If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
MSk Lab, Department of Surgery and Cancer, Imperial College London, London, UKHealth Education Kent, Surrey and Sussex Higher Surgical Training Programme, London, UKDunhill Medical Trust and Royal College of Surgeons of England Joint Research Fellowship, London, UK
Address correspondence and reprint requests to: R.J. van Arkel, Department of Mechanical Engineering, Imperial College London, Exhibition Road, London, SW7 2AZ, UK.
For a population with knee osteoarthritis (OA), determine: 1) the prevalence of single compartmental, bicompartmental and tricompartmental OA, 2) the prevalence of isolated medial tibiofemoral, lateral tibiofemoral, or patellofemoral OA, and combinations thereof.
Methods
PubMed and Web of Science databases, and reference lists of identified studies, were searched to find studies which reported on the compartmental distribution and prevalence of knee OA. Two independent reviewers assessed studies against pre-defined inclusion criteria and prevalence data were extracted along with subject characteristics. The methodological quality of each included study was assessed. A random-effects model meta-analysis was performed for each OA category to estimate the relative prevalence of OA in the knee compartments amongst people with knee OA.
Results
16 studies (3,786 knees) met the inclusion criteria. High heterogeneity was measured. Normalised for knees with OA, estimated prevalence rates (95% CI) were: single compartmental 50% (31.5–58.3%), bicompartmental 33% (23.1–37.2%) and tricompartmental only 17% (8.8–24.8%). Isolated medial tibiofemoral OA, isolated patellofemoral OA, and combined medial tibiofemoral and patellofemoral OA were more common than tricompartmental disease, occurring in 27% (15.2–31.1%), 18% (9.9–22.7%) and 23% (14.1–27.3%) of people respectively. Single/bicompartmental patterns of disease involving the lateral tibiofemoral compartment were less common, summing to 15% (8.5–18.7%).
Conclusion
Three-quarters of people with knee OA do not have tricompartmental disease. This is not reflected in the frequency with which partial and combined partial knee arthroplasties are currently used.
Osteoarthritis (OA) is the deterioration of the articular cartilage and subchondral bone of the joint - leading to pain, loss of function and ultimately destruction of the joint
. In the UK, OA is the most prevalent joint disease, and the knee is the most commonly affected joint: a fifth of people aged over 45 years will develop knee OA in their lifetime
undergoing a total knee arthroplasty (TKA), where all three knee compartments are replaced with monolithic femoral and tibial components. However, single compartmental disease can be treated with partial knee arthroplasty (PKA), using medial or lateral unicondylar or patellofemoral implants. There is also renewed interest in combining PKA implants to treat bicompartmental disease
; in June 2018, the National Joint Registry of England, Wales and Northern Ireland modified its data collection to record combined partial knee arthroplasties (CPKA) as primary procedures
, but has been associated with a greater risk of serious complications such as myocardial infarction, stroke, and deep infection, as well as higher mortality
Adverse outcomes after total and unicompartmental knee replacement in 101330 matched patients: a study of data from the National Joint Registry for England and Wales.
Unicompartmental knee arthroplasty vs total knee arthroplasty for medial compartment arthritis in patients older than 75 years: comparable reoperation, revision, and complication rates.
Is unicompartmental knee arthroplasty (UKA) superior to total knee arthroplasty (TKA)? A systematic review and meta-analysis of randomized controlled trial.
Larger range of motion and increased return to activity, but higher revision rates following unicompartmental versus total knee arthroplasty in patients under 65: a systematic review.
. Hence, it is of great interest to investigate how knee OA is distributed across compartments to identify the proportion of patients that could be suitable for PKA compared to those for whom the only option is TKA. Moreover, these data will also prove useful as advances in cartilage tissue engineering translate to clinical practice; understanding the pattern of disease will enable more targeted therapies for focal knee OA
: single compartmental OA includes-isolated medial tibiofemoral, isolated lateral tibiofemoral, isolated patellofemoral OA; bicompartmental OA includes-combined medial and lateral tibiofemoral, combined medial tibiofemoral and patellofemoral, combined lateral tibiofemoral and patellofemoral OA; and tricompartmental OA is when disease is located in all three compartments.
