Effectiveness of exercise therapy in patients with knee osteoarthritis: an overview of systematic reviews

Abstract

Objective

This study aimed to assess the methodological quality of published systematic reviews of exercise therapy in knee osteoarthritis and summarise their reported effectiveness on quality of life, knee joint function, or adverse events.

Design

Overview of systematic reviews.

Data sources

PubMed, Embase, CINAHL, Web of Science and CENTRAL (searched on 14 April 2025), plus grey literature (PROSPERO, Epistemonikos, OpenGrey).

Eligibility criteria for selecting studies

We included systematic reviews of randomised controlled trials in patients diagnosed with knee osteoarthritis by imaging or clinical criteria and treated conservatively with exercise therapy; we excluded reviews that enrolled patients scheduled for surgery, with acute inflammation or osteoarthritis of other joints (hand, hip, ankle), for which relevant author data could not be obtained after one contact attempt, or that did not report at least one primary outcome (quality of life, knee joint function or adverse events).

Data extraction and synthesis

Two reviewers independently extracted data on study characteristics, interventions and outcomes, and assessed methodological quality using the AMSTAR 2 (A MeaSurement Tool to Assess systematic Reviews 2) tool. Due to heterogeneity in outcome measures across systematic reviews, meta-analysis was not conducted. Effectiveness was defined as any reported beneficial outcome of exercise therapy on predefined outcomes, including quality of life, physical function, pain or adverse events.

Results

58 systematic reviews were selected. Muscle-strengthening (74.1%) and aerobic (48.2%) exercises were the most commonly prescribed exercise-based interventions. SF-36 (36-Item Short Form Health Survey) and the WOMAC (Western Ontario and McMaster Universities Osteoarthritis Index) were the most popular outcome-evaluation tools. Furthermore, 63.7% of the systematic reviews revealed that exercise therapy improved all outcomes. The number of intervention-related adverse events was small. Notably, almost all systematic reviews (87.4%) had a critically low quality.

Conclusions

Current evidence on exercise therapy for knee osteoarthritis is inadequate. Nevertheless, exercise therapy can be considered for conservative treatment of knee osteoarthritis. Future studies should use network meta-analyses to compare the effects of different exercise therapies and determine their superiority over other conservative therapies.

STRENGTHS AND LIMITATIONS OF THIS STUDY.

• This umbrella review followed the Joanna Briggs Institute methodology and was prospectively registered.

• A comprehensive and updated literature search was conducted across multiple databases without language restrictions.

• Study selection, data extraction and quality assessment were independently conducted by two reviewers, with a third reviewer resolving discrepancies.

• The quality of included systematic reviews was rigorously assessed using the AMSTAR 2 (A MeaSurement Tool to Assess systematic Reviews 2) tool.

• Meta-analysis could not be performed due to considerable heterogeneity in outcome definitions and measurement tools across reviews.

Introduction

Osteoarthritis (OA) involves the degeneration of articular components (bone, cartilage and synovium)¹ and has increased disability-adjusted life years and healthcare costs, rendering it a major issue with the ageing global population.² OA commonly affects the knee joint³; the reported lifetime risk for symptomatic knee OA (KOA) is 14–45%.⁴ KOA is characterised by pain, functional impairment and a diminished quality of life (QoL).^{3 5}

Conservative therapies, such as exercise and diet, constitute the first-line treatments for OA.^{1 3 6} Several guidelines have highlighted the effectiveness of physiotherapist-administered exercise therapy for KOA.⁷,¹⁰ Researchers have extensively used systematic reviews (SRs) to report the effectiveness of exercises for KOA.¹¹,¹⁴ However, these SRs are limited by their high heterogeneity. For example, some SRs included randomised controlled trials (RCTs) investigating a wide variety of exercise therapies. Multiple SRs have been published recently; however, their lack of quality has been highlighted.¹⁵,¹⁷ Poor-quality SRs can mislead clinicians and impact their decision-making ability.

A comprehensive review of SRs and meta-analyses (MAs) examined the impact of physical activity on OA; however, it did not analyse KOA, hip OA or the effects of conservative interventions.¹⁸ Another comprehensive review of SRs and MAs examined the effectiveness of non-pharmacological and non-surgical interventions for KOA¹⁹; however, the broad spectrum of interventions analysed introduced heterogeneity.

