Characteristics for Describing Patients With Knee OA
Characteristics for Describing Patients With Knee OA
We assembled an international panel of clinical experts on osteoarthritis to generate a list of baseline characteristics for describing patients included in clinical trials with knee osteoarthritis. A preliminary list was prepared and sent to these experts with the request that they add patient characteristics that they considered relevant. In a second round, the experts were asked to rate the relevance of each patient characteristic.
We searched Medline and EMBASE to identify clinical experts on osteoarthritis of the knee. The following MESH terms were used: Osteoarthritis, Knee, Physical Therapy Modalities, Steroids, Viscosupplementation, Anti-Inflammatory Agents, Non-Steroidal, Randomized controlled trial (a detailed list of the search strategy is available upon request from the corresponding author). The search was restricted to articles published between the years 2007 and 2012. We included only studies that evaluated the treatment effect of steroid injections, viscosupplementation, non-steroidal analgesics, or physical therapy. The aim, set arbitrarily, was to identify 20 experts that would participate in our survey. From the list of all authors, we selected those that co-authored three or more trials plus a random sample of authors that were listed on two publications. Based on the medical specialty and/or affiliation mentioned in the publication, we categorized authors into groups of clinically-oriented (e.g., rheumatologists, physiotherapists) or methodology-oriented (e.g., clinical epidemiologists, biostatisticians) researchers. Only authors categorized as clinically-oriented researchers were contacted for participation in the survey.
The selected experts were contacted by E-mail and informed about the aim of the study. Those that agreed to participate received a prepared form and a request to add characteristics to complete a preliminary list of baseline characteristics (indicated by § in Additional file 1).
The experts received the following information: Patients with knee osteoarthritis have been included in four randomized trials (A/B/C/D, Additional file 2). In each of the four trials, one group of patients received an active treatment X (non-surgical) and the other group a placebo. In trials A and B, the outcome of interest was pain; in trials C and D, the outcome was a functional parameter. The results between trials differed significantly. In trials A and C, treatment X showed a significant benefit, and in trials B and D, the identical treatment X showed no benefit. The execution of the trials (intervention, treatment, measurement of outcomes, randomization etc.) was identical; however, one reason for the contradictory results may have been the inclusion of different patient populations in all four trials. Then, the experts were asked "From your experience as a clinical expert, what baseline patient characteristics would be necessary to identify patients to whom the trial results might be applicable, and furthermore, to evaluate the comparability of the two populations in trials A/B and C/D? The participants were asked to add characteristics that they thought should be included in an attached list of baseline characteristics. Participants returned the completed lists by mail or fax.
Based on the answers from the first round of inquiries, we updated the list of patient characteristics. To avoid redundancy, nearly identical characteristics were merged into one parameter. The final list included 121 items that were assigned to one of six categories, including general information about the patient (e.g., age, gender), psychosocial factors (e.g., depression, anxiety), history (e.g., duration of pain, pain provoking maneuvers), physical examination results (e.g., periarticular tenderness, instability), laboratory tests (e.g., C-reactive protein, serum hyaluronic acid concentration), and imaging results (e.g., Kellgren-Lawrence grading, bone marrow lesion).
This updated list was sent to the experts that had returned the questionnaire from the first round. The participants were asked to rate the 'importance' of each baseline characteristic on a scale of 0 to 10, where 0 indicated no importance, and 10 indicated the utmost importance. We informed participants that the 'degree of importance' was related to two issues; first, the relevance of the characteristic in identifying patients in daily practice to whom the results of a study were applicable; and, second, its usefulness in meta-analyses for assessing the comparability of patient populations from different primary trials for a potential pooling of results.
We reasoned that the experts would probably estimate the importance of baseline characteristics differently, depending on the outcome of interest. Therefore, we asked the participants to rate each baseline characteristic according to three different outcome categories; pain (e.g., VAS), function (e.g., WOMAC-function sub-score), and structure (e.g., change of the joint space width over time).
Participants that completed both questionnaires were asked to provide information about their age, gender, and primary medical specialty. Completed questionnaires were returned by E-mail or fax. All experts that responded in the first round were informed that they would receive a voucher for $100 after returning the completed questionnaire for round 2.
For this study, the medians and interquartile ranges for each parameter were calculated to quantify the importance assigned to single items. The median is a measure for the average and the 25–75% interquartile range (IQR) and the range are measures of dispersion of values. The median value means that half of the values are below and half the values above the median value. The 25–75% IQR is the difference between the values of the 25th and 75th percentiles. The 0th and 100th percentiles (minimal and maximal values) define the range. The final list only included baseline characteristics with a median rating ≥7 (on a 0 to 10 point scale). We arbitrarily defined a consensus among experts as a rating with an interquartile range ≤4 (±2) points.
This work did not involve human subjects or animals. Thus, according to national laws and institutional regulations, review board (IRB) approval was not necessary. All experts acknowledged in the manuscript gave their permission to list their names.
