Clinical prediction rule identifies risk for rheumatoid arthritis

A cut point of 8 or higher on the Leiden clinical prediction rule is linked to a good predictive value in determining which patients with undifferentiated arthritis are at risk for developing rheumatoid arthritis, according to a meta-analysis of data sets examined with the tool.

A number of research studies have indicated that early aggressive treatment of patients with the first signs of rheumatoid arthritis (RA) can delay or diminish joint damage and functional disability from disease progression. An estimated one-third of patients with undifferentiated arthritis (UA) will progress to RA, while approximately half will remit their disease; the remainder tend to develop other conditions, such as osteoarthritis, psoriatic arthritis, and reactive arthritis. Thus, it is important to identify which patients will progress to RA so they can benefit from early treatment, as well as prevent unnecessary treatment in those who won’t develop RA, the investigators said (Semin. Arthritis Rheum. 2013 Oct. 17 [doi:10.1016/j.semarthrit.2013.08.005]).

Emma McNally of the Royal College of Surgeons in Ireland, Dublin, and her associates performed this meta-analysis, using the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guidelines and the QUADAS (Quality Assessment of Diagnostic Accuracy Studies). Their search of the MEDLINE, Cochrane Library, EMBASE, Cinahl, and Google Scholar databases identified more than 5,000 papers; they also performed a hand search of references found in the retrieved articles. They winnowed the papers down to four articles containing six data sets. A total of 1,084 patients were included in the analysis of the Leiden clinical prediction rule (CPR), which showed that as the total Leiden CPR score increased, the sensitivity decreased and the specificity increased.

"The pooled data indicate that a cut-point of greater than or equal to 8 is better for ruling in RA, with a higher specificity (0.95) than sensitivity (0.49). However, the cut points of greater than or equal to 9 and greater than or equal to 10 were associated with higher specificities (0.99 [for both])," Ms. McNally and her associates wrote, and the latter "may offer a more optimal marker for ruling in RA and determining initiation of treatment."

The duration of follow-up in the studies in the meta-analysis varied from a minimum of 6 months to 30 months; three data sets followed patients for 1 year, as did the study from which the Leiden CPR was derived. The percentage of UA patients who developed RA ranged from 31% to 76% across the six data sets. Four data sets used a modified version of the Leiden CPR, replacing the variable "severity of morning stiffness" with "duration of morning stiffness."

In two studies, some patients started receiving disease-modifying antirheumatic drugs (DMARDs), and in one study, some patients also received glucocorticoids.

In addition to the fact that some patients were treated during the studies, which would likely increase the discriminative ability of the CPR, other considerations to make when evaluating this meta-analysis include the fact that revised classification criteria have been developed for the diagnosis of RA since the Leiden CPR was developed. The American College of Rheumatology (ACR)/European League Against Rheumatism (EULAR) 2010 criteria were developed to "facilitate the identification of RA at an earlier stage (prior to the development of bone damage), compared with the 1987 ACR criteria," the investigators explained. A review of the two criteria showed that the new 2010 criteria had a higher sensitivity but lower specificity, compared with the 1987 criteria. That and other factors led the authors to conclude that use of the 1987 criteria in the studies in this meta-analysis was appropriate.

Nonetheless, Ms. McNally and her associates said that no impact-analysis studies of the Leiden CPR have been performed, and assessing the clinical effect of a CPR is key. Such studies, usually randomized, controlled trials, would explore the effect of the rule on "patient outcomes such as progression to RA, prescription of DMARDs, prevention of joint destruction, and disease remission."

"Further studies also need to be conducted to determine the predictive value of the Leiden CPR in patients with UA, if the new ACR/EULAR 2010 criteria for the diagnosis of RA are implemented in clinical practice," they wrote. The results of one study so far suggest that in this situation, the rule might be more useful as a marker of disease persistence of UA, unless the CPR was modified to accommodate the new criteria.

This research was funded by the Health Research Board in Ireland. The authors listed no relevant financial disclosures.