Exercise therapy for KOA has not been investigated in detail. With a marked increase in evidence, comprehensive summaries of SRs and MAs focusing on available exercise therapies and their impact on the QoL would aid clinicians in their decision-making. Therefore, this overview aimed to systematically evaluate the quality and summarise the findings of SRs that investigated the effects of exercise therapy in individuals with KOA. Specifically, we aimed: (1) to assess the methodological quality of included SRs, and (2) to summarise the reported effectiveness of exercise interventions across key clinical outcomes such as QoL, knee joint function, pain and adverse events.

Materials and methods

Protocol

The protocol of this overview of SRs adhered to the Joanna Briggs Institute guidelines²⁰ and was prospectively registered in the Open Science Framework (https://osf.io/tjxwy/). We included SRs and MAs of RCTs investigating the QoL, knee joint function or adverse events following conservative treatment for KOA. We compared outcomes between patients receiving exercise therapy and those receiving standard care or no intervention. This review complies with the ‘Preferred Reporting Items for Overviews of Reviews’ guidelines.²¹

Study selection

The inclusion criterion was analyses involving patients diagnosed with KOA based on imaging or clinical findings and treated with conservative therapy. SRs were not required to include QoL as an outcome for eligibility, as our aim was to comprehensively assess the effectiveness of exercise therapy across a range of clinically relevant outcomes including QoL, knee function, pain and adverse events. Inclusion was limited to SRs that reported at least one of the predefined primary outcomes: QoL, knee joint function or adverse events. In cases where these primary outcomes were not explicitly stated, SRs were included if they reported secondary outcomes such as pain or physical function in sufficient detail and were judged to be clinically related to the review’s objectives, given their relevance to functional and QoL outcomes in KOA. This ensured consistency with the study’s dual focus on both effectiveness and methodological quality. The exclusion criteria were as follows: (1) analyses including patients studied immediately before or while awaiting surgery; (2) analyses of quasi-RCTs, non-RCTs or observational studies; (3) analyses including patients with acute inflammation; (4) failure to obtain data from relevant authors after one attempt; (5) analyses including patients with OA of the hand, hip, ankle or other joints (reports were included if the analysis considered joints other than the knee, but the KOA results were presented separately); (6) studies with outcomes other than the QoL, knee joint function or adverse events; and (7) analyses of abstracts, commentaries, methodological studies, reviews, narrative reviews, conference abstracts, guidelines and network MAs (NMAs).

Primary outcomes included the QoL (assessed using the 36-Item Short Form Health Survey (SF-36) or the European Quality of Life Five Dimension), knee joint function (assessed using the Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC) or the Knee injury and Osteoarthritis Outcome Score (KOOS)) and adverse events (exacerbation of pain or falls). Adverse events were defined broadly and included any negative health outcomes potentially related to the exercise intervention, such as exacerbation of joint pain, general body pain, dizziness, fatigue, muscle soreness, falls or other clinical symptoms reported in the included studies and consistent with American College of Rheumatology (ACR)-defined manifestations of OA. These three outcomes were selected based on their clinical relevance and alignment with core domains commonly recommended in KOA trials and guidelines. They reflect patient-centred (QoL), functional (joint function) and safety (adverse events) outcomes, which are crucial in evaluating the comprehensive impact of exercise interventions. Secondary outcomes included pain and physical function (assessed using the Timed Up-and-Go Test or the 6-minute walk test).

Data sources and searches

PubMed, Embase, the Cumulative Index to Nursing and Allied Health Literature database, Web of Science and the Cochrane Central Register of Controlled Trials were searched for articles published until 5 April 2022. PROSPERO, Epistemonikos and the OpenGrey platform were searched for grey literature. Database searches were updated as of 14 April 2025. Medical Subject Headings were screened to identify keywords associated with KOA, exercise, SR and MA (online supplemental appendix S1). No language restrictions were imposed. We contacted the authors of the included SRs and MAs to obtain additional information, whenever necessary.

Two authors initially screened and selected the eligible studies by independently reviewing titles and abstracts after removing any duplicate records using Rayyan.²² In the case of disagreements, a third author made the final decision. The selected articles were screened a second time, and the same two authors independently reviewed their full texts. Disagreements were resolved by consensus or by the third author’s decision.