Previous systematic reviews have reported on the compartmental prevalence of OA in the knee and highlighted the mismatch between the prevalence of isolated patellofemoral OA and rates of patellofemoral arthroplasty
. To enable comparison with arthroplasty treatment options, it is necessary to analyse this further, separating out the medial and lateral tibiofemoral compartments. The aim of this study is thus to quantify the relative prevalence and location of single and multi-compartmental knee OA within the population of patients with knee OA through a systematic literature review with meta-analysis.
Objectives
To determine for a population with knee OA:
1)
The extent of disease; the prevalence of single compartmental, bicompartmental and tricompartmental OA (Fig. 1)
2)
The location of disease; the prevalence of the seven patterns of knee OA: isolated medial tibiofemoral, isolated lateral tibiofemoral, isolated patellofemoral, combined medial tibiofemoral and patellofemoral, combined lateral tibiofemoral and patellofemoral, combined medial and lateral tibiofemoral, and in all three compartments (Fig. 1)
Methods
Protocol and registration
Prior to investigation, the protocol was registered with the International prospective register for systematic reviews PROSPERO (ID: CRD42019140345), following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement and checklist.
Eligibility criteria
Studies which included participants with ‘confirmed’ knee OA, with or without simultaneous OA in other joints were included. OA could be confirmed intraoperatively, or through any imaging modality. Radiographically, OA was considered present through assessment with an established atlas (e.g., Kellgren-Lawrence
). Scoring systems for the severity of OA intraoperatively, using magnetic resonance imaging (MRI), or any other imaging modalities, are less established, and so the structural features observed (cartilage defects, bone marrow lesions, osteophytes etc.,), along with assessments such as the Whole-Organ MRI score
, were used to define the presence or absence of OA. No lower limit was set on the severity of the OA structural changes detected, though studies were later grouped by the minimum OA severity measured to enable subgroup analyses.
Studies had to assess the presence of knee OA by compartment for inclusion in the meta-analysis with the frequency of the combinations of compartments involved reported in full, not simply the frequency with which each compartment was involved
. The medial and lateral tibiofemoral compartments had to be reported separately; analysis of studies where the tibiofemoral compartments were considered monolithically is included as Supplementary Material.
Animal, computational and case studies were excluded, as were studies which focused on disease in only one or two compartments, or stipulated disease in a specific compartment as a baseline for inclusion. Studies with subjects having inflammatory arthropathies such as rheumatoid arthritis and other systemic joint diseases were excluded.
Information sources and search strategy
Terms associated with “compartment”, “osteoarthritis”, “knee”, “prevalence” and “pattern” were searched (Supplementary Material) using the electronic databases PubMed and Web of Science up to 9th July 2020, with no other data restrictions. The reference lists of studies assessed for data extraction were also screened to identify additional relevant studies.
Study selection
Two independent reviewers (JCS, OD) assessed the titles and abstracts of all the studies identified and discarded studies that met any of the exclusion criteria. For example, a study where patient selection was based on the presence of medial tibiofemoral OA
was excluded. The full text of all remaining studies was then assessed against the inclusion criteria by the two independent reviewers; disagreements were resolved by a third reviewer (RvA).
Data collection process and data items
A standardised form was used for data extraction. Data relating to the extent and location of knee OA were extracted. The country, cohort type and size, number of knees with OA, proportion of female subjects, mean age and mean BMI of the subjects were recorded. The imaging modality, OA features used in assessment of the knee (e.g., osteophytes, joint space narrowing), OA atlas used (e.g., Kellgren & Lawrence, OARSI etc.,), minimum severity of OA recorded (‘Low’ or ‘High’), and whether the tibiofemoral compartment was considered monolithically or distinguished between its medial and lateral compartments, were also recorded.
Methodological quality was assessed with the Loney critical appraisal tool
. Two independent reviewers (JCS, OD) scored each study out of 8; disagreements were resolved by a third reviewer (RvA). Studies with score ≥6 were considered high quality, and ≤3 low quality.
Summary measures
Analyses followed the methodology described by Borenstein et al.
using Comprehensive Meta-Analysis Software (Biostat Inc., USA), with plots produced using MATLAB (The Mathworks, Natick, MA, USA). The summary effect was calculated for the extent and location of OA based on study weightings calculated according to the random effects model. This model was used as it was assumed that the true effect size varied from study to study, since the population characteristics and methodologies varied amongst the included studies. The resulting summary statistic estimated the mean of the distribution of the size of the true effects of all the studies. Two principal meta-analyses were performed: one for the extent of OA, one for the location of OA.
