Development and initial validation of the vasculitis damage index for the standardized clinical assessment of damage in the systemic vasculitides.код для вставкиСкачать
ARTHRITIS & RHEUMATISM Vol. 40, No. 2, February 1997, pp 371-380 0 1997, American College of Rheumatology 371 DEVELOPMENT AND INITIAL VALIDATION OF THE VASCULITIS DAMAGE INDEX FOR THE STANDARDIZED CLINICAL ASSESSMENT OF DAMAGE IN THE SYSTEMIC VASCULITIDES A. R. EXLEY, P. A. BACON, R. A. LUQMANI, G. D. KITAS, C. GORDON, C. 0. S. SAVAGE, and D. ADU Objective. To develop and validate the Vasculitis Damage Index (VDI)for the standardized clinical assessment of damage in the systemic vasculitides. Methods. Using a nominal group consensus approach, the Birmingham Vasculitis Group generated guiding principles for assessment of damage in all systemic vasculitides. Damage was defined as irreversible change resulting from scars. Consensus principles were developed into the VDI, including guidelines for use, a list of items of damage, and a glossary. Results. For 100 surviving patients with systemic vasculitis, the median VDI score at last observation was 3 (range 0-8). Within the Wegener’s granulomatosis subgroup, the median VDI score for 12 non-survivors was higher than for 47 survivors (non-survivors median score 7, interquartile range 5-8 versus survivors median score 4, interquartile range 2-5; P = 0.003). VDI scores for 100 patients with systemic vasculitis increased from initial presentation to last observation by a median score of 3 (range 1-4; P C 0.001). The VDI assesses more items and is more sensitive to change than other indices of damage (P < 0.001). Using the VDI, trained observers can produce moderately consistent damage scores. Conclusion. The VDI is a sensitive, reproducible, comprehensive, and credible clinical tool for quantifying damage. The data presented herein should enable furSupported by the Arthritis and Rheumatism Council. A. R. Exley, MD, MRCP, P. A. Bacon, FRCP, R. A. Luqmani, DM, MRCP, G. D. Kitas, PhD, MRCP, C. Gordon, MD, MRCP, C. 0. S. Savage, PhD, FRCP, D. Adu, FRCP: The University of Birmingham, Birmingham, UK. Address reprint requests to A. R. Exley, MD, MRCP, Department of Rheumatology, The Medical School, Birmingham B15 2TT, UK. Submitted for publication February 23, 1996; accepted in revised form August 21, 1996. ther validation and testing of the VDI in specific vasculitic syndromes, and should facilitate the comparison of different therapies. Earlier studies based on postmortem series established the systemic vasculitides as severe life-threatening diseases (1,2). The introduction of cyclophosphamide for the treatment of Wegener’s granulomatosis altered the clinical course of this disease, improved survival, and dramatically reduced overall mortality (3,4). The systemic vasculitides still have an early acute mortality, but they have become chronic relapsing diseases. Thus, within 5 years of remission, systemic vasculitis may relapse in the majority of patients (5). Active disease accounts for half the patient-years of followup, reflecting both a poor initial treatment response in some patients and relapses in others. Remission is not achieved without cost, since more than half the patients experience serious morbidity due to disease or to drug toxicity (6). When the systemic vasculitides are viewed as chronic relapsing diseases, mortality is less appropriate as an outcome measure. Quality of survival becomes critical, and global assessment of disease severity and outcome is required. However, different aspects of disease should be assessed separately. Disease activity, damage, and functional ability differ conceptually and indicate very dissimilar treatment needs (7). We have previously described the Birmingham Vasculitis Activity Score (8), a disease activity score based on the intention to treat active, potentially reversible disease with cytotoxic drugs. These drugs have a narrow therapeutic range in patients with chronic relapsing diseases. Damage or scarring is unlikely to respond to immunosuppression. Therefore, we developed the Vasculitis Damage Index (VDI) to aid in the separation of damage from disease activity and to rationalize selection of therapy. Herein we describe the development and initial valida- EXLEY ET AL 372 tion of the VDI for the standardized clinical assessment of damage in the systemic vasculitides. MATERIALS AND METHODS Development of the VDI. The standardized assessment of damage in chronic diseases is a concept in evolution. In both systemic lupus erythematosus (SLE) and systemic vasculitis, the need to measure changes in morbidity has arisen from the substantial reduction in mortality achieved by modern therapy. The multidisciplinary Birmingham Vasculitis Group formed a nominal group to discuss standardized damage assessment in systemic vasculitis, borrowing from the principles of the Systemic Lupus International Cooperating Clinics (SLICC) and American College of Rheumatology (ACR) Damage Index for SLE (9). Guiding principles for damage assessment in systemic vasculitis were generated by consensus. The principles of damage assessment include a definition, a time frame, pro forma recording of data, attribution, accumulation, weighting, and feasibility. In group discussions, damage was defined as an irreversible process, the result of scars, not acute inflammation nor grumbling disease activity. Evidence to support a distinction between activity and damage, based on the duration of symptoms, was sought. In our own series of patients with active disease, remission usually occurred within 3 months of treatment (5). It was agreed that scoring should include all possible events recorded since the vasculitic syndrome began, with the understanding that events not necessarily related to vasculitis would be included. All events prior to the onset of vasculitis would be excluded. At presentation, the specific diagnosis may be unclear and patients with systemic vasculitis may be reclassified (10). Therefore, items of damage characteristic of each type of systemic vasculitis were included. A framework based on organ systems was chosen to facilitate the recording of any items of damage a patient might have. Achieving consensus on the pro forma recording of items of damage was difficult. Assessors from different specialties had different perspectives on damage. For example, chronic