Data extraction and quality assessment

Two authors collected the following data: (1) SR characteristics, including the authors’ countries, participants’ characteristics (including age range and mean body mass index), and reviewed articles; (2) characteristics of the interventions and research in the reviewed articles (treatments in the intervention/comparison groups, funding information and main findings of each review); and (3) outcomes of the included studies (including the QoL, total knee function, pain, physical function outcomes and adverse events). The quality of the included SRs and MAs was assessed independently by two authors using ‘A MeaSurement Tool to Assess systematic Reviews 2’ (AMSTAR 2).²³ Conflicts between the two authors were resolved by consensus or the third author’s decision.

Data synthesis and analysis

An MA could not be performed owing to variations in the outcomes and measurement methods among the SRs identified in our preliminary search. In this review, ‘effectiveness’ was defined as any reported beneficial outcome of exercise therapy on at least one of the predefined outcomes (QoL, physical function, pain or adverse events), based on the conclusions or summary findings presented in each SR. We focused on the participants, interventions and outcome measures when assessing the clinical heterogeneity of the included reports.

Patient involvement

Patients were not directly involved in formulating the research question, choosing the outcome measures, designing or conducting the study, or interpreting the findings.

Results

Search results

We initially identified 3498 articles. After removing duplicates, the titles and abstracts of the remaining 2057 articles were screened. Overall, 58 SRs met the eligibility criteria (figure 1).¹⁴²⁴,⁸⁰ Among the 58 included SRs, 15 assessed QoL outcomes, 28 addressed knee joint function and 20 reported on adverse events. Many reviews evaluated more than one of these outcomes concurrently. Online supplemental table S1 presents a list of the excluded studies.

Figure 1. Flow diagram of study selection. CINAHL, Cumulative Index to Nursing and Allied Health Literature.

Study characteristics

The included SRs were conducted between 2005 and 2025 in Africa (Ethiopia),⁶² the Americas (Canada,²⁹ the USA^{41 59} and Brazil^{27 33 36 50 52 60 75}), Asia (China,^{2931 35 42},^{48 63} India⁵⁸ and Japan⁵³,⁵⁷), Europe (Cyprus,⁴⁹ Denmark,³⁸ France,³⁰ the Netherlands,³⁷ Portugal³² and the UK⁵¹), the Middle East (Saudi Arabia^{24 66} and Turkey^{26 39}) and Oceania (Australia^{14 25 28 34 40 72}). Online supplemental figure S1 presents a list of the included articles according to the publication year.

The most common participant selection criterion was KOA diagnosed according to the ACR criteria (n=19 (32.8%)),2427 33 35 37 38 41 42 44 45 63 64 66,68 71 73 74 77 followed by KOA diagnosed radiographically^{14 27 31 32 34 45 63 66 75} and KOA diagnosed using the Kellgren-Lawrence classification.²⁶²⁸,30 38 46 47 60 72 Age restrictions were commonly imposed (n=9 (15.5(%)).^{27 28 35 40 48 62 64 68 69} The median number of RCTs per SR was 9 (range: 1–139). The median number of patients in the RCTs was 841.5 (range: 52–12 468). The ranges for participant age and body mass index were 35–89 years and 21.5–38.1 kg/m², respectively; however, approximately half of the SRs did not provide body mass index data (online supplemental table S2)

Characteristics of the interventions/comparisons in the included SRs

The intervention groups received exercise-based therapies, such as muscle-strengthening training (n=43 (74.1%)),¹⁴²⁴,^{28 30 32} aerobic exercises (n=28 (48.3%))²⁵,^{2831 34} and mind-body exercises (n=23 (40.0%)).2527 29 34 35 37 44 51,54 56 57 60 63 64 66 67 71 73 74 76 78 The control groups received patient education (n=29 (50.0%)),²⁵,^{2729 30 35} no intervention (n=28 (48.3%)),^{2527 31},34 36 37 42 44 45 48 50 usual care (n=14 (24.1%))^{2526 29 34},36 45 48 49 52 61 62 75 79 or attention control (n=9 (15.5%))^{29 40 42 53 64 65 68 72 74} or the members were placed on waiting lists (n=9 (15.5%)).^{25 35 36 41 53 63 67 68 73} Of the 58 SRs included, 27 (46.6%) were funded by public or non-profit organisations.2425 27 29 31 34 36 38 40 43,47 51 52 55 60 61 69 71 Conversely, 12 (20.7%) did not detail any funding.^{26 32 39 50 53 54 57 59 62 70 79 80} Slightly more than half of the SRs (n=37 (63.8%)) found exercise therapy to be effective for each outcome (online supplemental table S3).²⁴²⁶,^{29 32}

