Abstract
Background The added value of magnetic resonance imaging (MRI) in primary care is still being debated. A high diagnostic yield can be expected in young and active patients with post-traumatic knee complaints.
Aim To determine the frequency of MRI abnormalities in young and active patients (aged 18–45 years) and the associations with patient, trauma, and clinical characteristics.
Design and setting A subgroup analysis of 174 patients, aged 18–45 years with knee trauma of <6 months, allocated to MRI in a randomised controlled trial on the yield of MRI in primary care. Patients were recruited by 150 GPs in the Netherlands from October 2012 to November 2015.
Method Associations were expressed using mean differences, odds ratio (OR) and predictive values.
Results Sixty-seven out of 174 patients (39%) had a positive MRI finding, predominantly anterior cruciate ligament (ACL) ruptures (22%) and/or traumatic meniscal tears (15%). Patients with a pre-existing musculoskeletal comorbidity had a two-fold lower prevalence of positive MRI findings (21%), OR 3.0 (95% confidence interval [CI] = 1.3 to 7.0). A ‘sports related trauma’ showed the highest OR of 4.6 (95% CI = 2.2 to 9.3) for a positive MRI finding. Clinical scores were statistically, significantly worse in patients with positive MRI findings, with mean differences ranging from 10 to 20%. Furthermore, increasing duration of complaints was correlated with decreasing prevalence rates of positive MRI findings. Overall, a popping sound and direct swelling showed the highest positive predictive value of 65% for the presence of positive MRI findings.
Conclusion The results from this study enable a preselection of patients to increase the diagnostic yield of MRI in primary care.
- anterior cruciate ligament injuries
- early diagnosis
- knee injuries
- magnetic resonance imaging
- tibial meniscus injuries
INTRODUCTION
GPs are often consulted for post-traumatic knee complaints.1 Of all sport injuries, the knee is the most commonly affected joint.2 A trauma of the knee is known to have the highest burden of all sport injuries causing a substantially lowered quality of life in predominantly young and physically active patients.2,3 These patients suffer from knee pain and impaired knee function resulting in less mobility, work absenteeism, and decreased sports participation.2,3 Post-traumatic knee injuries can be diagnosed with high diagnostic accuracy using magnetic resonance imaging (MRI).4 As an established diagnostic tool in secondary care, it is mainly requested by orthopaedic surgeons to select treatment strategy, including whether surgery is needed.5,6 During the past decades MRI has been increasingly used in primary care, with the MRI of the spine requested most, followed by the knee and the brain respectively.5,7,8
Negative MRI findings could reassure patients and treatment in primary care can be continued. In case of positive findings, an early MRI scan could contribute to an earlier diagnosis and dedicated treatment. Patients could return to sports or work earlier with possible improved healthcare outcomes. A high diagnostic yield can be expected especially in young patients with post-traumatic knee complaints.5,9–11 Earlier studies in primary care have evaluated prevalence rates of MRI abnormalities in patients with knee complaints, however, these studies included older patients and patients with chronic knee complaints.6,11–13 None of these studies assessed the associations of clinical outcomes with the overall MRI result, including combinations of important MRI findings.
The goal of this study was, therefore, to determine the frequency of MRI findings in younger patients (18–45 years) with post-traumatic knee complaints seen in primary care and to determine the associations with patient and trauma characteristics; duration of complaints; and clinical scores.
METHOD
Study design
For the current study, data were used from the TraumAtic Complaints of the Knee — Leiden University Medical Center (LUMC) and Erasmus MC (TACKLE) Trial, a multicentre randomised controlled trial (RCT) to determine the (cost-) efficacy of MRIs requested by the GP in patients with a recent knee trauma. The full research protocol was published earlier.14 Patients aged 18–45 years with knee complaints after a recent knee trauma in the preceding 6 months were eligible. Exclusion criteria were indications for direct referral (fracture or a locked knee), referrals to secondary care and/or MRI already performed, previous surgery of the affected knee, knee osteoarthritis diagnosed by a physician and contraindications for MRI. Patients were enrolled from October 2012 to November 2015 by 150 GPs in the western part of the Netherlands. Eligible patients were selected during consultation by the GP or by an invitational letter from the GP within 12 weeks after the consultation. After the baseline measurement, patients were evenly allocated to usual care or to MRI. In the MRI arm patients received an MRI scan within 2 weeks after the baseline measurement, as well as usual care. Only the results of the MRI group are presented in the current study.