sinusitis was regarded as either a clinical or a radiologic diagnosis. Hearing loss might be due to either impaired conduction or nerve damage. Cranial neuropathy might include sensorineural deafness and blindness from optic nerve damage. Group members wished to separate subglottal stenosis that did not require surgery from subglottal stenosis that required surgery. It became clear that a final arbiter, available to each assessor, would be required. A glossary of definitions for each item of damage was thus compiled. During this process, other aspects related to recording items of damage were highlighted. For example, some acute events, such as resection of infarcted gut, are essentially irreversible, whereas other acute events may subsequently heal or result in damage. The nominal group felt that reproducibility might be impaired if the assessor made additional judgments of attribution and weighting. These judgments could be made separately. An alternative approach, in which damage would be recorded as simply present or absent, was considered. A global view of outcome was preferred, in which all events since the appearance of first symptoms would contribute to the assessment of each patient’s outcome. Thus, early events should contribute to each subsequent assessment of damage. In addition, group members agreed that damage assessment should be considered only part of a comprehensive clinical assessment. Therefore, standardized assessment of damage should be practical, easy to complete in a short period of time, and accesd~leto rtatistical analysis. The consensus guiding principles were used to generate a list of items of damage occurring in the systemic vasculitides. This list was further refined by extensive discussions between members of the Birmingham Vasculitis Group and other members of the European Community Concerted Action Group for the Study of Therapeutic Trials in Systemic Vasculitis (ECSYSVASTRIAL) (12). Preliminary results were subject to peer review in presentations at international meetings on systemic vasculitis (13). The consensus principles were developed into the VDI, including guidelines for its use (described below), a list of items of damage (Table I), and a glossary (Appendix 1). Guidelines for application of the VDI. Damage was defined as the presence of irreversible scars which have occurred since the onset of vasculitis. Onset of vasculitis was defined as the onset of first symptoms attributable to vasculitis (it is useful, for later analyses, to record dates of presentation to the hospital and date of diagnosis). Events must be present for at least 3 months before they can be scored as damage; this time limit was set arbitrarily to differentiate from current active inflammation. Acute irreversible events, such as gut resection, should be scored 3 months after the acute event to ensure consistency in scoring. A repeat episode can be scored, for items indicated, provided that it occurred at least 3 months since the first event. Damage can be scored as either present or absent. This is irrespective of whether damage may be later attributable to the vasculitic syndrome, complications of treatment, infection, or concurrent diseases, provided that the damage occurred since the onset of vasculitis. The index is cumulative; thus, the VDI score can only remain stable or increase. All previously scored items count at each subsequent assessment. The VDI score is the simple sum of the items of damage. Statistical analysis. Statistical analysis was performed using Minitab for Windows, Release 10.2 (Minitab, State College, PA). It was assumed that all data did not follow a Gaussian distribution. Therefore, nonparametric tests were used. Median values and interquartile ranges are given, unless otherwise stated. The Wilcoxon signed rank test was used for comparison of paired data, and the Mann-Whitney U test was used for comparison between 2 groups. The Kruskal-Wallis test was used as an extension of the Mann-Whitney U test for comparisons between more than 2 groups. Correlations between paired data were examined by determining Pearson’s correlation coefficients. All tests were 2-tailed, unless otherwise stated. The reproducibility of the VDI was examined in an informal assessment exercise. In May 1995, 4 members of ECSYSVASTRIAL from 3 different countries assessed patients with systemic vasculitis in the Birmingham Vasculitis Clinic. The VDI was scored for 8 patients who were seen in random order, according to a latin square design. A series of 2 x 2 tables was constructed between pairs of observers for each system within the VDI. The kappa statistic was calculated to determine agreement greater than that expected by chance (11). DEVELOPMENT AND VALIDATION OF THE VDI 373 Table 1. Items of damage in the Systemic Necrotizing Vasculitis (SNV) Damage Index and the Systemic Lupus International Cooperating ClinicsiAmerican College of Rheumatology (SLICCiACR) Damage Index compared with the Vasculitis Damage Index (VDI)* Items in common with VDI System, item I. Musculoskeletal 1. Significant muscle atrophy or weakness 2. Deforming or erosive arthritis 3. Avascular necrosis 4. Osteoporosis with fractures or vertebral collapse 5. Osteomyelitis 11. Skin 1. Alopecia 2. Skin ulceration 3. Oral ulceration 111. Ear, nose, and throat 1. Hearing loss 2. Nasal blockage or chronic discharge or crusting 3. Nasal bridge collapse or septa1 perforation 4. Chronic sinusitis or radiologic evidence of bone destruction 5. Subglottal stenosis without surgery 6. Subglottal stenosis with surgery IV. Pulmonary 1. Pulmonary hypertension 2. Pulmonary fibrosiskavity 3. Pleural fibrosis 4. Pulmonary infarction 5. Chronic asthma 6. Significant chronic breathlessness 7. Impaired pulmonary function tests V. Cardiovascular 1. Anginaicoronary artery bypass 2. Myocardial infarction 3. Second myocardial infarction 4. Cardiomyopathy 5. Valvular disease 6. Pericarditis 7. Hypertension VI. Renal 1. Estimated or measured GFR <50% 2. Proteinuria of >0.5 gmi24 hours 3. End-stage renal failure SNV SLICCiACR + + + + + + + - + + + + - + + + + + + + - +s +s +$ Items in common with VDI System, item VII. Gastrointestinal 1. Gut infarction 2. Mesenteric insufficiency or pancreatitis 3. Chronic peritonitis 4. Esophageal stricture or upper GI tract surgery VIII. Peripheral vascular 1. Absent peripheral pulse in 1 limb 2. Second episode of