SEARCH

SEARCH BY CITATION

Abstract

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. SUBJECTS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. AUTHOR CONTRIBUTIONS
  8. Acknowledgements
  9. REFERENCES

Objective

To compare differences in knee varus and valgus angular laxity and passive mechanical stiffness between asymptomatic knees and those with mild, moderate, and severe knee osteoarthritis (OA).

Methods

A total of 127 participants with symptomatic medial tibiofemoral OA and 32 asymptomatic controls participated. OA knees were stratified according to radiographic severity. Varus–valgus laxity was evaluated using a customized dynamometer, providing continuous measurement of varus–valgus joint angle and torque. The following indices were calculated: 1) varus, valgus, and total angular laxity, 2) end-range varus and valgus stiffness, and 3) midrange stiffness.

Results

There was no difference in varus, valgus, and total angular laxity, or varus and valgus end-range stiffness between the groups (P > 0.05 for all). The OA groups were less stiff in the midrange compared with the controls (P = 0.004–0.043).

Conclusion

The absence of differences in total angular laxity is contrary to previous findings, and may be associated with the failure of previous work to account for body size effects. Less midrange stiffness in OA participants compared with controls may indicate less rotational support provided by passive joint structures in knee OA within the functionally important range. The role of passive varus–valgus stiffness in disease onset and progression is worthy of further investigation.


INTRODUCTION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. SUBJECTS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. AUTHOR CONTRIBUTIONS
  8. Acknowledgements
  9. REFERENCES

Knee osteoarthritis (OA) commonly affects the medial tibiofemoral compartment and mechanical factors play a role in disease pathogenesis (1–3). One mechanical factor that has received attention is frontal plane knee laxity, which refers to the behavior of the joint during passive varus–valgus rotation. The fact that frontal plane laxity may be affected in knee OA is not surprising, given pathologic and biomechanical changes associated with the disease such as varus malalignment (4), high knee adduction moments (5), joint space narrowing, meniscal damage, osteophytosis, and bony attrition (6). Greater laxity may result from varus malalignment combined with a high adduction moment during gait that induces chronic stretching of the lateral soft tissue around the knee (7, 8). Increases in laxity may also occur as a result of cartilage loss or bony erosion (often termed “pseudolaxity”). Conversely, laxity may also be reduced in knee OA because the presence of osteophytes may provide a mechanical barrier to movement (9). Importantly, with increasing severity of medial knee OA, increases in varus malalignment (10, 11), knee adduction moments (5), and bony attrition (12) are observed. Therefore, it may be expected that the passive laxity of the knee is affected not only by the presence of OA, but also by its severity.

The laxity measure used in most knee OA studies is an adaptation of clinical laxity tests and quantifies total varus–valgus angular motion when a specific torque is applied. These studies tend to indicate that laxity is increased in the presence of knee OA (13, 14); however, findings are not consistent across the literature (15, 16). A weakness of using total angular motion to quantify laxity is its failure to separate motion into varus and valgus components. Because knee OA predominantly affects the medial compartment, it is likely that laxity will differ under varus and valgus loading conditions. Therefore, greater understanding of the changes in the mechanical properties of the knee with OA may be gained by separating measures of total laxity into its varus and valgus components. Although some recent studies have separated these measurements (17, 18), they have not investigated differences in varus and valgus laxity with the presence and increasing severity of knee OA.

Despite laxity potentially influencing the risk of OA disease onset and progression (3, 13), the functional significance of current laxity indices remains unclear. Reports suggest that approximately 3° to 6° of total varus–valgus motion occurs at the knee during gait (19–23), but reported passive laxity values typically exceed this, being in the order of 5° to 15° (9, 13–16, 21, 22). Therefore, measurement of the mechanical behavior of the joint within its functional range may provide additional important information about changes in laxity with knee OA. The resistance to varus–valgus angular motion of the knee within its functional range is represented by its passive mechanical rotational stiffness (torque per unit of angular movement [Nm/°]). The varus–valgus stiffness of the knee has been described in studies of cadavers and young asymptomatic participants (24, 25); stiffness varies through the range of motion and is nonlinear. In the midrange, the joint provides relatively little resistance to varus–valgus motion (25), i.e., the knee is less stiff and more compliant than when it is approaching its range of motion limits. In contrast, toward its end range in varus and valgus, the knee becomes stiffer and less compliant (25). Given the nonlinear stiffness response of the knee, measures of varus–valgus stiffness in the end range and midrange may reveal important information about joint changes with OA that have not previously been reported.

