Original Paper Ophthalmologica 2008;222:408–413 DOI: 10.1159/000161555 Received: April 19, 2007 Accepted: April 20, 2007 Published online: October 10, 2008 Interobserver Variability of the Würzburg Bleb Classification Score Thomas Klink Sybille Schrey Uta Elsesser Janine Klink Günther Schlunck Franz Grehn Department of Ophthalmology, Julius Maximilian University, Würzburg, Germany Abstract Background: The Würzburg bleb classification score (WBCS) serves to assess filtering blebs in a standardized fashion. The purpose of this prospective masked agreement study was to evaluate the WBCS interobserver variability. Methods: The WBCS provides a scheme to grade clinical bleb morphology. It evaluates the following parameters: vascularity, corkscrew vessels, encapsulation, microcysts and bleb height. Thus, 113 eyes of 104 consecutive patients at various times after surgery were examined (slit lamp biomicroscopy) by 3 ophthalmologists with each observer being unaware of the findings reported by the others. To calculate the interobserver variability of the WBCS, the interobserver consistency and absolute agreement were determined with an intraclass correlation coefficient (ICC) using a 2-way random model. Results: The ICC values of a single rater’s judgment were: vascularity +0.62, corkscrew vessels +0.67, encapsulation +0.63, bleb height +0.53, microcysts +0.52 and total score +0.74. The ICC values of the mean of all 3 raters were: +0.83 vascularity, +0.86 corkscrew vessels, +0.84 encapsulation, +0.77 bleb height, +0.76 microcysts and +0.90 total score. Conclusion: The WBCS is a bleb morphology score with high levels of interobserver consistency and absolute agreement in clinical practice. Copyright © 2008 S. Karger AG, Basel © 2008 S. Karger AG, Basel 0030–3755/08/2226–0408$24.50/0 Fax +41 61 306 12 34 E-Mail email@example.com www.karger.com Accessible online at: www.karger.com/oph Introduction Scarring of the filtering bleb is the main reason for failure of penetrating glaucoma surgery. Recently, several wound healing modulating agents have entered routine clinical practice [1, 2]. In order to guide their use, a thorough slit lamp examination is essential and allows for a timely start of wound-modulating therapy [2–8]. In clinical practice, scoring systems offer a framework for structured patient examination and facilitate the interpretation of clinical findings. The Würzburg bleb classification score  (WBCS) was designed to allow for a well-structured assessment of filtering bleb morphology. Since 1998, the WBCS is regularly used in experimental investigations and clinical routine, especially as a basis for intensified postoperative care [1, 9–13]. Recently published grading systems for filtering blebs use criteria similar to the WBCS [2, 3, 8]. Here we analyze the interobserver variability of the WBCS as a prerequisite for its consistent use in clinical routine. Materials and Methods From July to October 2004, 113 eyes of 104 consecutive patients, who had undergone primary fornix-based trabeculectomy for medically uncontrolled primary or secondary open-angle glaucoma or chronic angle-closure glaucoma, were enrolled in a prospective masked agreement study at the Würzburg University Eye Hospital. The patients had undergone surgery between 2 days Dr. Thomas Klink Universitäts-Augenklinik Würzburg Josef-Schneider-Strasse 11, DE–97080 Würzburg (Germany) Tel. +49 931 201 20610, Fax +49 931 201 20490 E-Mail firstname.lastname@example.org Downloaded by: Siriraj Medical Library, Mahidol University 22.214.171.124 - 1/27/2016 6:18:16 AM Key Words Bleb grading ⴢ Glaucoma surgery ⴢ Filtering bleb ⴢ Trabeculectomy and 10 years prior to the examination (median 67 days). The study was approved by the local ethics committee, and informed consent was obtained from all participants. All eyes were examined once (slit lamp biomicroscopy) by 3 ophthalmologists at different clinical training levels (F.G., T.K., S.S.) with each observer being unaware of the findings reported by the others. A fourth person (U.E.) selected the patients and dispensed and collected the data entry forms. The WBCS evaluates vascularity, corkscrew vessels, encapsulation, microcysts and bleb height (table 1). Most parameters are scored from 0 to 3, as compared to standard photographs (fig. 1). The presence of microcysts is scored from 0 to 3 according to thirds of the bleb area bearing microcysts (fig. 2) and bleb height is estimated as multiples of corneal thickness [5, 10, 11]. Finally, the total score was calculated (vascularity, corkscrew vessels, encapsulation and microcysts, examples of