How this fits in
At a time of on ongoing discussion about the added value of magnetic resonance imaging (MRI) in primary care, a subgroup analysis was conducted with all patients allocated to MRI in a randomised controlled trial. A potential subgroup of patients that could benefit from MRI was studied. It contained relatively young patients (18–45 years) with post-traumatic knee complaints. The MRI findings were expounded and several patient, trauma, and clinical characteristics associated with higher prevalence rates of positive MRI findings were identified. The study findings help GPs to preselect patients with post-traumatic knee complaints in whom an increased diagnostic yield of MRI can be expected.
MRI abnormalities
Prior to the start of this study, two orthopaedic surgeons, employed in two different participating centres of the TACKLE Trial, identified ‘positive’ MRI findings that may have needed further specialised assessment and treatment by an orthopaedic surgeon. These were a trabecular fracture, a complete rupture of a collateral ligament, a meniscus tear, a cruciate ligament rupture and a full thickness cartilage defect. A standardised knee MRI report was developed and the involved radiologists were trained accordingly (further details on MRI imaging in the TACKLE Trial are available from the authors).
Patient, trauma, and clinical characteristics
The following patient characteristics and data were collected:
age;
sex;
body mass index (BMI);
presence of musculoskeletal comorbidities, defined as pre-existing pain in the hip/s, ankle/s or spine;
duration of complaints;
trauma characteristics, including the activity during trauma (sports, work, at home, traffic); trauma mechanism (rotational, fall, bumping); a popping sensation; direct pain, direct swelling inability to continue activity and combinations.
clinical scores — knee injury and osteoarthritis outcome score (KOOS) pain; KOOS symptoms; and KOOS sports.15
Statistical analysis
The data were described using absolute numbers with frequencies for binary and categorical data, means with standard deviation (SD) for normally distributed data and medians with interquartile ranges (IQR) for non-normally distributed data. Several categorical variables were dichotomised into sensible and clinically relevant variables, for example, sports-related versus non-sports-related trauma. Prior to the start of the study all meniscal tears were defined as a positive MRI finding. However, in the analyses, presence of a horizontal meniscal tear was associated with higher age and not associated with duration of complaints, nor with any clinical outcome measure. Therefore, patients with a horizontal meniscal tear were excluded from the positive MRI findings group. Furthermore, presence of effusion and/or synovitis were combined into one variable ‘effusion-synovitis’ to assess the clinical correlations. The associations of MRI findings with the collected patient, trauma, and clinical characteristics were expressed using mean differences for continuous data assessed with the Student’s t-test and odds ratios (OR) for dichotomous data using binary logistic regression analysis. Additionally, 95% confidence intervals (95% CIs) for all associations were calculated. The mean difference was preferred because this measure provides more directly relevant information than the correlation coefficient.16 Furthermore, negative predictive values (NPVs) and positive predictive values (PPVs) were calculated for the trauma characteristics and the presence of MRI findings, using cross tabulations. To express the association of duration of complaints with the presence of MRI findings and with mean clinical scores, patients were stratified into three equal groups according to (increasing) days from trauma to MRI scan. The statistical significance of differences in prevalence and mean scores across the three strata were determined using the 1-way ANOVA test. Statistical analyses were performed with IBM SPSS (version 23.0).
RESULTS
Study population
In the TACKLE trial 174 patients received an MRI scan (Table 1). The median time from baseline measurement to MRI scan was 6 days (IQR 4–9 days), with 93% of the patients receiving the MRI scan within the scheduled 2 weeks after randomisation. Median duration of knee complaints was 42 days (IQR 17–83 days).
Prevalence of MRI findings
A positive MRI finding was present in 67 patients (39%), predominantly anterior cruciate ligament (ACL) ruptures in 38 patients (22%) and/or a traumatic meniscal tear in 26 patients (15%) (Table 2). A positive MRI finding accompanied with effusion-synovitis was found in 49 patients (73%); in 43 patients (64%) accompanied by a bone bruise; and in 59 patients (88%) accompanied by effusion-synovitis and/or a bone bruise (Figure 1). No abnormalities were found in 30 patients (17%) found, leaving 77 patients (44%) with minor MRI findings who did not require an evaluation by an orthopaedic surgeon.
