Dactylitis describes a uniform swelling of a digit with inflammation causing a “sausage digit” and is one of the key identifying features of PsA. Dactylitis is one of the features used to make a diagnosis of PsA using the recently developed CASPAR classification criteria for PsA (Taylor et al. 2006). Dactylitis can be further characterised as acute/tender dactylitis where the digit is tender to touch and often erythematous and warm to touch, or as chronic/sub-acute/non-tender dactylitis where the digit is swollen but non-tender. It has been hypothesised that the chronic form occurs following an episode of acute dactylitis in some patients but this has not
been confirmed in studies. Acute or tender dactylitis has been shown to be a presenting feature of 33.5% of patients at their first visit to the clinic and to occur in 48% of patients at some point in their disease (Brockbank et al. 2005).
Our understanding of dactylitis, its pathogenesis and how to define it is limited. Studies using imaging have confirmed that physical examination can identify pathology in tender dactylitis. Olivieri et al scanned 12 tender dactylitic fingers and their matching contralateral normal digits with both MRI and US. Both scans identified distension of the flexor tendon sheaths and thickening of the flexor tendons, with no involvement in the extensor tendons. Physical examination by one physician identified flexor tendonitis in all of the involved digits with clinically normal extensor tendons (Olivieri et al. 1996). Thus it seems that clinical instruments for assessing dactylitis do have some criterion validity. However this study only assessed 12 obviously swollen digits and the contralateral normal digits. Therefore it seems likely that in normal clinical practice, there will be some variation between observers resulting in lower agreement particularly in “grey” cases, where digits may be slightly swollen or where matching digits on both sides are involved when it is harder to classify as dactylitis. This idea was later confirmed by a reliability study performed in Canada. This showed a moderate agreement (kappa 0.57, 95%CI=0.34, 0.82) between 10 experienced observers for number of digits with dactylitis (Gladman et al. 2004).
Clinical measures of dactylitis have been used as secondary outcome measures in clinical trials but the majority have used non-validated measures. The simplest measure used is a simple count of dactylitic digits, although some have counted only acute/tender dactylitis (Salvarani et al. 2003) and some have included both tender and non-tender digits (Clegg et al. 1996). Other large RCTs of new therapies have attempted to grade dactylitis by physician-rated severity in addition to counting affected digits (Kaltwasser et al. 2004; Antoni et al. 2005b) but none of these measures were validated prior to their inclusion in studies.
The Leeds Dactylitis Instrument (LDI) was developed in response to this need for a clinical, objective, validated outcome measure for dactylitis. This also provided the first numerical definition of dactylitis as an increase in circumference of the digit of more than 10% compared to the contralateral non-affected digit (Helliwell et al. 2005). This was based on the evaluation of the median difference in digital circumference between dactylitic digits and control digits.
The aim of the LDI is to provide a quantification of both the size of the swollen digit and the tenderness so that the score can differentiate between tender and non-tender dactylitis. The LDI tool is a circumferometer which is used to measure the circumference of the affected digit as near to the web space as possible.
This is compared to the contralateral unaffected digit as a ratio. If the contralateral digit is also dactylitic, then normative values based on population averages are provided. The tenderness scoring can be based on the RAI with tenderness scored from 0-3 (LDI scoring) or can be simplified to a dichotomous score of 0 for non- tender and 1 for tender (LDI basic) (Helliwell et al. 2005).
The first study of this instrument tested its reliability, investigating inter- and intraobserver agreement. A cross-sectional examination of seven patients by five examiners using the tool was performed. This showed a relatively poor interobserver reliability for identifying tender dactylitis and a poor agreement on non-tender dactylitis. This was improved significantly by using the LDI scoring system. Inter and intraobserver reliability for the LDI score was good, but was increased further using the LDI basic, suggesting that some of the variability was due to the inaccuracy of grading tenderness (Helliwell et al. 2005).
A longitudinal study was then performed to further investigate the use of this clinical tool and to compare it to other methods used previously. This study recruited 28 patients with dactylitis who were having a change in treatment for their PsA. All five available measures for dactylitis (tender dactylitis count (Salvarani et al. 2003), all dactylitis count (Clegg et al. 1996), IMPACT1 (Antoni et al. 2005b), LDI (Helliwell et al. 2005), LDI basic (Helliwell et al. 2005)) were performed at baseline and then regularly following the change in therapy. All measures showed a change with treatment after three and six months with a reasonable effect size for the variety of treatments used. The majority of these correlated with other clinical disease activity measures such as joint counts and VAS for disease activity (Healy and Helliwell 2007). Only the count of all dactylitic digits performed badly probably due to the inclusion of non-tender dactylitis which may not be thought to represent disease activity in patients’ and physicians’ opinions.
A subgroup of patients in the above study also had MRI scans performed at baseline and six months to assess the inflammation in the dactylitic digits. Similar to the Olivieri study (Olivieri et al. 1996), this showed that clinically tender dactylitic digits had significant MRI abnormalities compared to non-involved digits or non-tender dactylitis. However the correlation between the level of inflammation on MRI and clinical evaluation was moderate at best (0.37 for LDI local score and MRI score) (Healy et al. 2008). The identification of more inflammation on MRI when compared to a clinical score is unsurprising given the poor sensitivity of clinical examination compared to MRI as seen in other aspects of inflammatory arthritis (Szkudlarek et al. 2004; Szkudlarek et al. 2006).
Although the LDI and LDI basic measures do take longer to perform, particularly if multiple digits are involved, these measures perform better in terms of
both truth and discrimination when considering the tool in the context of the OMERACT filter (Healy and Helliwell 2007). Thus, it is the most validated clinical outcome measure available for dactylitis.