3.10]	0.425
Revisions	1	1		1.000
Follow Up	2.8 (0.6)	2.8 (0.7)	[-0.3 – 0.2]	0.938
Radiolucent lines	0	0		1.000
AP-laxitya
3-months postoperative	0	1 (degree 1)		0.294
1-year postoperative	0	1 (degree 1)		0.294
2-years postoperative	1 (degree 1)	1 (degree 2)		0.367

^aMeasured with Lachman test.

4. Discussion

In this study we investigated the radiological alignment of the bi-cruciate retaining TKA compared to the cruciate-retaining TKA using standard radiographs and long leg radiographs. As hypothesized, there was no significant difference in the number of outliers for the individual components in BCR-TKA compared to CR-TKA. However, a significant difference in the number of outliers in HKA-axis is observed in favour of the CR-group. As described earlier, a difference in alignment of > 3-degrees compared to the preoperative calculated position of each component (FFC, FTC, LFC and LTC), was considered to be an outliers.^16,19 This measure of misalignment correlates for accelerated wear and impaired general performance of a TKA.^16,19 In accordance to this, Bäthis et al. described improved long term outcome after TKA a HKA-axis with <3° varus or valgus is associated with improved long term outcome.¹³ Proper component alignment and restauration of the HKA-axis is considered one of the key elements in determining the long-term outcomes after TKA, and is believed to decrease both the mechanical and shear stresses placed on the bearing surfaces, as well as the bone/prosthesis interfaces.²² Using the median preoperative HKA-axis deviation angle of 6° as a cut off value, reveals a significant risk (OR 4.030) of malalignment after BCR-TKA compared to CR-TKA when HKA-axis exceeds these 6°. Based on these results, we advise to approach the indication for BCR-TKA in cases with preoperative HKA-axis deviation >6° with caution. Future studies and accurate follow-up of BCR-TKA patients is mandatory in order to determine whether the minimal statistical difference of 1° in postoperative HKA-axis leads up to clinical significant and relevant differences of increased wear or possible (early) failure mechanisms.

The authors of this study found no reason for the increased amount of outliers of the HKA-axis in BCR-TKA as outliers in femur and tibia component occurred in the same extent. Possibly, the observed differences are the result of measurement errors or a difference in alignment guide between BCR- and CR-TKA. Despite the fact that the PSI method is bone referenced to determine the size and alignment, a different insert thickness medially and laterally can be opted for during BCR-TKA surgery in order to optimally balance the knee joint. Furthermore, the BCR-TKA provides the choice of a bilaterally different insert, what could possibly have an additional influence on the postoperative HKA axis. Based on our data, patients implemented with bilaterally different inserts (n = 7) showed no additional risk in obtaining a deviant postoperative HKA-axis due to a low odds ratio (OR: 1.179 [0.239–5.801]) and is therefore in itself not a sufficient explanation for the higher incidence of outliers in HKA axis in the BCR TKA group.

No significant difference of predicted component size by the surgeon and actual implanted size is found between BCR-TKA and CR-TKA. However, a significant difference between technician and implanted size of the tibia component is shown in favour of the CR-TKA. Therefore, as described in previous literature, the preoperative planning should always be approved by the operating surgeon.¹⁵ Since these results support the ability to preoperatively predict the perioperative size¹⁵ and alignment^16,19 without loss of accuracy.