absent pulse in 1 limb 3. Absent peripheral pulses in 2 2 limbs 4. Major vessel stenosis 5. Claudication of the extremities 6. Complicated venous thrombosis 7. Minor tissue loss 8. Major tissue loss 9. Second episode of major tissue loss IX. Ocular 1. Cataract 2. Retinal change 3. Optic atrophy 4. Visual impairment/diplopia 5. Blindness in 1 eye 6 . Blindness in second eye 7. Orbital wall destruction X. Neuropsychiatric 1. Cognitive impairment 2. Major psychosis 3. Seizures 4. Cerebrovascular accident 5. Second cerebrovascular accident 6 . Cranial nerve lesion 7. Peripheral neuropathy 8. Transverse myelitis XI. Other damage 1. Premature gonadal failure 2. Marrow failure 3. Diabetes mellitus 4. Chronic chemical cystitis 5. Malignancy 6. Other features# SNV SLICCiACR + +- + + + +I - - - - - - - +t +t + + + + + + + + + + + - + +$ +* - - - - - ++ + + + +TI - - + + +$ +$ + + + +$ +$ + + + + * + or - indicates presence or absence of a common item. Items of damage listed as “second” are repeat events that are scorable if >3 months have elapsed since the first event. GFR = glomerular filtration rate; GI = gastrointestinal. i Items of damage in the VDI that are grouped together as single items in the SNV Damage Index. $ Adjacent items of damage in the VDI that are grouped together as single items in the SLICCiACR Damage Index. 5 An item of damage in the VDI that is listed as 2 items in the SLICCiACR Damage Index. llAn item of damage in the VDI that is listed as 2 items in the SNV Damage Index. # Any other items of damage the assessor wishes to record that are not listed in the VDI (these items are not included in the VDI score). RESULTS Content and face validity (comprehensiveness and credibility). The VDI was applied to 100 patients diagnosed with systemic vasculitis according to standard criteria (10,14-16), to derive VDI Scores for the last observation of each patient (Table 2). One assessor (ARE) carried out all the assessments. The median duration of disease from first symptoms to last observation was 6 years (range 3.2-10.0 years). The median VDI was 3 (range 0-8) (Figure 1). No items of damage were scored in 5 patients (5%). A wide variety Of items of damage were x m e d in the 100 Patients with systemic vasculitis, and some items of damage were scored in every organ system. Seventeen items, including 4 repeat items, were not used in these patients. A further 6 items EXLEY ET AL 374 Table 2. Diagnoses of 100 patients with systemic vasculitis No. of patients Diagnosis 13 12 20 21 Secondary systemic vasculitis* Polyarteritis nodosa Classic Wegener’s granulomatosis Localized (non-renal) Wegener’s granulomatosis Giant cell arteritist Unclassified systemic vasculitis Hennch-Schonlein purpura Churg-Strauss syndrome Microscopic polyangiitis All I 4 4 5 8 100 * Systemic rheumatoid vasculitis (n = 4), systemic lupus erytematosus vasculitis (n = l), and BehGet’s syndrome (n = 8). t Takayasu arteritis (n = 6) and giant cell (temporal) arteritis (n = 1). were used in a second group made up of 20 nonsurvivors with systemic vasculitis. The remaining items were avascular necrosis, pulmonary hypertension, pleural fibrosis, cognitive impairment, transverse myelitis, chronic pericarditis, esophageal stricture or upper gastrointestinal tract surgery, and 4 repeat items. Nineteen different items, including 6 cases of sicca syndrome, were recorded once in the category “Other features.” These items are not representative of damage in systemic vasculitis and therefore were not added to the VDI score. The ability of the VDI to record the accumulation of damage over time was examined by determining the change in VDI score over time. The increase in VDI score in 100 patients with systemic vasculitis from pre- Number of Patients 35 -I 30 25 20 15 10 5 0 0 1 2 3 4 5 6 7 8 9 10 Damage Score Figure 1. Damage index scores at last observation, as determined for 100 patients with systemic vasculitis. Gray bars = unweighted Systemic Lupus International Cooperating ClinicsiAmerican College of Rheumatology Damage Index score. Open bars = unweighted Systemic Necrotizing Vasculitis Damage Index score. Solid bars = Vasculitis Damage Index score. sentation to last observation was determined. The median interval was 5 years (range 2.5-8 years). There was a significant increase in the VDI score, by a median score of 3 (range 1-4; P < 0.001). Criterion validity (agreement with external criteria). In the absence of a true gold standard for damage in systemic vasculitis, death was used as a surrogate gold standard. The VDI score for all patients with Wegener’s granulomatosis who were non-survivors was compared with that for survivors: 12 non-survivors versus 47 survivors at last observation. Irreversible events present at the final admission were scored as damage in the final VDI score. The VDI score for non-survivors was significantly higher than that for survivors (non-survivors median score 7, interquartile range 5-8 versus survivors median score 4, interquartile range 2-5; P = 0.003). Construct/convergent validity (comparison with other indices). The convergent validity of the VDI was examined by comparison with 2 other damage indices: the Systemic Necrotizing Vasculitis (SNV) Damage Index (17) and the SLICCiACR Damage Index (18). The VDI was designed to apply to the systemic vasculitides. There are 64 items of damage, including 5 items of repeat episodes of damage, and 1 item to record any other damage that is not scored. The SNV Damage Index was used in a comparison of damage in patients with polyarteritis nodosa and Churg-Strauss syndrome (17). The SNV Damage Index has 34 items, including 4 items not found in the VDI, namely arrhythmia, peripheral vascular repair, infarction of abdominal organ (e.g., spleen, gall bladder), and chronic peptic ulceration. No items of damage in the ear, nose, and throat (ENT) system are included (Table 1). All aspects of renal damage count as 1 item, peripheral neuropathy is separated into motor and sensory neuropathies, and 9 items are weighted 2-fold. Events are required to persist for at least 6 months before they can be scored as damage, apart from certain irreversible events including cerebrovascular accident, myocardial infarction, and gut infarction. Although the SLICC/ACR Damage Index was developed for SLE, vasculitis is an important component of this disease. The SLICC/ACR Damage Index includes 40 