Associations of MRI findings with patient and clinical characteristics
Higher age was correlated with the presence of a horizontal meniscal tear, but with no other MRI abnormality. Patients with a horizontal meniscal tear had a mean age of 38 (SD 7) years, compared to a mean age of 32 (SD 8) years in patients without a horizontal meniscal tear; mean difference 6 (95% CI = 3 to 9) years. Furthermore, patients with pre-existing musculoskeletal comorbidities had a statistically significant lower prevalence of positive MRI findings: 8 out of 39 patients (21%) compared to 59 out of 135 patients (44%) without a musculoskeletal comorbidity; OR 3.0 (95% CI = 1.3 to 7.0). No statistically significant associations were found between BMI, or sex, and the presence of MRI findings.
As the duration of complaints increased, fewer MRI abnormalities were observed and fewer knee specific symptoms were reported (Table 3). Only patients with a horizontal meniscal tear were evenly distributed across the three strata.
The group with positive MRI findings had statistically significant worse scores on all three tested KOOS subscales, with the highest mean difference of 22 (95% CI = 15 to 30) in the KOOS sports score. Most single positive MRI findings showed a statistically significant association with clinical scores, with the exception of the horizontal meniscal tear. Effusion/synovitis and the presence of a bone bruise were associated with significantly worse scores on the three KOOS subscales, with mean differences ranging 10–19 (95% CI = 4 to 27).
Associations of MRI findings with trauma characteristics
All trauma characteristics were statistically significantly associated with the presence of one or more positive MRI finding, of which sports-related trauma showed the strongest association (Table 4). Of the six trauma characteristics, four were associated with a higher prevalence of ACL rupture, of which ‘direct pain after trauma’ showed the strongest association with an OR of 3.9 (95% CI = 1.3 to 11.8). Overall high NPVs were observed: highest for sports-related trauma and direct pain after trauma, with NPVs ranging from 90% to 93%. Furthermore, highest PPVs for a positive MRI finding were seen for the combinations of a popping sound during trauma and a direct swelling, and for a rotational trauma with a direct swelling, with PPVs ranging from 62% to 65% (Table 4). The classical combination of a popping sound and a direct swelling showed a 41% PPV for an ACL rupture.
DISCUSSION
Summary
In this study the MRI findings in young adults with a recent knee trauma seeking medical attention in primary care were investigated. Almost 40% of the patients had one or more positive MRI findings potentially requiring further evaluation by an orthopaedic surgeon. These findings were usually accompanied by effusion-synovitis and/or a bone bruise. Patients without a pre-existing musculoskeletal comorbidity showed a twofold higher prevalence of these positive MRI findings. Increasing duration of complaints was associated with decreasing prevalence rates of positive MRI findings. Six trauma characteristics showed a statistically significant association with positive MRI findings of which sports-related trauma showed the highest OR. A popping sound with a direct swelling showed the highest PPV for a positive MRI finding. Patients with a positive MRI finding showed statistically significantly more impaired knee function and more knee pain.
Strengths and limitations
A strength of this study is the selection of patients in which the highest yield of MRI imaging is expected, including young patients aged 18 to 45 years with post-traumatic knee complaints, thereby excluding older patients with possible degenerative findings that are often confusing to clinicians and lack therapeutic options. Furthermore, the overall MRI result was focused on and correlated with clinical scores, not exclusively concentrating on single ligamentous or meniscal injuries.
There were some limitations. No information was gathered on the findings of physical examinations performed by the GP. This decision was based not only on logistical reasons with the primary goal of the TACKLE trial being the yield of MRI imaging in primary care, but also on evidence from the literature showing the minor to absent added value of physical examination of the knee by GPs for the diagnosis of ACL ruptures and meniscal tears.13,17,18 Furthermore, the tested associations in this study were based on MRI findings and not confirmed by arthroscopy. However, MRI imaging is known to have high diagnostic accuracy for post-traumatic knee injuries.19 A few false positive or false negative findings could have been made, but were unlikely to influence the main results.
