Article Text

This article has a correction. Please see:


Concise report
A diagnostic algorithm for persistence of very early inflammatory arthritis: the utility of power Doppler ultrasound when added to conventional assessment tools
  1. J E Freeston,
  2. R J Wakefield,
  3. P G Conaghan,
  4. E M A Hensor,
  5. S P Stewart,
  6. P Emery
  1. Academic Section of Musculoskeletal Disease, University of Leeds, Chapel Allerton Hospital, Leeds, UK
  1. Correspondence to Dr P Emery, Academic Section of Musculoskeletal Disease, University of Leeds, Chapel Allerton Hospital, Chapeltown Road, Leeds LS7 4SA, UK; p.emery{at}


Objectives: The aim of this study was to assess the value of power Doppler ultrasound (PDUS) in combination with routine management in a cohort of patients with very early inflammatory arthritis (IA).

Methods: 50 patients with ⩽12 weeks of inflammatory symptoms with or without signs had clinical, laboratory and imaging assessments. Diagnosis was recorded at 12 months. Assuming a 15% pre-test probability of IA, post-test probabilities for various assessments were calculated and used to develop a diagnostic algorithm.

Results: All patients positive for rheumatoid factor (RF) and/or cyclic citrullinated peptide (CCP) developed persistent IA, so the added value of PDUS was assessed in the seronegative (RF and CCP negative) group. The probability of IA in a seronegative patient was 6%. The addition of clinical and radiographic features raised the probability of IA to 30% and, with certain ultrasound features, this rose to 94%.

Conclusions: In seronegative patients with early IA, combining PDUS with routine assessment can have a major impact on the certainty of diagnosis.

Statistics from

Evidence suggests that the earlier treatment is started for rheumatoid arthritis (RA), the better the radiological and functional outcome,1 2 so there is a need to identify and suppress inflammation as early as possible. At this early stage, however, the diagnosis is often unclear and may be termed “undifferentiated arthritis” (UA) when diagnostic criteria for RA are not fulfilled. Some patients will develop progressive erosive disease whereas, in others, symptoms will resolve.3 With the potential consequences of overtreating a significant number of patients at a stage when many are often reluctant to accept aggressive treatment regimens, clinicians are increasingly looking for accurate tools to predict outcome.

Biomarkers such as rheumatoid factor (RF) and anti-cyclic citrullinated peptide (ACPA) antibodies combined with conventional hand and feet radiography (CR) are used to assess patients with early UA,4 as for RA.5 There are, however, potential problems with relying on such tools alone. Serology will fail to identify those who develop seronegative inflammatory arthritis (IA) and it takes time for erosions to appear on CR. Inflammatory markers such as C-reactive protein (CRP) can be influenced by other biological processes such as infection.

Sensitive imaging such as power Doppler ultrasound (PDUS) is able to identify both subclinical synovitis6 and early erosive disease preceding changes seen on CR.7 This study aimed to assess the diagnostic role of PDUS in combination with current management in a cohort of patients with very early IA, with subsequent development of a diagnostic algorithm for the prediction of persistent IA in this group.


Fifty patients (38 women, age range 21–80 years) were recruited consecutively from an Early Arthritis Clinic with inflammatory hand symptoms (defined as early morning stiffness for at least 1 h in the hands with or without clinical synovitis) for ⩽12 weeks. Baseline clinical assessment involved 28 tender and swollen joint counts and measurement of RF, ACPA and CRP. Imaging involved hand and feet CR (assessed for erosions) and grey scale (GS) and PDUS using a Philips HDI 5000 machine (Best, The Netherlands) with a medium wall filter applied. Bilateral metacarpophalangeal joints, flexor tendons and wrists were scanned by an experienced ultrasonographer (RJW) who was blind to the other assessments. Each joint was scored for GS and power Doppler (PD) on a 0–3 semi-quantitative scale8 and the presence or absence of flexor tenosynovitis was noted. There was no formal power calculation as this was a pilot study. The sample size of 50 patients was too small to perform logistic regression analysis.

Sensitivities, specificities and likelihood ratios of the parameters described above were determined. For simplicity, parameters were dichotomised (eg, any joint with a PD score ⩾1; table 1). A semi-quantitative cut-off of 3 was used for GS but ⩾1 for PD, as low-level GS can be seen in normal subjects.9

Table 1

Sensitivity, specificity and likelihood ratios for IA features

The prevalence of IA among patients with any kind of hand pain was first addressed. The only published study examining the prevalence of IA in patients with hand pain presenting from an unselected population such as primary care was that by Spies-Dorgelo et al.10 In the study cohort, 7.9% were found to have RA and 15.8% tenosynovitis. From these data a maximum estimated prevalence of IA of 15% was assumed, and this figure was used as the pre-test probability (PreTP).10

The knowledge of the sensitivity and specificity of a diagnostic test in combination with the PreTP allows calculation of the post-test probability (PostTP) of IA after a positive or negative test.11 The probability of the presence of persistent IA at 12 months, depending on the presence or absence of the above features of IA, was calculated using the formula based on Bayes’ theorem.11


One patient was lost to follow-up at 12 months so baseline data were analysed on 49 patients. At 12 months, 23/49 (47%) had RA, 15/49 (31%) had “other” IA and 11/49 (22%) had non-IA (including osteoarthritis and treated crystal arthropathy). The “other” IA group consisted of UA (n = 12), reactive IA following Chlamydia infection (n = 1) and connective tissue disease-associated IA (n = 2)). All diagnoses were made by the treating rheumatologist who was blind to the US results. Overall, 30/49 patients were negative for both RF and anti-cyclic citrullinated peptide (anti-CCP) antibodies.