items, with 5 repeat items (Table 1). There are 2 items unique to this index, namely shrinking lung and ruptured tendons, and no items of damage in the ENT system. The presence of end-stage renal failure excludes scoring other items of renal damage and is weighted 3-fold. Cranial and peripheral neuropathies count as 1 item. Events are required to persist for at least 6 months before they can be scored as damage. DEVELOPMENT AND VALIDATION OF THE VDI Table 3. Correlation between damage scores at last observation, in 100 patients with systemic vasculitis* Damage index ~ Scores in original format SNV SLICC/ACR Scores with weightings removed SNV SLICCiACR VDI SNV 0.594 0.570 0.807 0.670 0.632 0.786 ~~ * All correlations (using Pearson’s correlation coefficients) were statistically significant (P< 0.001). See Table 1 for definitions. The 3 damage indices were compared by disregarding the time limits for differentiating activity from damage. Weighted scores for the S N V Damage Index and for the SLICC/ACR Damage Index were calculated (9,17,18). Unweighted scores were also derived by scoring items of renal damage separately and removing weightings. No patients had any of the items unique to the SLICC/ACR Damage Index, nor the first 3 items of damage unique to the SNV Damage Index. Ascertainment of chronic peptic ulceration was incomplete and, therefore, this item was disregarded. The 3 damage indices were applied to 100 patients with systemic vasculitis (Table 2), and scores at last observation were derived. There was good correlation between the indices (Table 3), although the correlation was not absolute. There was a lower correlation between the VDI and weighted SNV Damage Index and between the VDI and the weighted SLICC/ACR Damage Index. The VDI scored significantly more items of damage than either the unweighted SNV Damage Index or the unweighted SLICC/ACR Damage Index (VDI median score 3, range 2-5 versus SNV median score 2, range 1-3 versus SLICC/ACR median score 1, range 1-2; P < 0.001) (Figure 1). The VDI also scored significantly more items of damage than either the weighted SNV Damage Index or the weighted SLICCi ACR Damage Index (VDI median score 3, range 2-5 versus SNV median score 3, range 1-4 versus SLICC/ ACR median score 1, range 1-2; P < 0.001). The VDI has a wider range of scores than either the unweighted SNV Damage Index or the unweighted SLICC/ACR Damage Index. The number of additional items scored by the VDI accounts for a sizable proportion of the VDI score at last observation. In 61% of patients, additional items scored accounted for at least half the total VDI score. In 22% of patients, additional items scored accounted for at least three-quarters of the total VDI score. Sensitivity is increased by maximizing the number of items of damage scored. 375 The ability of the 3 different damage indices to record the accumulation of damage over time was examined. Each damage index was applied to 100 patients with systemic vasculitis (as previously described), and the number of additional items scored between presentation and last observation was compared. The additional items of damage scored showed a good correlation among the different indices (Table 4), although the correlation was not absolute. There was a lower correlation between each pair of damage indices after weighting the SNV Damage Index and SLICC/ ACR Damage Index. The VDI scored significantly more additional items of damage between presentation and last observation than did the SNV Damage Index or the SLICC/ ACR Damage Index (VDI median number of additional items scored 3, range 2-5 versus SNV median number 2, range 1-3 versus SLICC/ACR median number 1, range 1-2; P < 0.001) (Figure 2). The increase in VDI score (median increase 3, range 2-5) between presentation and last observation was significantly greater than the increase in the weighted SNV Damage Index (median increase 2, range 0.25-3; P = 0.002) or in the weighted SLICC/ACR Damage Index (median increase 1, range 0-2; P < 0.001). Construct/discriminant validity (comparison of disease groups). The power of the damage indices to detect damage in each of the systemic vasculitides was examined in 100 patients with a range of vasculitic syndromes (Table 2). In particular, the ability of the VDI to detect more items of damage than the SNV Damage Index and the SLICC/ACR Damage Index was compared for each vasculitic syndrome. The additional sensitivity of the VDI was clearly seen in vasculitic syndromes poorly recognized by the other indices. VDI scores at last observation were higher, particularly in Wegener’s granulomatosis, Churg-Strauss syndrome, gi- Table 4. Correlation between damage scores for sensitivity to change, in 100 patients with systemic vasculitis* ~~~ ~ Damage index Scores in original format SNV SLICCiACR Scores with weightings removed SNV SLICCiACR VDI SNV 0.651 0.699 0.695 0.721 0.704 0.819 * Sensitivity to change determined by subtracting damage scores at presentation from damage scores at last observation. All correlations (using Pearson’s correlation coefficients) were statisitcally significant ( P < 0.001). See Table 1 for definitions. EXLEY ET AL 376 Number of Patients 35 1 30 25 20 15 10 5 0 0 1 2 3 4 5 6 7 a - Increase in Damage Score (Last observation Presentation) Figure 2. Sensitivity to change among the 3 damage indices. The incremental change in damage scores from presentation to last observation was determined for 100 patients with systemic vasculitis. Gray bars = change in unweighted Systemic Lupus International Cooperating ClinicsiAmerican College of Rheumatology Damage Index score. Open bars = change in unweighted Systemic Necrotizing Vasculitis Damage Index score. Solid bars = change in Vasculitis Damage Index score. ant cell arteritis, and the secondary vasculitides (P < 0.001) (Figure 3). A similar pattern was seen for the additional items of damage scored between presentation and last observation (P = 0.002) (Figure 4). The additional sensitivity of the VDI over the other damage indices was unaffected by weighting the SNV Damage Index and the SLICC/ACR Damage Index. Reliability. Reliability or reproducibility was examined by comparing the VDI scores as determined by independent assessors. A novice assessor was given a brief introduction to the principles and practice of scoring the VDI. Paired VDI scores were then determined in 50 