Since random-effects model weightings are non-linear
, the summary effects for the extent and compartmental distribution did not add to 100% (whilst the raw data from the individual studies did). Therefore, the pooled values were divided by the total, then multiplied by 100 to re-normalise the results. These normalised results were plotted as a pie chart (extent of OA) and Venn diagram (location of OA) and reported in the abstract and discussion.
Subgroup analyses
Pooled subgroup analyses were performed for:
•
minimum severity of OA (‘Low’/‘High’)
•
cohort type (symptomatic/population-based)
•
imaging method
•
geography
•
mean age
•
mean BMI
For each pooling, heterogeneity was assessed with the I2 statistic–the ratio of the true heterogeneity to the total observed variation. The significance of heterogeneity across studies was measured using the Q statistic–a measure of weighted squared deviations–testing the null hypothesis that all studies share a common effect size (α = 0.05) from a Chi-squared distribution with degrees of freedom, , based on the number of studies included in the meta-analysis, n. I2 is related to Q: .
Cumulative meta-analyses based on study size and study quality were performed to assess whether there was any discernible publication bias.
Results
Study selection and characteristics
173 unique records were identified from the databases (Fig. 2). 16 studies (3,786 knees) were suitable for inclusion in the principal meta-analysis
Incidence of bicompartmental osteoarthritis in patients undergoing total and unicompartmental knee arthroplasty: is the time ripe for less radical treatment?.
(Table I). The data extracted from these studies is available as a downloadable dataset. An additional 14 studies (3,847 knees–7,633 total) passed all inclusion criteria except for considering the tibiofemoral compartments monolithically, and have been analysed in the supplementary material
Gender difference in symptomatic radiographic knee osteoarthritis in the Knee Clinical Assessment - CAS(K): a prospective study in the general population.
High prevalence of knee osteoarthritis, pain, and functional limitations in female soccer players twelve years after anterior cruciate ligament injury.
‘Low’ group includes population-based studies with no OA threshold inclusion criteria, studies that did not use an established quantitative metric, and studies that included patients with K&L scores ≤2, or equivalent; ‘High’ group includes all papers with quantitative metric inclusion criteria of K&L ≥ 3, or equivalent.
Incidence of bicompartmental osteoarthritis in patients undergoing total and unicompartmental knee arthroplasty: is the time ripe for less radical treatment?.
∗ JSN = Joint space narrowing, BME = Bone marrow edema, OST = Osteophyte, K&L = Kellgren and Lawrence, PR = Patient reported.
† Paper only had sufficient data for inclusion in extent meta-analysis (single/bi/tricompartmental).
‡ ‘Low’ group includes population-based studies with no OA threshold inclusion criteria, studies that did not use an established quantitative metric, and studies that included patients with K&L scores ≤2, or equivalent; ‘High’ group includes all papers with quantitative metric inclusion criteria of K&L ≥ 3, or equivalent.
Mean methodological quality was 5.3 (range 3–8) (Supplementary Material). 88% of papers used standard measures to assess OA, had an adequate response rate (>70%), described refusers and described the study subjects. 63% had outcomes measured by unbiased assessors and reported confidence intervals. 44% of studies had a sample size >300, 25% used an unbiased sampling frame, and 31% used a random sample or whole population.
Results of principal meta-analyses
Extent of OA
The meta-analysis found differences between the relative prevalence of uni-, bi- and tricompartmental OA (Fig. 3). Normalised: 50% (31.5–58.3%) (normalised pooled estimate of prevalence rate (95% CI)) of the OA population had only one compartment assessed to have osteoarthritic changes, while 33% (23.1–37.2%) had bicompartmental and 17% (8.8–24.8%) had tricompartmental disease (Fig. 4). Results from the extended meta-analysis with the additional studies set at the midpoint of their possible range of values showed the same proportions as found in the principal meta-analysis (Supplementary Material).
Fig. 3Forest plots showing the estimated random-effect model prevalence and 95% CIs of single, bi- and tricompartmental knee OA (raw meta-analysis output, pre-normalisation). The sizes of the squares indicate the relative weight assigned to each study in the random-effect meta-analysis. The result of a meta-analysis of symptomatic cohorts (n = 3,266) for single (I2 = 98%, Q = 759), bi- (I2 = 95%, Q = 299) and tricompartmental (I2 = 98%, Q = 650) is shown below the result for all studies.