The aim of this study was to investigate differences in passive frontal plane laxity of asymptomatic control knees and those with mild, moderate, and severe medial tibiofemoral OA. Using a novel methodology (26) to separate the measures into varus and valgus and to assess passive stiffness, the specific laxity indices considered were: 1) varus, valgus, and total angular laxity, 2) end-range varus and valgus stiffness, and 3) midrange stiffness.

SUBJECTS AND METHODS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. SUBJECTS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. AUTHOR CONTRIBUTIONS
  8. Acknowledgements
  9. REFERENCES

Participants.

One hundred fifty-nine participants, including 127 with medial compartment knee OA and 32 asymptomatic, healthy controls, were recruited from the community in Melbourne, Australia, via advertisements in newspapers and local clubs. Participants with OA were combined from 2 separate but similar cohorts. The first cohort consisted of a subset of 54 participants from a randomized controlled trial examining the efficacy of laterally wedged insoles (27), and the second cohort consisted of 73 participants involved in another randomized controlled trial investigating the effects of quadriceps strengthening on the external knee adduction moment (28). All of the data reported in this study were collected at baseline prior to any intervention. OA participants were diagnosed based on the American College of Rheumatology classification criteria (29): age >50 years, had knee pain on most days of the previous month (average level >3 cm on a 10-cm visual analog scale), and osteophytes apparent on radiographs. In addition, to ensure that participants had medial tibiofemoral joint OA, other inclusion criteria consisted of predominance of pain/tenderness over the medial region of the knee and medial radiographic OA defined as at least grade 1 medial joint space narrowing (medial greater than lateral) and grade 1 medial tibial or femoral osteophytes (6). Participants were excluded if the presence of knee OA was questionable, i.e., a Kellgren/Lawrence (K/L) scale grade 1.

Control participants were age >50 years. They were excluded if they 1) reported a history of knee pain, injury, or pathology that interfered with function or caused them to seek treatment within the past year, 2) reported past lower extremity surgery, or 3) reported any condition that affected mobility for more than 1 week in the past year.

Exclusion criteria for both of the participant groups consisted of a history of lower extremity joint replacement, knee surgery, or intraarticular steroid or hylan G-F 20 injection (in the previous 6 months); systemic arthritic condition; were seeking or currently receiving physiotherapy for knee OA; or had a severe medical condition. The participants were screened over the telephone, and if deemed eligible for the symptomatic group, they underwent weight-bearing radiographic analysis. Participants fulfilling the radiographic eligibility criteria were enrolled in the study. The control participants did not undergo radiographic evaluation to exclude radiographic signs of OA due to ethical constraints.

The most painful eligible knee was deemed the study knee for OA participants. Where both knees were equally painful and eligible for inclusion, the dominant knee was evaluated. A randomly selected knee was tested in asymptomatic controls. The research was approved by the University of Melbourne Human Research Ethics Committee. All of the participants provided written informed consent.

Disease severity and knee alignment.

In participants with knee OA, standardized short-extremity weight-bearing posteroanterior knee radiographs were taken with the participant standing barefoot. Radiographs were taken in a semiflexed position for participants in the first cohort, and extended in the second cohort. Radiographic severity of tibiofemoral OA was assessed with the K/L system, whereby 0 = normal; 1 = possible osteophytes; 2 = minimal osteophytes and possible joint space narrowing; 3 = moderate osteophytes, some narrowing, and possible sclerosis; and 4 = large osteophytes, definite narrowing, and severe sclerosis (30). Frontal plane mechanical alignment was derived from anatomic alignment measured from the radiographs using published regression equations (31, 32).

Varus–valgus knee joint laxity.

Laxity was assessed using the Kin-Com 125-AP dynamometer (Chattecx) with customized modifications, as described previously (17, 26). The participants were seated with the knee relaxed and flexed at 20° (13, 33), the ankle secured in a 90° fixed flexion ankle–foot orthosis to a load cell on the horizontal lever arm of the dynamometer, and the tibiofemoral joint directly above, and intersected by, the lever arm axis of rotation. In this gravity-neutral position, the leg was moved passively by the dynamometer 10 times from varus to valgus at 5° per second. Varus and valgus angles were determined at the points where 12 Nm of passive resistance was reached (13).