patients with systemic vasculitis by 1 expert assessor and 1 novice assessor. The paired VDI assessments were carried out in random order, and scores for 6 months after presentation were determined. Paired VDI scores were identical in 64% of patients, and within 1 point in 78%. Reliability was dependent on strict adherence to the definition and guidelines for the application of the VDI. The differences in scoring were attributable to a few consistent errors, and a learning effect was apparent. In particular, the novice assessor did not consistently score all damage since the onset of first symptoms, and did not consistently score damage only if events were present for at least 3 months. These errors were seen for patients with secondary systemic vasculitis and for those with polyarteritis nodosa versus all other diagnoses; for these groups, the median difference in scoring was 3.5 for systemic vasculitis versus 1 for polyarteritis nodosa versus 0 for all other diagnoses (P < 0.001). In the assessment exercise, there was complete agreement between all assessors for two-thirds of the VDI, although these were items of damage scored as absent in all 8 patients. In order to highlight areas of disagreement, the analysis was confined to the 21 items in which any assessor scored damage. The mean value of the kappa statistic for each assessor was 0.41, 0.38, 0.36, and 0.34. Analysis of agreement for each of the systems in the VDI revealed important differences between assessors. There was good agreement between all assessors for renal/neurologic systems (mean kappa statistic 0.61) and the ENT system (mean kappa statistic 0.41). In the other systems, there was good agreement between 2 assessors versus poor agreement between the other assessors. Thus, the mean kappa values were 0.65 versus 0.1 for musculoskeletaVskin systems, 0.53 versus 0.15 for the pulmonary system, and 0.53 versus 0.24 for cardiovascular/peripheral vascular systems. Feasibility. The score sheet for the VDI is a tabulated list of items of damage grouped into 10 Additional Items of Damage 4- 321 - A I+ T f T A 01 SSV PAN WG LWG GCA USV HSP CSS I MPA Specific Diagnosis Figure 3. Power of the Vasculitis Damage Index (VDI) to detect more damage than the Systemic Necrotizing Vasculitis Damage Index (SNV) or the Systemic Lupus International Cooperating Clinics/ American College of Rheumatology Damage Index (SLICC/ACR), in each of the systemic vasculitides. Damage scores in 100 patients with systemic vasculitis were measured at last observation using the VDI, the SNV, and the SLICCIACR. All weightings were removed. The number of additional items scored by the VDI was dependent on the specific diagnosis (n = 2 X 100 scores; P < 0.001). Results are expressed as the median and interquartile range for each diagnosis. SSV = secondary systemic vasculitis; PAN = polyarteritis nodosa; WG = classic Wegener’s granulomatosis; LWG = localized (non-renal) Wegener’s granulomatosis; GCA = giant cell arteritis; USV = unclassified systemic vasculitis; HSP = Henoch-Schonlein purpura; CSS = Churg-Strauss syndrome; MPA = microscopic polyangiitis. DEVELOPMENT AND VALIDATION OF THE VDI Additional Items of Damage *w 3 T T 41 01 - SSV PAN WG LWG f€ GCA USV HSP * T CSS MPA Specific Diagnosis Figure 4. Power of the VDI to detect a greater increment in damage than the SNV or the SLICC/ACR, in each of the systemic vasculitides. The increment in damage scores (sensitivity to change) between presentation and last observation in 100 patients with systemic vasculitis was measured using the VDI, the SNV, and the SLICC/ACR. All weightings were removed. The VDI detected a greater increment in damage scores (was more sensitive to change) than the other indices, and the additional items scored vary depending on the specific diagnosis (n = 2 X 100 scores; P = 0.002). Results are expressed as the median and interquartile range for each diagnosis. See Figure 3 for definitions. organ-based systems, 1 general category, and a space for other comments (Table 1). The tabulated score sheet and glossary enable the clinician to rapidly (within 5 minutes) record items of damage present at the assessment visit. At subsequent assessments, additional items of damage are added to new columns on the score sheet. A simple computer program, written in dBase IV (Borland International, Scott Valley, CA), enables scores to be transferred from the score sheets or directly entered. An integrated package, the Vasculitis Integrated Total Assessment Log, has been developed for the standardized assessment of disease activity, damage (using the VDI), and function in patients with systemic vasculitis (12). DISCUSSION The quality of survival is critical in chronic relapsing diseases with a high morbidity, such as the systemic vasculitides. We have developed a vasculitis damage index to enable a detailed assessment of damage in the systemic vasculitides. The VDI is an integrated package that includes guidelines for its application, a list of items of damage, and a glossary. The tabulated layout of the VDI, based on organ systems, enhances its clinical 377 utility. Some of the principles underlying the VDI are shared by the SLICC/ACR Damage Index and the SNV Damage Index. There is good correlation between damage indices, but the more comprehensive VDI has increased sensitivity, both at single assessments and over time. The power of the VDI to record the accumulation of items of damage over time supports its validity as a clinical index of outcome. This initial validation suggests that the VDI is a sensitive, reproducible, comprehensive, and credible clinical instrument for recording the accumulation of damage. Thus, the VDI should prove to be a useful clinical tool for the investigation of different therapies. The distinction between disease activity and damage is fundamental (7). The Birmingham Vasculitis Activity Score was based on an intention to treat disease with immunosuppressive therapy. Implicit in the development of the Birmingham Vasculitis Activity Score was the concept that disease activity is potentially reversible. Disease activity was defined as the development of disease features that were new or worse within the last 4 weeks (8). Remission induction in patients with systemic vasculitis is, in our experience, generally achieved within 3 months of treatment (5). In contrast, the fundamental attribute