Fig. 4The extent of knee OA: pie chart demonstrating the normalised prevalence of single (dark purple), bicompartmental (mid green) and tricompartmental (light yellow) disease.
The compartmental distribution of OA was determined through the second meta-analysis (Fig. 5). OA affected the isolated medial tibiofemoral compartment most often; normalised it affected 27% (15.2–31.1%) of the knee OA population. The bicompartmental combination where both the medial tibiofemoral and patellofemoral compartments were involved was the next most common disease pattern (23% [14.1–27.3%]), followed by isolated patellofemoral OA (18% [9.9–22.7%]), and then tricompartmental OA (17% [7.4–22.2%]). The combination of lateral tibiofemoral and patellofemoral OA was estimated to occur in 5% (2.8–6.5%) of the OA population. Similarly, concurrent medial and lateral tibiofemoral OA had a prevalence of 5% (2.9–6.3%), and isolated lateral tibiofemoral OA had a normalised prevalence of 5% (2.8–5.9%) within the knee OA population (Fig. 6).
Fig. 5Forest plots showing the estimated random-effect model prevalence and 95% CIs of the location of knee OA (raw meta-analysis output, pre-normalisation). The sizes of the squares indicate the relative weight assigned to each study in the random-effect meta-analysis. The result of a meta-analysis of symptomatic cohorts (n = 3,029) for isolated medial tibiofemoral (I2 = 96%, Q = 377), isolated lateral tibiofemoral (I2 = 66%, Q = 38), combined medial and lateral tibiofemoral (I2 = 83%, Q = 75), isolated patellofemoral (I2 = 96%, Q = 289), combined medial tibiofemoral and patellofemoral (I2 = 96%, Q = 297), combined lateral tibiofemoral and patellofemoral (I2 = 79%, Q = 63) and combined medial and lateral tibiofemoral and patellofemoral (I2 = 98%, Q = 574) is shown below the result for all studies.
Fig. 6The location of knee OA: Venn diagram demonstrating the normalised prevalence of each pattern of disease. Each circle represents a compartment of the knee (top-patellofemoral compartment; left-medial tibiofemoral compartment; right-lateral tibiofemoral compartment). Non-overlapping regions represent single compartmental OA (dark purple), intersections between two circles (mid green) represents bicompartmental configurations, while the intersection of all 3 circles (light yellow) represents tricompartmental disease prevalence.
Limiting the meta-analysis to those examining symptomatic cohorts only had minimal effect on the results (Fig. 3 and Table II): the symptomatic subgroup analysis showed the normalised prevalence of single compartment OA decreased by 3 percentage points (PP) (from 50% to 47%), bicompartmental OA increased by 3 PP (from 33% to 36%), while tricompartmental OA prevalence was unchanged (17%). There were insufficient population-based cohorts for subgroup analysis.
Table IIEffect of running meta-analyses on subgroups of studies (‘Low’ vs ‘High’ OA minimum severity, Population-based vs Symptom-based cohorts, Radiograph vs other imaging modalities, UK population vs US population, Mean age >65 y vs < 65 y)
Category
Subgroup
n studies
Effect compared to no pooling (PP change from main analysis)
Single Compartment
Bicompartmental
Tricompartmental
Minimum OA severity
Low
9
↓ (−9 PP)
↑ (+1 PP)
↑ (+8 PP)∗
High
6
↑ (+8 PP)
↓ (−1 PP)
↓ (−7 PP)
Cohort type
Population-based
2
N/A
N/A
N/A
Symptom-based
15
↓ (−3 PP)
↑ (+3 PP)
–
Imaging Modality
Radiograph
11
↑ (+12 PP)∗
↓ (−5 PP)
↓ (−7 PP)
Other
5
↓ (−30 PP)∗
↑ (+10 PP)∗
↑ (+20 PP)∗
Geography
UK
5
↑ (+15 PP)∗
↓ (−3 PP)
↓ (−12 PP)∗
US
6
↓ (−33 PP)∗
↑ (+6 PP)∗
↑ (+27 PP)∗
Age
Under 65 y
6
↓ (−14 PP)
↑ (+2 PP)
↑ (+12 PP)∗
Over 65 y
8
↑ (+8 PP)
↓ (−3 PP)
↓ (−5 PP)
The arrows indicate whether normalised prevalence of single, bi- or tricompartmental OA increased (↑) or decreased (↓) compared to the full meta-analysis. The values give the percentage point (PP) change in the distribution compared to Fig. 4: low numeric PP values indicate that the subgrouping had little effect, whereas large numeric values indicate that the compartmental distribution is influenced by subgrouping. For each table row, the PP changes sum to zero.