Comparison with existing literature
Prevalence rates of MRI findings in this study are comparable with earlier studies in patients with post-traumatic knee complaints seen in primary care. In the present sample, 15% of the patients had a traumatic meniscal tear, this is comparable with the prevalence rate of 17% in a similar group of primary care patients.12 An overall meniscal tear prevalence rate of 31% was observed in this study, in contrast to 47% in the aforementioned study. This could be due to the relatively older patients included in their study with a mean age of 41 years.12 For ACL ruptures, those authors found a prevalence rate of 17%. In another study in primary care (with a trauma rate of 66%) 18% of the patients had an ACL rupture; this was 22% in the present cohort.13
Another marked finding was the relative high frequency (17%) of horizontal meniscal tears in the present targeted population, which were not associated with higher scores on pain or function loss. A horizontal tear is generally believed to be of degenerative origin. First a linear signal change inside the meniscus develops, reflecting slowly developing meniscal degradation; most of these signal changes later progress to horizontal meniscal tears.20 Earlier studies already suggested these tears were an incidental finding, but these studies were performed in older patients.12,20,21 Results from this study provide additional evidence that an isolated horizontal meniscal tear should be considered an age-related phenomenon, not requiring referral to an orthopaedic surgeon, even in relatively young adults with post-traumatic knee complaints.
An important finding in this study was the decreasing prevalence rates of most MRI findings with increasing duration of complaints. This was also true for the prevalence rates of positive MRI findings. The clinical outcome scores were expected to improve over time but this clear association was not foreseen. An earlier study assessed duration of complaints and the association with the prevalence of positive MRI findings, but found no statistically significant difference.22 However, in that study, the mean duration of symptoms was over 6 months, a time period in which the effect is probably diminished. The finding in this study can be explained by the natural recovery of patients in which bone bruises resolve, effusion diminishes, and injuries to the meniscus and collateral ligaments heal. There is also evidence that partial and even complete ACL ruptures can regain continuity over time, although, the time span in this study might be too short for this event to occur.23 Another possible explanation is that earlier consultation was achieved by patients with more severe complaints, resulting in a shorter period from trauma to MRI imaging.
A new finding in this study is the association of pre-existing musculoskeletal comorbidities with a lower prevalence of positive MRI findings. A musculoskeletal comorbidity is known to be associated with a higher score on knee specific pain subscales,24 but to the authors’ knowledge, no earlier studies reported on the lower prevalence of MRI findings in these relatively younger patients with recent knee trauma. Future studies should confirm the finding in this study. The strongest association of a trauma characteristic with positive MRI findings was ‘sports-related trauma’, with a 30% higher prevalence rate of positive MRI findings compared to a non-sports-related knee trauma. This association was not found in previous studies.17,18,25 However, a recent study confirmed a statistically significant association of a sport-related trauma with a higher prevalence of medial meniscal tears, compared to patients with non-sports-related knee trauma.26
Implications for practice
This article presents an overview of MRI findings observed in a population of young patients, aged 18–45 years, with traumatic knee complaints. Several patient, trauma, and clinical characteristics associated with positive MRI findings were identified. Duration of complaints should be considered by the GP when referring a patient for an MRI scan to appraise the possible yield of MRI imaging. Sports-related trauma is a strong predictor for positive MRI findings, while the presence of a musculoskeletal comorbidity results in less positive MRI findings. Furthermore, horizontal meniscal tears are, even in relatively young adults with a recent knee trauma, an incidental finding without clinical relevance. The predictive values of the trauma characteristics could enable a preselection of patients with a greater likelihood of having an abnormal MRI scan, thereby increasing the yield of MRI in primary care.
Acknowledgments
The authors are indebted to all patients for their contribution in this trial and would like to thank all GPs involved in this study for the recruitment of patients. Furthermore, the authors thank all radiologists for the reporting of the MR scans made in this trial.
Notes
Funding
This trial was financially supported by the Netherlands Organisation for Health Research and Development (ZonMW), (project number 171202005).
Ethical approval
All patients approved and signed the informed consent form. The study has been approved by the Medical Ethics Committee of the Erasmus Medical Centre (Dutch Trial Registration: NTR3689).
Provenance
Freely submitted; externally peer reviewed.
Competing interests
The authors have declared no competing interests.
Discuss this article
Contribute and read comments about this article: bjgp.org/letters
- Received March 1, 2017.
- Revision requested May 15, 2017.
- Accepted July 6, 2017.
- © British Journal of General Practice 2017