The analog force and lever arm angle were sampled directly from the Kin-Com at 100 Hz by 16-bit analog to digital conversion (Micro 1401; Cambridge Electronic Design) to a computer using Spike2 software (Cambridge Electronic Design). Joint torque (Nm) was computed as the product of the force (Newtons) recorded at the ankle and the lever arm (meters; measured from the axis of rotation at the knee to the force transducer at the ankle). In contrast to other methods that do not record force continuously (13, 16, 33), a neutral lever arm angle could be identified at zero force, and on this basis, varus and valgus ranges were separated from the total lever arm angle data. Stiffness was defined as the change in joint torque divided by the change in joint angle (Nm/°). End-range varus and valgus stiffness was calculated over the last 25% of the range moving in a varus and valgus direction, respectively (Figure 1). Midrange stiffness was calculated from the averaged varus and valgus movement over a 2° window, 1° on either side of mechanical neutral. Intrarater reliability of the angular laxity and stiffness measures was excellent when measured a week apart in 10 people with medial tibiofemoral OA (intraclass correlation coefficient2,1 0.87–0.97).

thumbnail image

Figure 1. Typical varus–valgus torque-angle data. Movement from valgus (positive) to varus (negative) is shown by the broken line, and varus to valgus movement is shown by the solid line. Varus (a), midrange (b; averaged from varus and valgus curves), and valgus stiffness (c) were computed across the intervals, as shown by the red lines.

Download figure to PowerPoint

Statistical analysis.

Data were analyzed using SPSS, version 15.0 (Norusis/SPSS). Data that were not normally distributed were logarithmically transformed and rechecked for normality prior to analyses (end-range varus stiffness and midrange stiffness). Participants with OA were categorized according to K/L scale grading; grade 2 was categorized as having mild disease, grade 3 as moderate disease, and grade 4 as severe disease.

Differences in demographics and sex distribution between groups were evaluated with an analysis of variance and a chi-square test, respectively. Differences in angular laxity and stiffness between groups were determined using general linear models. Group was included as a fixed factor, whereas age and sex were included as covariates, given their reported effect on the mechanical properties of tendons and ligaments (34–38). Post hoc analysis revealed significant correlations between all laxity indices and both body mass (r = 0.29–0.52, P < 0.001) and height (r = 0.20–0.53, P < 0.001–0.014). Correlations with body mass index (BMI) were not significant, or weak (r = 0.09–0.26, P = 0.004–0.265). Therefore, allometric scaling was used to compute an index-specific body massx for inclusion as a covariate in these analyses, where x = the slope of the log-transformed laxity index plotted against log-transformed body mass. The values of x were −0.97 for varus angular laxity, −0.71 for valgus angular laxity, −0.85 for total angular laxity, 0.56 for varus end-range stiffness, 0.43 for valgus end-range stiffness, and 0.99 for midrange stiffness. The adjusted models were followed by pairwise contrasts to locate the source of any significant differences. An a priori alpha level of 0.05 was set for all of the comparisons.

RESULTS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. SUBJECTS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. AUTHOR CONTRIBUTIONS
  8. Acknowledgements
  9. REFERENCES

Descriptive information about the OA and control groups is provided in Table 1. Groups were comparable for height and sex distribution (P > 0.05). Moderate and severe OA groups were significantly older, heavier, and had a higher BMI compared with controls, and in some cases, the mild OA group (P ≤ 0.05). Varus knee alignment was significantly greater in moderate OA compared with mild OA, and in severe OA compared with mild and moderate OA (P ≤ 0.05).

Table 1. Descriptive information for the OA and control groups*
VariablesControls (n = 32)Mild OA (n = 50)Moderate OA (n = 45)Severe OA (n = 32)
  • *

    Values are the mean ± SD unless otherwise indicated. A lower degree of knee alignment indicates greater varus. OA = osteoarthritis.

  • P ≤ 0.05 versus controls and mild OA.

  • P ≤ 0.05 versus controls.