An asterisk (∗) indicates when the subgroup result was outside of 95% confidence interval calculated for full meta-analysis.
The minimum severity of OA has a moderate effect on the compartmental distribution of OA (Table II). Limiting the meta-analysis to just those studies with a ‘High’ threshold for considering OA to be present (K&L ≥ 3 or equivalent)–a group more representative of an arthroplasty patient group
–increased the normalised rate of single compartment OA by 8 PP (from 50% to 58%), decreased the rate of bicompartmental OA by 1 PP (from 33% to 32%) and decreased tricompartmental OA by 7 PP (from 17% to 10%). The converse trend (decreased prevalence of single compartmental, bicompartmental largely unaffected, increased tricompartmental) was seen for those studies with a ‘Low’ threshold for reporting OA.
The imaging methodology had a large effect on the compartmental distribution of OA (Table II). Studies that used radiography were representative of the full meta-analysis: measured a 12 PP increase (from 50% to 62%) in the prevalence of single compartment OA, with a 5 PP decrease in the prevalence of bicompartmental OA (from 33% to 28%), and 7 PP decrease in tricompartmental OA (from 17% to 10%). However, when pooled according to alternative, typically more sensitive, imaging methods, a 30 PP decrease in single compartment was seen (from 50% to 20%), bicompartmental OA increased by 10 PP (from 33% to 43%) and tricompartmental by 20 PP (from 17% to 37%). These more sensitive imaging methods changed the prevalence ordering from single > bi- > tricompartmental to bi- > tri- > single compartmental.
Geographic location also had a large effect (Table II): the US subgroup had 33 PP decreased prevalence of single compartment OA (from 50% to 17%), balanced by increases in bicompartmental (6 PP from 33% to 39%), and tricompartmental (27 PP from 17% to 44%). The changes in the US subgrouping changed the prevalence order to tri-> bi-> single compartmental. The reverse trend was true for the UK subgroup and thus did not affect the prevalence ordering.
The age range of the cohorts had a moderate effect (Table II): mean age <65 years decreased the prevalence of single compartmental OA by 14 PP (from 50% to 36%), increased bicompartmental OA by 2 PP (from 33% to 35%) and tricompartmental by 12 PP (from 17% to 29%). The reverse trend was true for studies with mean age over 65.
There were too few studies for BMI subgroup analysis (8 studies reported mean BMI: 6 obese cohorts, one overweight and one healthy).
Heterogeneity
When no pooling was applied to the selected studies, heterogeneity was very high for all ten outcomes measured (Fig. 3, Fig. 5): I2 ranged from 95% to 98% for the extent of OA (df = 15, Q > 25.0 and P < 0.05), and from 71% to 97% for the location (df = 14, Q > 23.7 and P < 0.05). Heterogeneity remained high for each subgroup investigated, except moderate heterogeneity was calculated for bicompartmental OA in the ‘Other’ imaging modalities subgroup (I2 = 73%).
A sensitivity analysis was performed to see whether the heterogeneity of the meta-analyses could be reduced by removing lower quality studies, however this had little effect on I2 (Supplementary Material). As such the pooled estimates of prevalence rates should be treated with caution.
Risk of bias across studies
For the extent of OA, the point estimate for the relative prevalence of single compartment OA stabilised after the fourth largest, or the sixth highest quality study was added to the cumulative meta-analyses; the addition of the smaller or lower quality studies following this point did not introduce publication selection bias. Similarly, the point estimates for bicompartmental and tricompartmental knee OA stabilised after the addition of the third largest, or sixth highest quality study, again indicating low risk of publication selection bias. For the location of OA, cumulative meta-analyses based on sample size and study quality also provided no evidence to indicate the meta-analysis was subject to publication bias (Supplementary Material).
Discussion
The most important finding from this systematic review and meta-analysis was that both single and bicompartmental patterns of knee OA are more common than tricompartmental disease (Fig. 4). This contrasts greatly with infrequent use of partial and combined PKA compared to TKA. Currently, ∼90% of patients undergo TKA
, however based on the upper bound of the estimated 95% confidence interval, at most only a quarter of people with knee OA have tricompartmental disease; there is a potentially large number of patients being overtreated with TKA. Overuse of TKA may be beneficial for revision rates
Adverse outcomes after total and unicompartmental knee replacement in 101330 matched patients: a study of data from the National Joint Registry for England and Wales.