A cut point of 8 or higher on the Leiden clinical prediction rule is linked to a good predictive value in determining which patients with undifferentiated arthritis are at risk for developing rheumatoid arthritis, according to a meta-analysis of data sets examined with the tool.

A number of research studies have indicated that early aggressive treatment of patients with the first signs of rheumatoid arthritis (RA) can delay or diminish joint damage and functional disability from disease progression. An estimated one-third of patients with undifferentiated arthritis (UA) will progress to RA, while approximately half will remit their disease; the remainder tend to develop other conditions, such as osteoarthritis, psoriatic arthritis, and reactive arthritis. Thus, it is important to identify which patients will progress to RA so they can benefit from early treatment, as well as prevent unnecessary treatment in those who won’t develop RA, the investigators said (Semin. Arthritis Rheum. 2013 Oct. 17 [doi:10.1016/j.semarthrit.2013.08.005]).

Emma McNally of the Royal College of Surgeons in Ireland, Dublin, and her associates performed this meta-analysis, using the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guidelines and the QUADAS (Quality Assessment of Diagnostic Accuracy Studies). Their search of the MEDLINE, Cochrane Library, EMBASE, Cinahl, and Google Scholar databases identified more than 5,000 papers; they also performed a hand search of references found in the retrieved articles. They winnowed the papers down to four articles containing six data sets. A total of 1,084 patients were included in the analysis of the Leiden clinical prediction rule (CPR), which showed that as the total Leiden CPR score increased, the sensitivity decreased and the specificity increased.

"The pooled data indicate that a cut-point of greater than or equal to 8 is better for ruling in RA, with a higher specificity (0.95) than sensitivity (0.49). However, the cut points of greater than or equal to 9 and greater than or equal to 10 were associated with higher specificities (0.99 [for both])," Ms. McNally and her associates wrote, and the latter "may offer a more optimal marker for ruling in RA and determining initiation of treatment."

The duration of follow-up in the studies in the meta-analysis varied from a minimum of 6 months to 30 months; three data sets followed patients for 1 year, as did the study from which the Leiden CPR was derived. The percentage of UA patients who developed RA ranged from 31% to 76% across the six data sets. Four data sets used a modified version of the Leiden CPR, replacing the variable "severity of morning stiffness" with "duration of morning stiffness."

In two studies, some patients started receiving disease-modifying antirheumatic drugs (DMARDs), and in one study, some patients also received glucocorticoids.

In addition to the fact that some patients were treated during the studies, which would likely increase the discriminative ability of the CPR, other considerations to make when evaluating this meta-analysis include the fact that revised classification criteria have been developed for the diagnosis of RA since the Leiden CPR was developed. The American College of Rheumatology (ACR)/European League Against Rheumatism (EULAR) 2010 criteria were developed to "facilitate the identification of RA at an earlier stage (prior to the development of bone damage), compared with the 1987 ACR criteria," the investigators explained. A review of the two criteria showed that the new 2010 criteria had a higher sensitivity but lower specificity, compared with the 1987 criteria. That and other factors led the authors to conclude that use of the 1987 criteria in the studies in this meta-analysis was appropriate.

Nonetheless, Ms. McNally and her associates said that no impact-analysis studies of the Leiden CPR have been performed, and assessing the clinical effect of a CPR is key. Such studies, usually randomized, controlled trials, would explore the effect of the rule on "patient outcomes such as progression to RA, prescription of DMARDs, prevention of joint destruction, and disease remission."

"Further studies also need to be conducted to determine the predictive value of the Leiden CPR in patients with UA, if the new ACR/EULAR 2010 criteria for the diagnosis of RA are implemented in clinical practice," they wrote. The results of one study so far suggest that in this situation, the rule might be more useful as a marker of disease persistence of UA, unless the CPR was modified to accommodate the new criteria.

This research was funded by the Health Research Board in Ireland. The authors listed no relevant financial disclosures.