of damage is its irreversibility. Although the time limit of 3 months used to distinguish disease activity from damage is arbitrary, it does have a clinical validity. Moreover, a definite time limit is essential for reproducible assessment of damage. The measurement of damage rather than activity is particularly important in assessment of long-term outcome in chronic diseases (19). Indeed, an assessment of damage revealed an unexpectedly poor outcome in patients with Wegener’s granulomatosis, despite treatment with cytotoxic drugs (6). Studies can now be undertaken to look for a lack of correlation between current activity and damage score, which would confirm that these 2 elements are different measures of disease. The VDI is a purpose-built package that incorporates items of damage applicable to each of the systemic vasculitides. There are more items of damage in the VDI than in either of the other damage indices. The VDI includes the additional items of musculoskeletal and peripheral vascular damage. Items of damage in the ENT system are also incorporated, to reflect their significant impact in patients with Wegener’s granulomatosis. These items include early-onset deafness, chronic sinusitis, nasal deformity, and subglottal stenosis. The absolute number of additional items scored for each patient is low. Nevertheless, this increment does contribute a significant proportion to the total score. 378 Moreover, there is a wider range of damage scores in the VDI than in the other damage indices. The VDI detects more additional items of damage between presentation to hospital and last observation. In this initial validation study, the power of the VDI was highlighted by its ability to detect more damage, compared with the other indices, in common vasculitic syndromes, e.g., Wegener’s granulomatosis, giant cell arteritis, and secondary systemic vasculitis. The increased power of the VDI was unaffected by weighting other damage indices, and reflects more than the larger number of items of damage listed in the VDI. Our data indicate that these additional items of damage do occur in the systemic vasculitides, both early and later in the course of the disease, depending on the vasculitic syndrome. It seems likely that the design of the VDI will result in greater power to discriminate between different interventions. An index that provides a large number of items maximizes information and sensitivity. However, too large a number of items will restrict the feasibility of the index in a clinical setting. Using the VDI, items from each organ system were scored in our patients with systemic vasculitis. Few items of damage from the VDI were not represented in 100 patients with systemic vasculitis. Nevertheless, the optional open category “Other features” failed to reveal significant items of damage not already included. Although 19 other features were noted, each occurred in only 1 patient and was therefore not included in the VDI score. Identification of a few items among many is aided by the format of the VDI. The tabulated layout enables early items of damage to be readily carried forward to each subsequent assessment. Data collection is not dependent on computer support, although a computer program will assist in data collection and analysis. Reproducibility of the VDI is enhanced by training. Errors made by inexperienced assessors highlighted a need to establish clear guidelines for application of the VDI. An informal assessment exercise revealed good agreement between some assessors, while others had particular problems scoring items of damage in the musculoskeletal/skin and pulmonary systems. Therefore, a more explicit glossary was devised. It is likely that reproducibility will be enhanced by the results and written text presented herein. New assessors are advised to carefully study the guidelines for application of the VDI (as previously described) and the glossary (Appendix 1). We are currently establishing a training program for the VDI, with the help of other members of the ECSYSVASTRIAL group. EXLEY ET AL The VDI records the presence or absence of items of damage, to indicate the cumulative toll of systemic vasculitis, its treatment, and comorbidity. We believe this approach minimizes interobserver variation. The distinction between damage due to the vasculitic syndrome, drug toxicity, infection, or concurrent disease may not be clear when the VDI assessment is undertaken. Damage may be multifactorial, and thus, attribution of damage is controversial. Reproducibility is preserved by retaining a basic database which can be transformed as desired. The itemized design of the VDI readily allows analysis in terms of the total score or the number of organs involved. Items can be selected and grouped according to a variety of criteria. A nominal group consensus approach can highlight items that may be significant in a variety of outcomes. This process of selection can be tested against gold standards of damage. In the absence of a true gold standard, the VDI score for non-survivors was chosen as a surrogate. As expected, non-survivors had high damage scores. This analysis could be extended to other surrogate standards, e.g., the development of end-stage renal failure. Damage can also be judged against the important outcome functional class, since preliminary studies have revealed significant functional impairment, even in patients with inactive disease (20). Serial studies of the VDI could be used to investigate early indicators of poor prognosis. Selection rather than weighting may be a more flexible, powerful, and reproducible approach to analysis of outcome. The development and initial validation of the VDI, as described herein, suggests that this index should be applicable to each of the systemic vasculitides. However, further validation is required; e.g., to test discriminant validity, studies could compare the VDI scores for different vasculitic syndromes such as microscopic polyangiitis versus polyarteritis nodosa or localized (nonrenal) versus classic Wegener’s granulomatosis, as well as the VDI scores for patients who experience a single relapse versus those who have multiple relapses. We propose that the VDI should be used as a clinical tool to compare different therapeutic regimens and to investigate the contribution of damage to morbidity and mortality in the systemic vasculitides. Clearly, we