scoring: fig. 3, 4). After recording their bleb evaluation, the observers received additional patient data [intraocular pressure (IOP), diagnosis and time of follow-up]. Based on the filtering bleb evaluation and the additional patient data, the investigators noted their therapeutic decisions from the options available: no therapy; increasing dose of steroids; 5-fluorouracil injections; antiglaucomatous medication; needling the filtering bleb; a new surgical approach. Statistical analysis was performed using SPSS for Windows statistical software (version 13.0, SPSS, Chicago, Ill., USA). An analysis was performed to determine the consistency and absolute agreement between the observers concerning bleb morphology, in order to reflect the clinical reproducibility of the WBCS. For bleb morphology, the consistency and absolute agreement of a single rater’s judgment and of the mean of all 3 raters were calculated with the intraclass correlation coefficient (ICC) using a 2-way random model . The single rater’s judgment is a singlemeasure ICC value, which gives information about the expected consistency and absolute agreement between single users of the WBCS. The mean of all 3 raters is an average-measure ICC value, which gives the expected consistency and absolute agreement between groups of 3 investigators using the WBCS. Levels of agreement obtained with the ICC statistics of 1 0.81 were defined as excellent agreement, levels between 0.61 and 0.80 were defined as good agreement, levels between 0.41 and 0.60 were defined as moderate agreement, levels between 0.21 and 0.4 were defined as fair agreement and levels less than 0.21 were defined as poor agreement. The statistical analysis of therapeutic decisions was performed with cross-classified tables ( values). Table 1. Parameters and scoring of the WBCS Parameters Scoring Vascularity 3 = avascular 2 = similar to adjacent conjunctiva 1 = increased 0 = massive 3 = none 2 = in one third 1 = in two thirds 0 = entire bleb 3 = none 2 = in one third 1 = in two thirds 0 = entire bleb 3 = entire bleb 2 = lateral or medial of the flap 1 = over the scleral flap 0 = none multiples of corneal thickness Corkscrew vessels Encapsulation Microcysts Bleb height ic angle-closure glaucoma. Intraoperatively, 8 eyes had received no antimetabolites, 11 had received 5-fluorouracil and 94 had received mitomycin C. Grading of Bleb Morphology For all eyes, the consistency and absolute agreement of a single rater’s judgment and of the mean of all 3 raters were calculated with the ICC (established by Landis ) using a 2-way random model. With slit lamp examination, the consistency and absolute agreement of a single rater’s judgment were: vascularity +0.62/+0.62, corkscrew vessels +0.68/+0.67, encapsulation +0.63/+0.63, bleb height +0.56/+0.53, microcysts +0.53/+0.52 and total score +0.74/+0.74 (table 2). The means of all raters were in this setting: +0.83/+0.83 for vascularity, +0.87/+0.86 for corkscrew vessels, +0.84/+0.84 for encapsulation, +0.80/+0.77 for bleb height, +0.77/+0.76 for microcysts and +0.90/+0.90 for total score (table 2). In total, 113 eyes of 104 consecutive patients (63 male, 41 female) were examined. Both eyes of 5 male and 4 female patients were included. The mean interval between surgery (trabeculectomy) and bleb grading was 334.4 8 637.6 days (range 2–3,516 days, median 67 days). The mean age of all patients was 62.3 8 13.06 years. Overall, 74 eyes suffered from primary open-angle glaucoma, 35 from secondary open-angle glaucoma and 4 from chron- Therapeutic Decisions The assessment of therapeutic decisions showed an agreement of all 3 observers in 88 (78%) eyes. In all other cases, at least 2 observers decided for the same therapy. There was no case of total disagreement among all 3 graders. In order to compare the therapeutic decisions among the single raters we used cross-classified tables with values. The values indicate moderate (0.556) to good (0.662 and 0.706) agreements (table 3). Würzburg Bleb Classification Score Ophthalmologica 2008;222:408–413 409 Downloaded by: Siriraj Medical Library, Mahidol University 126.96.36.199 - 1/27/2016 6:18:16 AM Results Vascularity avascular similar to adjacent conjunctiva increased massive Microcysts Corkscrew vessels in one third in two thirds entire bleb under high magniﬁcation Encapsulation Fig. 1. Standard photographs of the WBCS. in one third entire bleb Discussion B B Fig. 2. Diagram of where to search for the presence of microcysts according to thirds of the bleb area bearing cysts (scored 0–3; see table 1). Early after the operation, microcysts appear above the scleral flap (A) and are found nasally and temporally of the scleral flap during follow-up (B). 