The location analysis revealed that isolated medial tibiofemoral disease, isolated patellofemoral, and combined medial tibiofemoral and patellofemoral disease patterns are more common than tricompartmental disease, indicating that a higher number of patients may be suitable for medial UKA and patellofemoral arthroplasty than currently undergo these procedures. Whilst patterns involving the lateral tibiofemoral compartment were less common, approximately a third of knees with OA were found to have lateral involvement (Fig. 6), many of whom may be suitable for a lateral UKA alone as patellofemoral arthroplasty would only be added in certain strict indications.
This review only considers OA, rather than the full indications for partial, combined partial or TKA. For example, a deficient anterior cruciate ligament (ACL) is often considered a relative contraindication for PKA
Incidence of bicompartmental osteoarthritis in patients undergoing total and unicompartmental knee arthroplasty: is the time ripe for less radical treatment?.
and the resulting arthroplasty and found that once all indications were considered, the rate of bicompartmental arthroplasty decreased by 17 PP, whilst the rates of single and tricompartmental arthroplasty increased by 9 PP and 8 PP respectively. Moreover, whilst the present study focused on a population with gonarthrosis prompting referral for imaging and specialist opinion, not all the osteoarthritic changes recorded would be severe enough to warrant arthroplasty. This would be the case particularly if the surgical group seeing the patient only practiced total joint replacement, for which the threshold for intervention is rather higher than for PKA. Similarly, treatment may not always match the presentation of the disease–for example, patients with medial facet patellofemoral disease in the presence of medial tibiofemoral arthrosis may not be treated with a patellofemoral arthroplasty since the patella tracks laterally and becomes asymptomatic once the varus deformity has been corrected. There was insufficient data to differentiate between medial and lateral faceted patellofemoral OA in this review. The ratio of medial to lateral tibiofemoral UKA procedures has only been recorded in the National Joint Registry of England, Wales and Northern Ireland since 2018, and so it is not possible to comment on how the UK subgroup analysis results compares to arthroplasties performed. Equivalent data from the US are also not available. Data from the specialist PKA referral practice of the senior author shows up to 40% of tibiofemoral UKAs treat the lateral compartment, either in isolation or in combination with a medial UKA or patellofemoral arthroplasty.
A high level of heterogeneity was calculated for each meta-analysis, and as such the estimates of the pooled prevalence rates reported in the results should be treated with caution. To mitigate any possible impact of the high heterogeneity, we conclude conservatively, based on the upper bound of the 95% CI for tricompartmental disease, rather than the summary effect. Other prevalence-based meta-analyses have similarly reported high values of the I2 statistic
. This is particularly relevant the subgroup analysis of population-based cohort data: only 2 studies were found which were drawn from a population-based cohort, preventing detailed analysis of the results from an etiological perspective. There is a need for more research in this area to understand if population-based studies also show a high prevalence of single compartmental disease which could indicate the role of local etiological factors
. Both the ‘Low’ severity and ‘Other’ imaging methods subgroups yielded an increased prevalence of tricompartmental disease, suggesting that minor structural changes are both more widespread and not well detected by radiography; the high proportion of symptomatic studies in the analysis means that this does not necessarily reflect the natural progression of the disease in the knee. This meta-analysis is more relevant to a surgical perspective, drawing from a large number of symptomatic cohorts. Moreover, the results of the ‘High’ OA severity sub-group analysis categorised to reflect OA changes severe enough to indicate arthroplasty, saw a further increase in the prevalence of single compartmental OA and a relative decrease in tricompartmental damage (Table II). The extended meta-analysis (Supplementary Material) had the advantage of increasing the number of studies in the subgroups–increasing the power of the test and its precision. The mid-point analysis for this extended meta-analysis agreed with the main findings of this paper. However, these studies were not included in the principal meta-analyses due to the inherent uncertainty introduced when full distribution data are not available. The non-linear random effects methodology also inevitably resulted in relative prevalence data that did not sum to 100%: for the extent of OA, the meta-analysis summed to 89.3% and for location to 83.1%. The re-normalised data, which summed to 100%, were within the 95% CI of the raw meta-analysis output and thus had no impact on the conclusions made.