A cut point of 8 or higher on the Leiden clinical prediction rule is linked to a good predictive value in determining which patients with undifferentiated arthritis are at risk for developing rheumatoid arthritis, according to a meta-analysis of data sets examined with the tool.

A number of research studies have indicated that early aggressive treatment of patients with the first signs of rheumatoid arthritis (RA) can delay or diminish joint damage and functional disability from disease progression. An estimated one-third of patients with undifferentiated arthritis (UA) will progress to RA, while approximately half will remit their disease; the remainder tend to develop other conditions, such as osteoarthritis, psoriatic arthritis, and reactive arthritis. Thus, it is important to identify which patients will progress to RA so they can benefit from early treatment, as well as prevent unnecessary treatment in those who won’t develop RA, the investigators said (Semin. Arthritis Rheum. 2013 Oct. 17 [doi:10.1016/j.semarthrit.2013.08.005]).

Emma McNally of the Royal College of Surgeons in Ireland, Dublin, and her associates performed this meta-analysis, using the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guidelines and the QUADAS (Quality Assessment of Diagnostic Accuracy Studies). Their search of the MEDLINE, Cochrane Library, EMBASE, Cinahl, and Google Scholar databases identified more than 5,000 papers; they also performed a hand search of references found in the retrieved articles. They winnowed the papers down to four articles containing six data sets. A total of 1,084 patients were included in the analysis of the Leiden clinical prediction rule (CPR), which showed that as the total Leiden CPR score increased, the sensitivity decreased and the specificity increased.

"The pooled data indicate that a cut-point of greater than or equal to 8 is better for ruling in RA, with a higher specificity (0.95) than sensitivity (0.49). However, the cut points of greater than or equal to 9 and greater than or equal to 10 were associated with higher specificities (0.99 [for both])," Ms. McNally and her associates wrote, and the latter "may offer a more optimal marker for ruling in RA and determining initiation of treatment."

The duration of follow-up in the studies in the meta-analysis varied from a minimum of 6 months to 30 months; three data sets followed patients for 1 year, as did the study from which the Leiden CPR was derived. The percentage of UA patients who developed RA ranged from 31% to 76% across the six data sets. Four data sets used a modified version of the Leiden CPR, replacing the variable "severity of morning stiffness" with "duration of morning stiffness."

In two studies, some patients started receiving disease-modifying antirheumatic drugs (DMARDs), and in one study, some patients also received glucocorticoids.

In addition to the fact that some patients were treated during the studies, which would likely increase the discriminative ability of the CPR, other considerations to make when evaluating this meta-analysis include the fact that revised classification criteria have been developed for the diagnosis of RA since the Leiden CPR was developed. The American College of Rheumatology (ACR)/European League Against Rheumatism (EULAR) 2010 criteria were developed to "facilitate the identification of RA at an earlier stage (prior to the development of bone damage), compared with the 1987 ACR criteria," the investigators explained. A review of the two criteria showed that the new 2010 criteria had a higher sensitivity but lower specificity, compared with the 1987 criteria. That and other factors led the authors to conclude that use of the 1987 criteria in the studies in this meta-analysis was appropriate.

Nonetheless, Ms. McNally and her associates said that no impact-analysis studies of the Leiden CPR have been performed, and assessing the clinical effect of a CPR is key. Such studies, usually randomized, controlled trials, would explore the effect of the rule on "patient outcomes such as progression to RA, prescription of DMARDs, prevention of joint destruction, and disease remission."

"Further studies also need to be conducted to determine the predictive value of the Leiden CPR in patients with UA, if the new ACR/EULAR 2010 criteria for the diagnosis of RA are implemented in clinical practice," they wrote. The results of one study so far suggest that in this situation, the rule might be more useful as a marker of disease persistence of UA, unless the CPR was modified to accommodate the new criteria.

This research was funded by the Health Research Board in Ireland. The authors listed no relevant financial disclosures.