Table 1 shows the sensitivity, specificity and LR for each parameter of IA assessed in the study. The diagnostic algorithm is shown in fig 1. RF and/or CCP positivity are generally accepted as the first test performed in clinical practice and are therefore used as the first diagnostic step. All patients in our study who were RF and/or CCP positive had developed persistent IA at 12 months, so most of the algorithm refers to patients who were negative for both RF and CCP. Figure 1 shows the changes in PostTP that occur with various combinations of positive and negative assessments. These changes are summarised as follows.

Figure 1

Suggested decision tree for diagnosing inflammatory arthritis (IA) in a very early cohort, incorporating pre-test and post-test probabilities (PreTP and PostTP). The percentage values shown in bold refer to the PostTP value/range depending on the test results shown in the preceding box. CCP, cyclic citrullinated protein; ↑CRP, C-reactive protein level above normal range; EMS, early morning stiffness; ero, erosion; GS, grey scale; PD, power Doppler; RF, rheumatoid factor; SwJ, number of swollen joints; XR, radiographs of hands and feet; +ve, positive test; −ve, negative test.

In our study cohort the probability of IA in a patient with a negative RF and CCP was only 6%, assuming a background prevalence of 15% (ie, negative serology reduces the chance of IA). The addition of typical features such as a positive CRP, at least one swollen joint and erosion on CR raised the PostTP to 30%. Subjects negative for these features had a PostTP of 1%.

In those who were seronegative for RF and CCP and positive for CRP, swollen joints and CR erosion, the presence of US features such as a semi-quantitative score of 3 for GS, ⩾1 for PD and at least one erosion increased the probability of IA from 30% to a PostTP of 94%. If only one US feature was positive, the probability of IA was 0–39% which increased to 8–85% if two US features were positive. If all US features were negative, the PostTP of IA was only 0–5%.


This is the first study to address the diagnostic benefit of adding US to conventional clinical tools in a cohort of patients with very early IA, providing crucial evidence-based support for this emerging technology in a climate where, increasingly, tools must prove their cost-effectiveness. We have shown that the probability of IA in patients presenting with very early hand symptoms, with or without signs, can be predicted according to the presence or absence of certain clinical features, laboratory tests and imaging findings. The final PostTP depending on combinations of features can then help in making a diagnosis of persistent IA.

Our pilot study data both confirm the predictive power of CCP and RF positivity and also suggest that the main role for US in the diagnosis of IA may be in the CCP/RF negative patient group where there is still significant diagnostic uncertainty. This is especially the case in those seronegative patients in whom 1–3 conventional features are positive, because positive US features can raise the probability from 2–30% to 50–94%.

The emphasis in this study on a cohort with very early IA stems from work by Green et al12 who showed that, before 12 weeks, the ACR criteria were not useful for identifying RA. Other studies on early cohorts include work from Leiden which has focused on a group of patients with UA (probably “pre-rheumatoid”).4 For example, Van der Helm-van Mil et al have developed a prediction rule for patients with UA, but the potential contribution from imaging was not assessed.13 Other studies such as that by Raza et al14 have examined cohorts with very early IA but did not look at the added benefit of imaging. Karim et al15 have previously examined the clinical utility of US in a cohort of general rheumatology patients referred for US, showing that it changed both the site-specific diagnosis and overall management. This study did not, however, examine the added benefit of US to conventional tools in a cohort with very early IA.

The main limitation of this study is the small number of study subjects which, when divided into their ultimate diagnosis at 12 months, produced even smaller numbers. The principal reason for this was the difficulty capturing patients with symptoms for ⩽12 weeks, as primary care physicians frequently refer patients with longer symptom durations. The fact that all seropositive patients developed IA limited the ability to examine the added value of US. This study can therefore only provide preliminary data, but it suggests that US may be a useful tool in the assessment of the clinically difficult seronegative group. In addition, this study did not attempt to show the potential contribution of US to patient management which is an important area for future work.

In summary, further research needs to be done to validate the results found in this study and, in particular, to validate the diagnostic algorithm in a different cohort of patients with very early IA. This work has highlighted the potential role for US in further characterising the seronegative group of patients, an important group of patients with few prognostic indicators.


View Abstract


  • Funding JEF was supported by a Centocor Research Fellowship.

  • Competing interests None.

  • Patient consent All patients gave their informed consent.

  • Provenance and Peer review Not commissioned; externally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Linked Articles

  • Miscellaneous
    BMJ Publishing Group Ltd and European League Against Rheumatism