Outcomes of the included SRs

Overall, 22 reviews (37.9%) examined the effectiveness of exercise therapy for QoL improvement in patients with KOA. SF-36 (n=15 (25.9%)),²⁷,2935 40 48 50 53 58 61 64 68 73 76 77 KOOS (QoL; n=10 (17.2%)),^{14 27 31 45 48 50 52 61 75 79} and Assessment of Quality of Life (n=3 (5.2%))^{14 48 77} were used as the assessment tools. 31 reviews (53.4%) examined the effects of exercise therapy on total knee joint function. The evaluation tools used in the SRs included the WOMAC (n=28 (48.3%)),1426 28 30 34 41 43 47,49 56 58 61 Arthritis Impact Measurement Scales (n=7 (12.1%)),^{30 37 40 48 56 61 77} KOOS (n=9 (15.5%))^{28 47 48 61 69 74 75 79 80} and Lequesne index (n=4 (6.9%)).^{30 37 40 46} 13 reviews (22.4%) provided information on adverse events.^{28 31 35 40 41 45 58 61 63 69 72 76 79} The most commonly reported adverse events were knee pain (n=10 (17.2%)),^{31 40 41 45 58 61 63 69 72 79} joint pain (n=5 (8.6%))^{35 40 41 58 63} and falls (n=5 (8.6%))^{28 41 63 72 79} (online supplemental table S4).

50 reviews (86.2%) examined the effectiveness of exercise therapy for pain, evaluated using the WOMAC pain scale (n=37 (63.8%)),^{2930 32},^{42 44 47} Visual Analogue Scale (n=35 (60.3%))^{1426 28},^{32 34 36} and KOOS pain scale (n=12 (20.7%)).1431 37 38 42 45 47 48 50,^{52 79} 50 reviews (86.2%) examined the effects of exercise therapy on physical function, evaluated using the WOMAC physical function scale (n=30 (51.7%)),³⁰³²,^{35 37} the 6-minute walk test (n=18 (31.0%))^{1425 26 29},31 35 40 43 47 48 50 and muscle strength measurements (n=16 (27.6%))^{24 25 29 30 33 39 40 43 48 51 52 54 58 64 69 70} (online supplemental table S5).

Quality assessment

Figure 2 describes the quality of the 58 SRs, and online supplemental table S6 details the AMSTAR 2 findings for each SR. In most included reviews, the authors used clearly defined ‘patient/population, intervention, comparison and outcome’ elements for the selection criteria (AMSTAR 2 item 1), clarified the study design(s) to be included in the review (item 3), carefully implemented study selection by two or more authors (item 5), appropriately assessed the bias risk (item 9), considered the bias risk when interpreting and discussing the results (item 13) and disclosed conflicts of interest (including funding sources; item 16). However, most included reviews involved insufficient work implementation or presented methodological concerns related to protocol registration (item 2), systematic literature search (item 4), presentation of the list of excluded studies (item 7), detailing of the included studies (item 8), reporting of funding sources of the included studies (item 10) and consideration of publication bias (item 15). Overall, approximately 87.9% of the included SRs (n=51) were rated as having a ‘critically low’ quality.

Figure 2. Quality of the included systematic reviews according to the AMSTAR 2 items. AMSTAR 2, A MeaSurement Tool to Assess systematic Reviews 2; N/A, not applicable.

Discussion

In this review, we identified 58 SRs and MAs on exercise therapy for KOA and evaluated their quality to investigate the effectiveness of exercise therapy and the robustness of the existing evidence. The ACR diagnostic criteria were used to define KOA in many SRs (32.7%). Muscle-strengthening (74.1%) and aerobic (48.2%) exercises were the most popularly analysed exercise-based therapies. SF-36 and the WOMAC were the most commonly used assessment tools for the QoL (25.8%) and knee joint function (48.2%), respectively. Adverse events were generally infrequent; however, the most frequent adverse event was knee joint pain (17.2%). Most SRs (87.9%) had a critically low quality.