  • §

    P ≤ 0.05 versus mild OA.

  • P ≤ 0.05 versus mild OA and moderate OA.

Age, years59.39 ± 6.9261.61 ± 7.1265.23 ± 7.7266.40 ± 9.39
Women, no. (%)17 (53)27 (54)20 (44)11 (34)
Height, meters1.68 ± 0.091.67 ± 0.101.69 ± 0.091.69 ± 0.08
Mass, kg75.06 ± 10.6778.99 ± 16.2285.43 ± 11.9984.77 ± 14.70
Body mass index, kg/m226.62 ± 2.8228.37 ± 4.7230.06 ± 4.3129.72 ± 5.20
Knee alignment, degrees178.90 ± 2.47176.73 ± 2.28§173.82 ± 2.58

Unadjusted and adjusted angular laxity and stiffness data are shown in Table 2. Group mean varus and valgus angles ranged from ∼8.5° to 10.5°. Total angular laxity ranged from ∼17.5° to 20°. There was no significant difference in varus, valgus, or total angular laxity between groups (P > 0.05).

Table 2. Unadjusted and adjusted angular laxity and stiffness in asymptomatic knees and knees with osteoarthritis (OA)
VariableControls, mean ± SDMild OA, mean ± SDModerate OA, mean ± SDSevere OA, mean ± SD
  • *

    These data were log-transformed for statistical analysis.

  • P < 0.05 versus controls (generalized linear model with age, sex, and body massx as covariates).

  • P < 0.01 versus controls (generalized linear model with age, sex, and body massx as covariates).

Unadjusted    
 Angular laxity, degrees    
  Varus−10.7 ± 3.7−10.5 ± 3.6−9.2 ± 2.7−8.5 ± 2.7
  Valgus8.4 ± 3.29.6 ± 3.28.8 ± 2.49.2 ± 2.9
  Total19.2 ± 6.520.1 ± 6.418.0 ± 4.717.7 ± 5.4
 Stiffness, Nm/°    
  Varus1.50 ± 0.481.43 ± 0.411.54 ± 0.421.76 ± 0.54
  Valgus1.70 ± 0.441.62 ± 0.421.69 ± 0.331.77 ± 0.38
  Midrange1.62 ± 0.681.42 ± 0.581.50 ± 0.581.43 ± 0.62
Adjusted    
 Angular laxity, degrees    
  Varus−10.1 ± 2.7−10.2 ± 2.6−9.6 ± 2.7−9.1 ± 2.7
  Valgus8.0 ± 2.79.3 ± 2.69.1 ± 2.69.6 ± 2.6
  Total18.1 ± 5.019.5 ± 4.818.7 ± 4.818.7 ± 4.8
 Stiffness, Nm/°    
  Varus*1.47 ± 0.471.39 ± 0.431.46 ± 0.461.64 ± 0.51
  Valgus1.75 ± 0.391.65 ± 0.371.66 ± 0.381.73 ± 0.38
  Midrange*1.60 ± 0.701.36 ± 0.581.34 ± 0.571.22 ± 0.52

Group mean end-range stiffness ranged from ∼1.4 Nm/° to 1.8 Nm/°. This indicates that while approaching the end range, ∼1.4 Nm to 1.8 Nm of torque must be applied to rotate the knee joint by 1°. There were no differences in varus or valgus end-range stiffness between the experimental groups (P > 0.05). However, midrange stiffness was significantly lower in the mild, moderate, and severe OA groups compared with the control group (P = 0.036). The SEM for midrange stiffness was less than the statistical differences observed, being 0.049 Nm/°. Therefore, the observed differences are deemed to reflect a real difference between groups rather than measurement error. Presented in percentage terms, the control group was 15%, 16%, and 24% stiffer in the midrange compared with the mild, moderate, and severe OA groups, respectively.

DISCUSSION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. SUBJECTS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. AUTHOR CONTRIBUTIONS
  8. Acknowledgements
  9. REFERENCES

Frontal plane laxity is a feature of the local mechanical environment thought to relate to the presence and severity of knee OA. To our knowledge, this is the first study to investigate differences in passive knee angular laxity and stiffness, separately in varus and valgus, across controls and people with differing severity of medial tibiofemoral OA. In our study, varus, valgus, and total angular laxity did not differ between the groups. Similarly, end-range stiffness did not differ between the groups. Of importance, the midrange knee stiffness was found to be lower in knees with OA compared with control knees. This may dispose the OA knee to abnormal motion and load during functional activities.

