This was a prospective, non-inferiority, multicentre randomised control trial, which aimed to assess the effectiveness of a standardised approach in the etiological diagnosis of uveitis versus an open strategy, where clinicians could perform any test. The authors developed their standardised strategy by incorporating data from recent studies and expert opinions and tailoring it to the anatomic type of uveitis. The first step of the strategy was common for all types of uveitis and included minimal work up, such as a full blood count, ESR, CRP, chest radiograph, syphilis serology and TB skin test. If no diagnosis was made then the clinician would order more complex, targeted investigations in a second and third step. Depending on the anatomical site of the uveitis and its duration, the second step could include HLA B27, ACE, chest CT, Toxoplasma serology, complement, antinuclear antibody, antiphospholipid antibodies, antineutrophil cytoplasmic antibodies. The third step would include anterior chamber tap, salivary gland biopsy, bronchoscopy, lumbar puncture, brain MRI or vitrectomy.
The study included 903 consecutive uveitis patients attending one of the 23 participating centres between June 2010 and May 2013. Randomisation was done by computer. As some patients were lost to follow-up, only 709 patients completed the trial (335 in the standardised and 374 in the open group). Exclusions due to protocol deviations occurred in 32 patients in the standardised group and one in the open group, altering the sample numbers to 303 versus 373 respectively. In the standardised group there were statistically more anterior (72.3% versus 60.8%) and acute uveitis (69.6% versus 60.7%). In the open group there were more recurrent (16.9% versus 9.6%) and posterior uveitis (17.8% and 8.2%). At baseline, each patient received a visual acuity test, slit-lamp examination with intraocular pressure check and dilated indirect ophthalmoscopy, as well as ancillary tests fundus fluorescein angiography (FFA) and optical coherence tomography (OCT) as required. They were followed up for six months and then re-examined at 12 months if an etiological diagnosis had not been reached. The primary outcome was the proportion of patients with an established a etiological diagnosis at six months for both strategies. Any diagnoses made in this group following this period (by free examinations) were considered as a standardised strategy failure. Secondary aims in the standardised group were the proportion of established diagnoses at the end of each step and the ratio of any additional investigations that may have contributed to it. At six months 50.4% of patients in the standardised group had an established etiological diagnosis versus 54.4% in the open group. However, because there were two strategy failures in the standardised group, who had required free complementary investigations to establish a diagnosis, the actual percentage fell to 49.5%.
Comparisons for non-inferiority proved inconclusive as the 95% confidence interval included zero and the non-inferiority margin. There were more investigations in the open group (15.39) versus the standardised group (12.41). In the two groups the commonest aetiologies were systemic diseases (60.8%), infections (25.1%), ocular specific conditions (10.7%), masquerade syndromes (1.7%) and medications (1.4%). The commonest entities included HLAB27 (22.2%), sarcoidosis (18%), spondylarthritis (11%), tuberculosis (10.7%) and herpes virus (8.5%), all of which accounted for 70% of all diagnoses. Behcet’s disease only accounted for 4.2% and syphilis for 1.7%. In the standardised group 16 patients had a clinical diagnosis, although the diagnosis was not ascertained at time of inclusion (five Fuch’s heterochromic cyclitis, six herpes virus infection, two Vogt-Koyanagi-Harada’s disease, one spondylathritis, one Birdshot chorioretinopathy and one pars planitis). In the remaining patients a diagnosis was established at the first step in 75.7%, the second step in 21.3% and the third step in 1.5%. There were only 20 patients in the standardised and 12 in the open group who required a 12 month review.
The authors acknowledge the limitations of their study, such as the difference in clinical characteristics in both groups at inclusion, the possibly less efficient randomisation which did not occur individually but by clusters, and despite 23 clusters, a few clusters contained the majority of patients and the fact that demographic information about race, whilst useful, was not collected due to national legislation (France). Another limitation affecting the study validity and power, is the reduction in patients in the standardised group that from 452 originally fell to 303. The authors suggest further studies to assess the cost-effectiveness of the strategy and the diagnostic yield of each test, in order to ascertain the best approach in the diagnosis of uveitis.