410 Ophthalmologica 2008;222:408–413 Klink/Schrey/Elsesser/Klink/Schlunck/ Grehn Downloaded by: Siriraj Medical Library, Mahidol University 188.8.131.52 - 1/27/2016 6:18:16 AM A Adequate healing of the filtering bleb area is the key to successful penetrating glaucoma surgery. Therefore, early detection of bleb scarring is essential as it guides the individualized application of antimetabolites . With slit lamp examination as the mainstay of decisive diagnostics, scoring systems provide guidelines for a wellstructured observation of clinical findings and their proper representation on record. This is of particular interest when different observers monitor the same patient, and benefits less-seasoned examiners in particular. The idea of the Würzburg bleb classification score (WBCS) was to establish an objective and reproducible grading system . To date, several bleb classification schemes have been proposed. The Indiana Bleb Appearance Grading Scale (IBAGS)  and the bleb classification according to Sacu et al.  are grading systems for filtering blebs, which used criteria similar to the WBCS. The most recent bleb grading system is the Moorfields Bleb Grad- Fig. 3. Filtering bleb 2 weeks after trabeculectomy, overfiltration with a total bleb score of 4 (vascularity: 0, corkscrew vessels: 0, encapsulation: 3, microcysts: 1). Fig. 4. Filtering bleb 4 weeks after trabeculectomy, working well with a total bleb score of 11 (vascularity: 2, corkscrew vessels: 3, encapsulation: 3, microcysts: 3). Table 2. Slit-lamp examination: interobserver consistency and absolute agreement for all eyes (n = 113) Measures Consistency/absolute agreement Vasc. CSV Encap. Micro. BH TS Single rater’s judgment consistency absolute agreement consistency absolute agreement 0.62 0.62 0.83 0.83 0.68 0.67 0.87 0.86 0.63 0.63 0.84 0.84 0.74 0.74 0.90 0.90 Mean of all 3 raters 0.53 0.52 0.77 0.76 0.56 0.53 0.80 0.77 Consistency and absolute agreement were calculated with the intraclass correlation coefficient using a 2-way random model for a single rater’s judgement and the mean of all 3 raters. Vasc. = Vascularisation; CSV = corkscrew vessels; Encap. = encapsulation; Micro. = microcysts; BH = bleb height; TS = total score. ing System (MBGS), which goes back to a telemedicine study [4, 8]. Sacu et al.  used the WBCS as a basis for their score, and expanded it by the parameters dimension of avascularity, conjunctival transparency, hemorrhage and thin-cystic avascular configuration. In a prospective clinical trial, Sacu et al.  showed that the detection of corkscrew vessels and microcysts in the first 2 weeks after surgery may serve as prognostic indicators of subsequent IOP development and outcome of surgery. Corkscrew vessels were associated with an increased IOP and microcysts with a reduced IOP after 12 months of follow-up . All other parameters did not correlate with long-term outcome. The IBAGS evaluates 4 parameters (bleb height, horizontal extent, vascularity and Seidel test), and its reproducibility was established using filtering bleb photo- graphs . The IBAGS shows a high consistency and absolute agreement between different observers. Unfortunately, the first report does not specify if the single or average ICC values were provided . The single ICC val- Würzburg Bleb Classification Score Ophthalmologica 2008;222:408–413 Table 3. Agreement in therapeutic decisions among the single Measure of the agreement () Absolute number of agreement (%) Grader A-B Grader A-C Grader B-C 0.706 96 (85) 0.662 90 (80) 0.556 83 (73) 411 Downloaded by: Siriraj Medical Library, Mahidol University 184.108.40.206 - 1/27/2016 6:18:16 AM graders using cross-classified tables ( values; n = 113 eyes; percentage in parentheses) References 412 ment. Both, the IBAGS and MBGS do not consider microcysts as these scores were designed to evaluate bleb photographs, which often fail to depict microcysts, even if present clinically. However, the work of Sacu et al.  indicates a relevant positive predictive value for microcysts, which makes their inclusion in a bleb score desirable. The consistency and absolute agreement of bleb height as scored by the WBCS was comparable to the IBAGS and slightly inferior to the MBGS . Consistency and absolute agreement of all other WBCS parameters were good (single ICC) or excellent (average ICC). The WBCS registers bleb encapsulation rather than bleb size (MBGS) or bleb extent (IBAGS), since indications for a scarring demarcation may deserve particular attention. ICC