A similar 2018 umbrella review included 41 SRs¹⁹; however, this review used broad classifications of interventions (such as non-pharmacological and non-surgical interventions) as the eligibility criteria, rather than focusing on exercise therapy. Furthermore, it included a small number of studies, although its findings support our conclusion that exercise therapy is effective for KOA. A strength of the present review is that the greater quantity of recent evidence allowed us to include 58 SRs after exhaustive searching and screening while focusing on exercise therapy as the intervention.

The earliest publication year for the SRs analysed in our review was 2005; however, most SRs were published after 2015. While we listed SRs by first author name to facilitate traceability and citation consistency, the publication year was also considered in the synthesis, as it reflects the growing research interest in exercise therapy for KOA in recent years. The poor quality of these SRs has been highlighted⁸¹ and may be attributed to their protocols being unregistered or unpublished in registries. High-quality SRs must be designed when developing and registering protocols. The ‘Preferred Reporting Items of Systematic Reviews and Meta-Analyses’ (PRISMA) 2020 guideline,⁸² a validated, updated and increasingly popular reporting guideline, could improve the methodological adequacy and transparency of SRs. Authors should follow the PRISMA checklist and refer to available tools for assessing the SR quality (AMSTAR 2) when developing protocols for higher-quality SRs.²³ AMSTAR 2 defines items 2, 4, 7, 9, 11, 13 and 15 as the critical domains. Most SRs included in this review were judged to be of a ‘critically low’ quality because several critical domains of AMSTAR 2 were poorly addressed. Authors should refer to AMSTAR 2 and carefully address these domains, particularly those related to review practices, to improve the quality of future SRs.

We did not perform an MA owing to the high predicted degree of data heterogeneity following the preliminary search. The selection criteria for patients with KOA varied across all SRs. Moreover, variations in the Kellgren-Lawrence classification grades and age-related limitations may have contributed to greater inconsistencies across the results of all SRs. Subgroup or meta-regression analyses can be useful for examining and adjusting for the potential effect modifiers identified in this review (such as Kellgren-Lawrence classification severity and age) and the impact of body mass index on exercise therapy outcomes.⁸³

Various exercise therapies were prescribed in the intervention groups, causing difficulty in assessing the superiority or inferiority of these interventions using data synthesis methods (such as MAs). Recently, to assess the effectiveness of multiple treatment interventions, a trend towards performing NMAs instead of pairwise comparisons has emerged.⁸⁴ NMAs allow both direct comparisons of treatments to estimate effects as well as indirect statistical comparisons of the impact of interventions that have not been compared pairwise.⁸⁵ Only two or three groups can be compared in typical clinical trials (such as RCTs), whereas NMAs can estimate the efficacy and priority ranking of each conservative therapy for KOA.⁸⁶ Future studies should use NMAs to compare the usefulness of different exercise therapies, including other conservative therapies.

Control-group treatments were carefully designed to avoid aggressive exercise therapies (such as those administered to the intervention groups). In some studies, patients were placed on waiting lists or assigned to attention-control training. However, the pharmacotherapy types and doses were insufficiently detailed in each SR and were heterogeneous across all the SRs. Drugs considerably affect OA outcomes (such as pain¹); therefore, the drug therapy criteria should be closely controlled in future SRs.

Approximately one-third of the SRs examined the QoL as a primary outcome. SF-36 scores are reportedly associated with the clinical status and functional capacity in OA and can be used as sensitive health status indicators in clinical practice.⁸⁷ Some SRs involved the KOOS QoL and Assessment of Quality of Life tools. The use of SF-36 in future MAs may aid in examining the quantitative integration of results. The number of adverse events was generally small; however, some SRs detailed adverse events from individual RCTs. Precautions should be taken to avoid knee pain, other joint pain and falls when prescribing exercise therapy for KOA. In terms of effectiveness, muscle-strengthening and aerobic exercises were associated with favourable outcomes across pain, physical function and QoL domains. However, a key limitation of the included SRs was the lack of consistent reporting on exercise parameters such as session duration, weekly frequency, number of repetitions, rest periods and intensity. This hindered our ability to determine optimal exercise prescriptions for each outcome. These findings support current clinical guidelines that recommend exercise—particularly muscle-strengthening and aerobic modalities—as core components of non-pharmacological treatment for KOA.