need to improve outcome without exchanging damage due to disease for additional damage due to treatment. The present VDI scoring system has been adopted by the ECSYSVASTRIAL group as an integral part of its record books for 3 core protocols in systemic vasculitis. Data from these therapeutic trials will allow further prospective validation of the VDI. DEVELOPMENT AND VALIDATION OF THE VDI ACKNOWLEDGMENTS W e ar e grateful for the support of o u r clinical colleagues who referred their patients for assessment. W e thank the members of t h e Birmingham Vasculitis G r o u p , Dr. Kirsten d e Groot, and the ECSYSVASTRIAL group, as well as Dr. Pradeep Bambury and Dr. Deborah Symmons, for their help and advice. REFERENCES 1. Godman GC, Churg J: Wegener’s granulomatosis: pathology and review of the literature. Arch Pathol 58533-553, 1954 2. Rose GA, Spencer H: Polyarteritis nodosa. QJM 26:26-43, 19.57 3. Novack SN, Pearson CM: Cyclophosphamide therapy in Wegener’s granulomatosis. N Engl J Med 284:938-942, 1971 4. Fauci AS, Haynes BF, Katz P, Wolff SM: Wegener’s granulomatosis: prospective clinical and therapeutic experience with 85 patients for 21 years. Ann Intern Med 98:76-85, 1983 5. Gordon M, Luqmani RA, Adu D, Greaves I, Richards N, Michael J, Emery P, Howie AJ, Bacon PA: Relapses in patients with a systemic vasculitis. QJM 86:779-789, 1993 6. Hoffman GS, Kerr GS, Leavitt RY, Hallahan CW, Lebovics RS, Travis WD, Rottem M, Fauci AS: Wegener granulomatosis: an analysis of 158 patients. Ann Intern Med 116:488-498, 1992 7. Fries JF, Hochberg MC, Medsger TA Jr, Hunder GG, Bombardier C, and the American College of Rheumatology Diagnostic and Therapeutic Criteria Committee: Criteria for rheumatic disease: different types and different functions. Arthritis Rheum 37:454462, 1994 8. Luqmani RA, Bacon PA, Moots RJ, Janssen BA, Pall A, Emery P, Savage C, Adu D: Birmingham Vasculitis Activity Score (BVAS) in systemic necrotizing vasculitis. QJM 87:671-678, 1994 9. Gladman D, Ginzler E, Goldsmith C, Fortin P, Liang M, Urowitz M, Bacon P, Bombardieri S, Hanly JG, Hay E, Isenberg D, Jones J, Nived 0, Petri M, Richter M, Sanchez-Guerrero J, Snaith M, Sturfelt G, Symmons D: The SLICCiACR Damage Index for SLE (abstract). Arthritis Rheum 35 (suppl 9):S209, 1992 10. Jennette JC, Falk RJ, Andrassy K, Bacon PA, Churg J, Gross WL, 379 Hagen EC, Hoffman GS, Hunder GG, CG, Kdknberg CGM, McClusky RT, Sinico RA, Rees AJ, van Es LA, Waldherr R, Wiik A Nomenclature of systemic vasculitides: proposal of an international consensus conference. Arthritis Rheum 37: 187-192, 1994 11. Brennan P, Silman A: Statistical methods for assessing observer variability in clinical measures. BMJ 304:1491-1494, 1992 12. Bacon PA, Moots RJ, Exley AR, Luqmani R, Rasmussen N: Vital assessment of vasculitis. Clin Exp Rheumatol 13:275-278, 1995 13. Exley A, Banbury P, Carruthers D, Moots R, Luqmani R, Kitas G, Bacon P: The Vasculitis Damage Index (abstract). Arthritis Rheum 38 (suppl 9):S340, 1995 14. Hunder GG, Arend WP, Bloch DA, Calabrese LH, Fauci AS, Fries JF, Leavitt RY, Lie JT, Lightfoot RW Jr, Masi AT, McShane DJ, Michel BA, Mills JA, Stevens MB, Wallace SL, Zvaifler NJ: The American College of Rheumatology 1990 criteria for the classification of vasculitis. Arthritis Rheum 33:1065-1144, 1990 15. Luqmani RA, Bacon PA, Beaman M, Scott DG, Emery P, Lee SJ, Howie AJ, Richards N, Michael J, Adu D: Classical vs non-renal Wegener’s granulomatosis. QJM 87:161-167, 1994 16. O’Neill TW, Rigby AS, Silman AJ, Barnes C: Validation of The International Study Group Criteria for BehGets Disease. Br J Rheumatol 33:115-117, 1994 17. Abu-Shakra M, Smythe H, Lewtas J, Badley E, Weber D, Keystone E Outcome of polyarteritis nodosa and Churg-Strauss syndrome: an analysis of twenty-five patients. Arthritis Rheum 37:1798-1803, 1994 18. Gladman D, Ginzler E, Goldsmith C, Fortin P, Liang M, Urowitz M, Bacon P, Bombardieri S, Hanly J, Hay E, Isenburg D, Jones J, Kalunian K, Maddison P, Nived 0, Petri M, Richter M, SanchezGuerrero J, Snaith M, Sturfelt G, Symmons D, Zoma A: The development and initial validation of the Systemic Lupus International Cooperating ClinicsiAmerican College of Rheumatology Damage Index for Systemic Lupus Erythematosus. Arthritis Rheum 39:363-369, 1996 19. Pincus T, Stein CM: What is the best source of useful data on the treatment of rheumatoid arthritis: clinical trials, clinical observations, or clinical protocols? J Rheumatol 22:1611-1617, 1995 20. Exley A, Moots R, Carruthers D, Luqmani R, Bacon P A Functional assessment in vasculitis using the SF-36 (abstract). Arthritis Rheum 38 (suppl 9):S340, 1995 APPENDIX 1: THE VASCULITIS DAMAGE INDEX GLOSSARY System, item 1. Musculoskeletal 1 .l. Significanr muscle arrophy or weakness: Demonstrated on clinical examination (not attributable to cerebrovascular accident). 1.2. Deforming or erosive arthritis: Deformities on clinical examination confirmed by radiographs (excluding avascular necrosis); demonstrated on clinical examination and radiographs. 1.3. 0,slropormic with fractures or veilebral collapse: By history, confirmed on radiographs (excluding avascular necrosis); ever, since the onset of vasculitis. 1.4.Avusculur necrosis: Demonstrated by appropriate radiologic techniques; ever, since the onset of vasculitis. I .5. 0.steomyelilU: Documented clinically, confirmed by radiographs and/or culture. 2. Skin 2.1. Alopecia: Major (c.g., requiring wig) chronic hair loss with or without scars, documented clinically. 2.2. Skin ulceration: Open Sore on skin surface, excluding that causcd by venous thrombosis. 2.3. Oral ulceration: Recurrent crops or persistent mouth ulcers requiring therapy. 3. Ear, nose, and throat (ENT) 3.1. Hearing loss: Any hearing loss due to middle ear involvement or to auditory nerveicochlear damage, preferably confirmed by audiometry. 3.2. Nasal blockagelchronic dischargelcnisting: Difficulties with breathing through the nose and/or with purulent discharge and/or with crust formation usually requiring nasal lavage. 3.3. Nasal bridge cullapseiseptal perfuratiun: Saddle nose deformity andlor perforation of nasal septum. 3.4. Chronic sinusitis/radiologic evidence uf bone destruction: Chronic purulent nasal discharge with sinus pain and/or radiologic evidence of sinusitis with or without bone destruction. 3.5. Subglottal stenosis without ,sup?ry: Persistent hoarseness and/or stridor, preferably confirmed by endoscopy and/or radiographs. 