At the 12 Nm threshold employed in the current study, we observed no differences between groups in total frontal plane angular laxity. Although increased (13) and decreased (15) total angular laxity has been reported with the presence and increasing severity of radiographic OA, our data are consistent with others that report no difference (16). This apparent ambiguity in the effect of OA on total angular laxity is likely to be related to differences in methodologic approaches across studies. In contrast to others, we have accounted for the relationship between joint mechanical properties and age, sex, and body mass. Although others have considered the relationship with BMI (13), this is not an absolute measure of body size and thus will not scale allometrically. In this study and others (17), we have demonstrated that frontal plane laxity indices are significantly correlated with absolute measures of body size such as height and body mass, but poorly correlated with BMI. Therefore, to remove potential bias due to body size (heavier people tend to be less lax and stiffer), our data were co-varied for the empirically derived exponent of body mass for each individual laxity measure. In addition, others have demonstrated a relationship of both sex and age with passive joint soft tissue mechanical properties (34–38). Age and sex were therefore also included as covariates in our statistical comparisons. Therefore, we believe that any potential bias that may have arisen from the reported age, sex, and body mass differences between our experimental groups has been eliminated.

Similar to other techniques quantifying knee joint laxity, our method rotated the shank relative to the thigh with a torque applied in either a varus or valgus direction. By continuously measuring the knee joint torque and angle and defining the neutral position at zero torque, we were able to distinguish varus and valgus laxity from total laxity as used in other studies of laxity in knee OA. Contrary to our expectations, no differences were observed in varus, valgus, or total angular laxity, despite being statistically powered to detect such differences. Conceivably, concurrent changes in various joint tissues may have masked individual effects of specific tissues on these laxity indices. For example, with increasing disease severity, medial compartment osteophytosis may reduce varus and total angular laxity (9); however, this effect may be balanced by increased laxity due to medial compartment cartilage loss. Irrespective of the underlying pathologic changes, because there was no net difference in these laxity indices, they are unlikely to play a significant role in the altered functional ability and loading mechanics observed with OA.

We measured the passive mechanical stiffness of the knee under varus loading, valgus loading, and within the midrange. Our data, the first to our knowledge to quantify in vivo frontal plane passive stiffness in OA knees, is in the range of 1–2 Nm/°. This agrees with the midrange stiffness of cadaver knees tested under similar conditions (25, 39). No differences were identified in varus or valgus end-range stiffness between the groups, and post hoc analysis indicated that we were powered to detect differences in these indices. This suggests that there is no effect of OA pathology on the net passive stiffness of the structures involved in this movement. That is not to say that individual structures are not affected, but rather, that it is their combined effect that is measured with this in vivo test.

To our knowledge, this is the first study to quantify the passive midrange stiffness of the OA knee, a position close to its operating range during gait, albeit at lower loading magnitude and rates than those likely to be experienced during gait. We found that OA knees were less stiff compared with asymptomatic control knees. Lower midrange stiffness indicates that the passive structures of the knee will offer less resistance to rotational motion when the joint is loaded while orientated close to its neutral alignment, such as the varus loading experienced during the stance phase of gait (8). Importantly, however, it is the net stiffness of the knee joint that will determine the effect of varus loading on joint kinematics. The passive stiffness (as measured in the current study) is only one component of this; the level of muscle activity around the joint will play an instrumental role in determining net joint stiffness. If on the one hand, the lower passive stiffness observed in OA knees translated to reduced net stiffness, combined with the higher adduction moments reported with OA (5), abnormal joint motion may occur. This abnormal motion may result in deleterious stress being placed on regions of cartilage that are unaccustomed to such high load; indeed, abnormal knee motion has been highlighted as a potential precursor for OA (2). Despite the relatively large percentage differences in passive midrange stiffness between groups (15–24%), the absolute differences are relatively small (0.24–0.38 Nm/°). Due to the relative scarcity of research investigating the clinical impact of laxity in knee OA, the clinical implications of our findings are unknown at present. Further research is required to determine if the magnitude of these absolute differences in passive stiffness are indeed large enough to have clinically significant consequences for joint health.