Methods
We assembled an international panel of clinical experts on osteoarthritis to generate a list of baseline characteristics for describing patients included in clinical trials with knee osteoarthritis. A preliminary list was prepared and sent to these experts with the request that they add patient characteristics that they considered relevant. In a second round, the experts were asked to rate the relevance of each patient characteristic.
Selection and Recruitment of Experts
We searched Medline and EMBASE to identify clinical experts on osteoarthritis of the knee. The following MESH terms were used: Osteoarthritis, Knee, Physical Therapy Modalities, Steroids, Viscosupplementation, Anti-Inflammatory Agents, Non-Steroidal, Randomized controlled trial (a detailed list of the search strategy is available upon request from the corresponding author). The search was restricted to articles published between the years 2007 and 2012. We included only studies that evaluated the treatment effect of steroid injections, viscosupplementation, non-steroidal analgesics, or physical therapy. The aim, set arbitrarily, was to identify 20 experts that would participate in our survey. From the list of all authors, we selected those that co-authored three or more trials plus a random sample of authors that were listed on two publications. Based on the medical specialty and/or affiliation mentioned in the publication, we categorized authors into groups of clinically-oriented (e.g., rheumatologists, physiotherapists) or methodology-oriented (e.g., clinical epidemiologists, biostatisticians) researchers. Only authors categorized as clinically-oriented researchers were contacted for participation in the survey.
First Round
The selected experts were contacted by E-mail and informed about the aim of the study. Those that agreed to participate received a prepared form and a request to add characteristics to complete a preliminary list of baseline characteristics (indicated by § in Additional file 1).
The experts received the following information: Patients with knee osteoarthritis have been included in four randomized trials (A/B/C/D, Additional file 2). In each of the four trials, one group of patients received an active treatment X (non-surgical) and the other group a placebo. In trials A and B, the outcome of interest was pain; in trials C and D, the outcome was a functional parameter. The results between trials differed significantly. In trials A and C, treatment X showed a significant benefit, and in trials B and D, the identical treatment X showed no benefit. The execution of the trials (intervention, treatment, measurement of outcomes, randomization etc.) was identical; however, one reason for the contradictory results may have been the inclusion of different patient populations in all four trials. Then, the experts were asked "From your experience as a clinical expert, what baseline patient characteristics would be necessary to identify patients to whom the trial results might be applicable, and furthermore, to evaluate the comparability of the two populations in trials A/B and C/D? The participants were asked to add characteristics that they thought should be included in an attached list of baseline characteristics. Participants returned the completed lists by mail or fax.
Second Round
Based on the answers from the first round of inquiries, we updated the list of patient characteristics. To avoid redundancy, nearly identical characteristics were merged into one parameter. The final list included 121 items that were assigned to one of six categories, including general information about the patient (e.g., age, gender), psychosocial factors (e.g., depression, anxiety), history (e.g., duration of pain, pain provoking maneuvers), physical examination results (e.g., periarticular tenderness, instability), laboratory tests (e.g., C-reactive protein, serum hyaluronic acid concentration), and imaging results (e.g., Kellgren-Lawrence grading, bone marrow lesion).
This updated list was sent to the experts that had returned the questionnaire from the first round. The participants were asked to rate the 'importance' of each baseline characteristic on a scale of 0 to 10, where 0 indicated no importance, and 10 indicated the utmost importance. We informed participants that the 'degree of importance' was related to two issues; first, the relevance of the characteristic in identifying patients in daily practice to whom the results of a study were applicable; and, second, its usefulness in meta-analyses for assessing the comparability of patient populations from different primary trials for a potential pooling of results.
We reasoned that the experts would probably estimate the importance of baseline characteristics differently, depending on the outcome of interest. Therefore, we asked the participants to rate each baseline characteristic according to three different outcome categories; pain (e.g., VAS), function (e.g., WOMAC-function sub-score), and structure (e.g., change of the joint space width over time).
Participants that completed both questionnaires were asked to provide information about their age, gender, and primary medical specialty. Completed questionnaires were returned by E-mail or fax. All experts that responded in the first round were informed that they would receive a voucher for $100 after returning the completed questionnaire for round 2.
Statistical Analysis
For this study, the medians and interquartile ranges for each parameter were calculated to quantify the importance assigned to single items. The median is a measure for the average and the 25–75% interquartile range (IQR) and the range are measures of dispersion of values. The median value means that half of the values are below and half the values above the median value. The 25–75% IQR is the difference between the values of the 25th and 75th percentiles. The 0th and 100th percentiles (minimal and maximal values) define the range. The final list only included baseline characteristics with a median rating ≥7 (on a 0 to 10 point scale). We arbitrarily defined a consensus among experts as a rating with an interquartile range ≤4 (±2) points.
Ethical Approval
This work did not involve human subjects or animals. Thus, according to national laws and institutional regulations, review board (IRB) approval was not necessary. All experts acknowledged in the manuscript gave their permission to list their names.