As secondary outcome of this study, clinical parameters such as ROM, AP-laxity and survival rate were compared between the CR-TKA and the BCR-TKA. No significant differences in ROM in the expense of the BCR-TKA compared to the CR-TKA was found. This contradicts results of prior research on this topic, in which increased stiffness in BCR-TKA was found, hypothetically due to the preservation of the ACL.²⁵ In our experience, a BCR-TKA is more sensitive to be implanted ‘too tight’ than a CR-TKA, resulting in loss of ROM. This highlights the importance of adequate surgical training addressing this issue prior to performing BCR-TKA’s. As the impairment in ROM was also seen in CR-TKA compared to PS-TKA,^27,28 the difference in postoperative ROM found between the BCR-TKA and PS-TKA²⁵ is there more likely a result of sparing of the PCL. Therefore, it can be assumed that sparing additional anatomical structures such as the ACL does not lead to further impaired of postoperative ROM when compared to CR-TKA. No significant difference of the presence of exceeded AP-laxity were shown as measured with the Lachman test. Mild hyperextension after 2-years FU was objectified in 3 cases in the CR-TKA group what could be caused by a degenerative or insufficient PCL. There were no cases of PCL insufficiency in BCR-TKA. Most likely, the PCL attachment is better protected when performing a BCR-TKA as a bony island is to be created, explaining this finding. This is in accordance with cadaveric studies implicating that retention of the ACL could potentially provide the ideal ligament tension and laxity leading up to optimal ROM and joint stability.^29,30 However, longer FU and validated clinical tests are needed to confirm this hypothesis. Lastly, this study obtained no differences in early revision rates between the BCR- and CR-TKA with an overall survival of 89.45 at 2.7 years FU. Since early term revision rates were excellent in both groups, survival of the BCR-TKA seems not to be inferior to that of the CR-TKA. This is in contrast with other short term literature of the BCR-TKA were a survival rate of 88% was found at 3-years FU.¹⁸ Still, long term FU in a prospective setting is needed to confirm these results.

The extensive amount of information provided in patients records is a particular strength of this study. This allowed the authors to match BCR patients with suitable patients operated with CR-TKA. Due to single surgeon data in this case controlled study, a better indication of actual differences between BCR-TKA and CR-TKA can be made. Nevertheless, general applicability could be doubted. This data is obtained from a high-volume surgeon in whom an easier adaptation to a new surgical technique can be assumed compared to low-volume surgeons or residents.¹⁶ However, the BCR-TKA technique could be a barrier for low-volume surgeons as it is known by its technical difficulty of implementation. A weakness of this study is that long leg radiographs and standard radiographs were used to perform the measurements. In general, to objectify the postoperative position of both the femoral and tibial implants, a wide variety of different analyses is used in the literature.¹⁴ Although measurements on standing long-leg radiographs are valid and reliable¹⁴ for routine clinical practise, 3D-CT has shown to be the gold standard for measuring orientation and position of both components.²⁴ In our hospital however, a postoperative 3D-CT is not routinely performed due to costs and radiation exposure. It cannot be excluded that part of the observed differences in alignment between BCR-TKA en CR-TKA are the result of measuring on conventional radiographs. Another major downside of this study is that no preoperative data on ROM and knee laxity was provided what could potentially lead up to selection bias. In addition, as AP laxity is measured by manually assessing the Lachman test, inter- and intra obeserver reliability might be impaired. This subjective measurement could lead up to measurement bias due to the retrospective character of this study. Therefore, a validated measurement of AP-laxity is highly needed.

5. Conclusion

This study illustrates that bi-cruciate retaining TKA resulted in a slightly higher number of outliers in HKA-axis compared to cruciate retaining TKA. Based on the clinical results it can be assumed that preserving the ACL provides necessary knee stability without additionally impairment of ROM compared to the CR-TKA. Randomized controlled trials comparing CR TKA and BCR TKA on the HKA-axis alignment are needed to confirm the presented results and should focus on possible cut off values concerning leg axis in order to define in what patients a BCR-TKA can safely be used.

Funding

There is no funding source.

Ethical approval

Ethical approval by the local ethical committee (email: METC@zuyderland.nl) was obtained for this study.

Informed consent

Informed consent was not necessary since it was a retrospective.

Authors contributions

Soufyan Kalaai conceived, designed, co-ordinated the study, collected and analysed the data and drafted the manuscript. Marlon Scholtes collected the data. Bert Boonen and Emil van Haaren critically reviewed the manuscript. Martijn Schotanus participated in the design of the study, co-ordinated the study and critically reviewed the manuscript. All authors have read and approved the final manuscript.

Declaration of competing interest

The authors have no conflicts of interest to declare.