An alternative scenario, however, where the net stiffness of the knee with OA during gait is equivalent, or in fact greater than, that of a healthy knee, is perhaps more likely (40, 41). It is hypothesized that the increased muscle activation and co-contraction observed around the OA knee (42, 43) may occur to compensate for less passive stiffness (44). Although such a compensation may limit the possible detrimental effects of abnormal joint motion resulting from less passive stiffness, co-contraction may also lead to increased compressive joint loading and accelerated disease progression (45). Future research should evaluate whether reduced passive stiffness is indeed accompanied by increased active musculotendinous stiffness during gait in OA.

Self-reported “stiffness” of the knee is common in OA (46), and this term has mechanical connotations. This self-report measure, however, bears little relation to the mechanical stiffness of the knee (40, 47), and may be more closely related to pain sensation. Indeed, “stiffness” as defined by the Western Ontario and McMaster Universities Osteoarthritis Index is reduced with pain relief (48). It is therefore important that a clear distinction is made between self-report measures and function of the mechanical knee “system.” Therefore, our objective measures of passive frontal plane knee stiffness should not be likened to self-reported “stiffness” symptoms of knee OA. Our measures provide potentially valuable information about differences in the local mechanical environment (rather than symptoms) with the presence of medial knee OA, i.e., reduced midrange stiffness with the presence of medial knee OA.

There are some limitations to our study. The cross-sectional design of this study means it is not possible to determine the cause-and-effect relationship between angular laxity, stiffness, and OA disease development. Further longitudinal work is required to establish this. We cannot conclusively rule out the presence of asymptomatic knee OA in our control group due to the lack of radiographs because of ethical and financial constraints. We did ensure, however, that these participants had no knee symptoms or any history of knee surgery likely to predispose to OA. Another limitation of the current study is that we were unable to confirm the cruciate ligament or meniscal status of participants. Because the menisci and cruciate ligaments play a role in frontal plane laxity, albeit small (25, 49), it is possible that their involvement may have affected measured laxity. In a study of OA knees, however, using a methodology similar to ours, Wada et al reported no differences in total angular laxity between those with intact, partially ruptured, or missing cruciate ligaments (16).

Our methodology for the assessment of varus–valgus laxity has enabled the evaluation of passive motion and stiffness in a nonfunctional situation. Although the position of the knee is similar to that observed during gait, the passive behavior of the joint may differ between gait and our assessment. Furthermore, although the importance of not activating the muscles around the knee was stressed to participants, muscle contraction may have occurred during the test. Muscle “guarding” of the knee would be more likely in the OA groups. Therefore, if unreported muscle contraction was present, it is most likely that the underlying difference in passive midrange stiffness would be greater than that observed in the current study.

In summary, we observed no differences in varus, valgus, and total angular laxity with the presence or increasing severity of medial knee OA. Our finding of no difference in angular laxity indices may be indicative of the competing effects of changes within the joint itself and the surrounding soft tissues, i.e., collateral ligaments. The passive stiffness of the knee under varus and valgus loading also appears unchanged with OA; however, the position of the knee during these measurements is outside of its operating range during functional activities. Of importance, in the functional midrange, passive knee stiffness was lower in OA knees compared with controls. This may have important implications for dynamic joint function. Further work is merited to establish how attenuated passive stiffness in OA affects gait mechanics and disease progression.

AUTHOR CONTRIBUTIONS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. SUBJECTS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. AUTHOR CONTRIBUTIONS
  8. Acknowledgements
  9. REFERENCES

All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be submitted for publication. Dr. Creaby had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study conception and design. Creaby, Wrigley, Hinman, Bennell.

Acquisition of data. Creaby, Lim, Bowles, Metcalf.

Analysis and interpretation of data. Creaby, Wrigley, Metcalf, Bennell.

Acknowledgements

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. SUBJECTS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. AUTHOR CONTRIBUTIONS
  8. Acknowledgements
  9. REFERENCES

We wish to thank Georgina Morrow and Fiona McManus for assisting with participant recruitment.

REFERENCES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. SUBJECTS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. AUTHOR CONTRIBUTIONS
  8. Acknowledgements
  9. REFERENCES