3.6. Subglottal stenosis with surgery: Confirmed by otolaryngnlogy surgeon. EXLEY ET AL 4. Pulmonary 4.1. Pulmonary hypertension: Right ventricular prominence or loud P2 (confirmed by cardiologic investigation if appropriate). 4.2. Pulmonary fibrosisicavity: According to physical signs and radiographs (confirmed by relevant tests if necessary); this may include patients who require pulmonary resection. 4.3. Pleural fibrosis: According to chest radiographs. 4.4. Pulmonary infarction: According to chest radiographs or ventilation/ perfusion scan. 4.5. Chronic asthma: Significant reversible airways obstruction. 4.6. Significant chronic breathlessness: Significant sympotomatic breathing difficulties and/or shortness of breath without bard signs on radiographs or lung function tests. 4.7. Impaired pulmonary function tests: Forced expiratory volume in 1 second or forced vital capacity 570%, or transfer coefficient for carbon monoxide (KCO) 570%. 5. Cardiovascular 5.1. Anginalcoronury artery bypass: On history, confirmed at least by electrocardiographic changes. 5.2. Myocardial infarction: On history, confirmed at least by electrocardiographic changes or cardiac enzyme elevation; ever, since the onset of vasculitis. 5.3. Second myocardial infarction: At least 3 months after first event. 5.4. Cardiomyuputhy: Chronic ventricular dysfunction, documented clinically or on appropriate investigation, 5.5. Valvular diseuse: Significant diastolic or systolic murmur, confirmed by cardio~ogictests if appropriate. 5.6. Pencarditis: Symptomatic pericardial inflammation or constriction for at least 3 months or pericardiectomy. 5.7. Hypertension: With a diastolic blood pressure >95 mm Hg or requiring antihypertensive drugs. 6. Renal 6.1. Estimated or measured glomerularfiltration rate <SO%: By any locally used method. 6.2. Proteinuria of >0.5 gm/24 hours: According to locally determined method. 6.3. End-stage renalfailure: Failure of native kidneys for >3 months regardless of subsequent dialysis or transplantation. 7. Gastrointestinal 7.1. Gut infarction: Infarction or resection of bowel below duodenum or of gall bladder, spleen, or liver: ever, since the onset of vasculitis. 7.2. Mesenteric insufficiencyipancreatitis: Typical abdominal pain confirmed on angiography or enzyme changes. 7.3. Chronic peritonitis: Typical abdominal pain and peritoneal irritation on clinical examination. 7.4. Esophageal stricture or upper gastrointestinal (GI) tract surgery: Documented endoscopically or radiologically: GI surgery ever, since the onset of vasculitis. 8. Peripheral vascular 8.1. Absent peripheralpulse: In 1 limb, detected clinically. 8.2. Second episode absent peripheral pulse: In 1 limb, detected clinically, at least 3 months after first event. 8.3. Ab.sent peripheral pulses: In at least 2 limbs, detected clinically. 8.4. Major vessel stenosis: Such as carotid or renal stenosis, documented on Doppler echocardiography or angiography. 8.5. Claudication: Exercise-related ischemic pain in peripheral large vessel present for at least 3 months. 8.6. Complicated venous thrombosis: With persistent swelling, ulceration, or clinical evidence of venous stasis. 8.7. Minor tissue loss: Such as loss of fingertip pulp space; ever, since the onset of vasculitis. 8.8. Major tissue loss: Such as the loss of digit(s) or limb(s), including by surgical resection; ever, since the onset of vasculitis. 8.9. Second episode of major tissue loss: At least 3 months after first event. 9. Ocular 9.1. Cataract: A lens opacity (cataract) in either eye, documented by ophthalmoscopy. 9.2. Retinal change: Any significant change, documented by ophthalmoscopic examination: may result in field defect, legal blindness. 9.3. Optic atrophy: Documented by ophthalmoscopic examination. 9.4. Visual impairment/diplopia: Restricted eye movements (not due to nerve palsies), reduced visual acuity, double vision, or tunnel vision. 9.5. Blindness: Complete loss of vision in 1 eye. 9.6. Blindness in second eye: At least 3 months after first event. 9.7. Orbital wall destruction: Determined by plain radiographs or computed tomography scan. 10. Neuropsychiatric 10.1. Cognitive impairment: Memory deficit, difficulty with calculation, poor concentration, difficulty in spoken or written language, impaired performance level, as documented on clinical examination (e.g., short mental test score) or by formal neurocognitive testing. 10.2. Major psychosis: Altered ability to function in normal activity due to psychiatric reasons. Severe disturbance of the perception of reality characterized by the following features: delusions, hallucinations (auditory, visual), incoherence, marked loose associations, impoverished thought content, marked illogical thinking, bizarre, disorganized, or catatonic behavior. 10.3. Seizures: Paroxysmal electrical discharge occurring in the brain and producing characteristic physical changes including tonic and clonic movements, and certain behavioral disorders. Only seizures requiring therapy for more than 3 months are counted as damage. 10.4. Cerebrovascular accident: Resulting in focal findings such as paresis, weakness, etc., or surgical resection for causes other than malignancy; ever, since the onset of vasculitis. 10.5. Second cerebrovascular accident: At least 3 months after first event. 10.6. Cranial nerve lesion: Cranial neuropathy, excluding optic newe or sensorineural deafness. 10.7. Peripheral neuropathy: Resulting in either motor or sensory dysfunction. 10.8. Transverse mye1iti.s: Lower extremity weakness or sensory loss with loss of sphincter control (rectal and urinary bladder). 11. Other damage 11.1. Premature gonadal failure: Onset of menopause before the age of 40 years. 11.2. Marrow failure: Leukopenia (white blood cells <4,000 per microliter) or thrombocytopenia (platelets <140) or anemia (hemoglobin <lo), preferably confirmed by marrow aspiration. 11.3. Diabetes mellitus: Requiring any type of therapy. 11.4. Chronic chemical cystitis: Persistent hematuria or shrunken bladder. This does not include acute hemorrhagic cystitis, which should be scored in the adverse drug reactions. 11.5. Malignancy: Documented by pathology, excluding dysplasias. 11.6. Other features: Any feature considered by patient or doctor to be an important scar or consequence which has arisen since onset of disease.