The WOMAC was the most popular assessment tool for the overall knee joint function and has sufficient reliability and validity for research use.⁸⁸ The WOMAC requires only 5–10 min for assessment, can simultaneously assess pain and physical function by separately considering the relevant subscales, and has been translated to and validated in many languages. The Arthritis Impact Measurement Scales and KOOS were also used. However, the use of the WOMAC to assess the knee joint function, pain and physical function is preferable to that of other scales to allow easier comparison of the effectiveness of interventions, thereby reducing patient and evaluator burden. The Visual Analogue Scale and KOOS pain scale were also used for pain assessment; however, these tools do not differ considerably. Furthermore, additional detailed physical parameters that cannot be assessed using the WOMAC, such as the 6-minute walk test distance, muscle strength and range of motion, were measured to evaluate physical function. The aforementioned measures can simultaneously assess exercise tolerance, muscle performance and knee joint mobility; therefore, using the WOMAC in different situations would be desirable. Most studies support the effectiveness of exercise therapy for improving primary outcomes; however, quantitative integration has not been achieved, and additional MAs are necessary to determine its effectiveness. Exercise therapy is recommended for conservative treatment of KOA, considering the low number of associated adverse events. Moreover, to improve the quality, transparency and reproducibility of exercise-based interventions, future SRs and primary studies should consider using standardised reporting tools such as the Preferred Reporting Items for Resistance Exercise Studies (PRIRES), the Consensus on Exercise Reporting Template (CERT) and the international Consensus on Therapeutic Exercise aNd Training (i-CONTENT) tool. These frameworks provide structured guidance for evaluating exercise intervention content, fidelity and reporting quality, and their broader application could enhance the rigour and comparability of future evidence syntheses.⁸⁹,⁹¹

None of the included SRs were sponsored by for-profit companies; therefore, the potential for industry bias was considered low. Sponsorship by manufacturers may result in reports that the intervention is more effective when using the sponsor’s product.⁹² Careful examination of the RCTs included in the SRs that investigated the effectiveness of whole-body vibration training revealed that medical device companies often supported individual studies. Therefore, the possibility of sponsorship bias should be considered when investigating certain exercise therapies, such as whole-body vibration training, since corporate interests can divert the research goals from the issues most relevant to public health.⁹³

This study has some limitations. First, we were unable to integrate the data using an MA. Researchers should perform an umbrella review wherein data are preferably extracted from each RCT and conduct an MA accordingly.²¹ An MA was not planned from the outset because we anticipated high data heterogeneity owing to the review’s broad scope. We recommend the following criteria for future similar umbrella reviews: (1) diagnostic criteria based on the ACR criteria, (2) inclusion criteria limited to Kellgren-Lawrence grades 1–3, (3) exercise therapy limited to strength training or aerobic exercise, (4) control-group interventions defined explicitly in advance and (5) inclusion of more detailed interventions and follow-up durations. Quantitative conclusions can be deduced from carefully prepared MAs. Second, publication bias was not adequately assessed. More than 10 studies should be visually examined using funnel plots to determine publication bias.⁹⁴ We included more than 10 studies with outcomes of QoL, total knee function, pain and physical function; however, no funnel plots were used to examine publication bias. Future umbrella reviews should evaluate publication bias using funnel plots, considering heterogeneity and other factors. Third, the certainty of evidence could not be assessed. Recent reporting guidelines recommend that this be mentioned in overviews of reviews²¹; therefore, future comprehensive reviews should include this assessment. Fourth, many included SRs lacked detailed reporting of exercise intervention characteristics, such as frequency, duration and total volume. This limitation prevented us from summarising such data in a standardised tabular format. Consistent and transparent reporting of these variables is essential for clinical translation and should be encouraged in future SRs and primary trials using tools such as CERT or PRIRES. Fifth, we were unable to systematically identify overlap in primary studies among the included SRs due to inconsistent reporting. Future overviews should consider using citation matrices or tools like the Corrected Covered Area to quantify overlap.

In conclusion, exercise therapy should be considered for conservative treatment of KOA. Future studies should use NMAs to compare the effects of different exercise therapies and determine their superiority over other conservative therapies. Quantitative data integration can be achieved using MAs with strict criteria.

Data availability statement

Data are available upon reasonable request.