This is the first systematic review and meta-analysis to look at the prevalence of uni-, bi- and tricompartmental OA, and at the compartmental distribution of OA, given that a knee has some OA. Hart et al. compared radiographic and MRI defined prevalence of patellofemoral OA with a meta-analysis and found that 43–57% of those with any radiographic tibiofemoral OA also had patellofemoral OA
; this is in agreement with our data where we identified that 25–56% of those with any tibiofemoral OA would have concurrent patellofemoral OA. Kobayashi et al.’s systematic review and meta-analysis examined the prevalence of patellofemoral OA and they found from population-based cohorts, 40% of people with any patellofemoral OA had the disease confined to the isolated patellofemoral compartment alone
. This is again comparable with the results of our study, which found that 29% of people would have isolated patellofemoral disease, given they have any patellofemoral involvement. Kobayashi's conclusion that patellofemoral OA was a common form of OA worthy of increased research attention is supported by our meta-analysis. However there remains a disconnect between pathology and surgical intervention here; whilst tibiofemoral disease severity matches symptoms quite well, in the patellofemoral compartment that relationship is less well defined
. We found that the medial tibiofemoral compartment was most likely to be involved in knees with OA (72%). The external knee adduction moment has been associated with increased loading in the medial tibiofemoral compartment
At least three-quarters of people with knee OA do not have tricompartmental disease. This contrasts greatly with the proportion of patients being treated with TKA. It is possible that for many patients, TKA is sacrificing healthy cartilage, ligament and bone. Research into treatments for uni- and bicompartmental osteoarthritis may help target treatment to the specific disease patterns seen in this large patient group.
Contributions
JC Stoddart: Conception and design of the study, collection, analysis and interpretation of the data, drafting the article, critical revision of the article for important intellectual content, final approval of the article.
O Dandridge: Collection, analysis and interpretation of the data, critical revision of the article for important intellectual content, final approval of the article.
A Garner: Conception and design of the study, interpretation of data, critical revision of the article for important intellectual content, final approval of the article.
J Cobb: Conception and design of the study, interpretation of the data, critical revision of the article for important intellectual content, final approval of the article.
RJ van Arkel: Conception and design of the article, collection and interpretation of data, drafting of the article, critical revision of the article for important intellectual content, final approval of the article.
Conflict of interest
O Dandridge is supported by The Sackler Trust. A Garner reports support from the Dunhill Medical Trust, with institutional support from Zimmer Biomet and Smith & Nephew. J Cobb has reported board membership of Orthonika, consultancy work for Zimmer Biomet, grants from Smith & Nephew and holds stock in Embody-Orthopaedic. R J van Arkel has reported grants from DePuy Synthes and other non-financial support from Zimmer Biomet. No other relationships present of potential conflict of interest.
Role of the funding source
Research was funded by the Peter Stormonth Darling Charitable Trust, who had no role in this work.
Acknowledgements
There are no other acknowledgements that contributed to this manuscript to disclose.
Appendix A. Supplementary data
The following are the Supplementary data to this article:
Adverse outcomes after total and unicompartmental knee replacement in 101330 matched patients: a study of data from the National Joint Registry for England and Wales.
Unicompartmental knee arthroplasty vs total knee arthroplasty for medial compartment arthritis in patients older than 75 years: comparable reoperation, revision, and complication rates.
Is unicompartmental knee arthroplasty (UKA) superior to total knee arthroplasty (TKA)? A systematic review and meta-analysis of randomized controlled trial.
Larger range of motion and increased return to activity, but higher revision rates following unicompartmental versus total knee arthroplasty in patients under 65: a systematic review.
Incidence of bicompartmental osteoarthritis in patients undergoing total and unicompartmental knee arthroplasty: is the time ripe for less radical treatment?.
Gender difference in symptomatic radiographic knee osteoarthritis in the Knee Clinical Assessment - CAS(K): a prospective study in the general population.
High prevalence of knee osteoarthritis, pain, and functional limitations in female soccer players twelve years after anterior cruciate ligament injury.
31197-3&id=gr1.jpg){kind=link}
31197-3&id=gr2.jpg){kind=link}
31197-3&id=gr3.jpg){kind=link}
31197-3&id=gr4.jpg){kind=link}
31197-3&id=gr5.jpg){kind=link}
31197-3&id=gr6.jpg){kind=link}