values for encapsulation were similar or higher than for the extension measures depicted in the IBAGS . With a similar rationale, the WBCS scores the presence of corkscrew vessels, which are easy to detect and which gave a good ICC in our study. The average measure ICC values for vascularity were comparable for all 3 bleb-grading systems . We chose a cross-sectional study design to include filtering blebs at various stages of conjunctival wound healing. To assess the absolute agreement of therapeutic decisions after WBCS grading and evaluation of additional clinical data, we analyzed the values in cross-classified tables comparing all 3 investigators. A complete agreement of all 3 examiners in 78% appears quite reasonable. Other bleb scoring schemes have not yet been analyzed in this regard. In summary, our data indicate that the WBCS is feasible for standardized filtration bleb analysis with satisfactory interobserver variability. The WBCS might be of particular value when postoperative care of a single patient is provided by different ophthalmologists, as it aids consistent evaluation and documentation of the clinical findings. 1 Marquardt D, Lieb WE, Grehn F: Intensified postoperative care versus conventional follow-up: a retrospective long-term analysis of 177 trabeculectomies. Graefes Arch Clin Exp Ophthalmol 2004;242:106–113. 2 Sacu S, Rainer G, Findl O, Georgopoulos M, Vass C: Correlation between the early morphological appearance of filtering blebs and outcome of trabeculectomy with mitomycin C. J Glaucoma 2003;12:430–435. Ophthalmologica 2008;222:408–413 3 Cantor LB, Mantravadi A, WuDunn D, Swamynathan K, Cortes A: Morphologic classification of filtering blebs after glaucoma filtration surgery: the Indiana Bleb Appearance Grading Scale. J Glaucoma 2003; 12: 266– 271. 4 Crowston JG, Kirwan JF, Wells A, Kennedy C, Murdoch IE: Evaluating clinical signs in trabeculectomized eyes. Eye 2004; 18: 299– 303. Klink/Schrey/Elsesser/Klink/Schlunck/ Grehn Downloaded by: Siriraj Medical Library, Mahidol University 220.127.116.11 - 1/27/2016 6:18:16 AM ues appear of higher value, as they reflect the clinical situation of a single observer investigating a bleb at a time, rather than a group of observers. Recently, the IBAGS was compared to the MBGS and also used in a prospective clinical trial to evaluate the bleb development after trabeculectomy with hyaluronate application [15, 16]. In the comparative report, IBAGS parameters vascularity and bleb height were of moderate (single measure ICC) to good (average measure ICC) agreement, whereas agreement for bleb extent was only poor, in contrast to an earlier report [3, 16]. The prospective study  found a significant effect of hyaluronate on horizontal bleb extent only. Interestingly, this parameter was reported as the most variable in the comparative study . Thus, a significant hyaluronate effect on bleb extent appears to be particularly strong. The MBGS was conceived as a research tool for studies of glaucoma surgery and is also based on photographs . The included parameters are area (2 subsections), bleb height and vascularity (3 subsections), in order to describe mixed bleb morphology particularly well. The MBGS initially used a grading range from 1 to 10, which was reduced to 1–5 in the comparison between IBAGS and MBGS to make it more practical [8, 16]. The WBCS was the first standardized bleb grading system used in a prospective clinical study to evaluate the development of filtering blebs . Since then, several other studies employed the WBCS to grade bleb morphology [5, 9, 12, 13]. The score was used to guide treatment decisions in postoperative follow-up, and to give a prognosis of bleb function and surgical outcome [9, 10]. The WBCS grades each bleb characteristic from 0 to 3, which might give less fidelity compared to the MBGS, for example, but has advantages in clinical routine for its simplicity. Our data reveal that not all WBCS parameters were unequivocally graded: microcysts and bleb height were detected with moderate consistency and absolute agree- Würzburg Bleb Classification Score 10 Klink J, Schmitz B, Lieb WE, Klink T, Grein HJ, Sold-Darseff J, Heinold A, Grehn F: Filtering bleb function after clear cornea phacoemulsification: a prospective study. Br J Ophthalmol 2005;89:597–601. 11 Klink T, Guthoff R, Grehn F, Schlunck G: Postoperative care after glaucoma filtration surgery (in German). Ophthalmologe 2006; 103:815–823, quiz 824–825. 12 Picht G, Welge-Luessen U, Grehn F, LutjenDrecoll E: Transforming growth factor beta 2 levels in the aqueous humor in different types of glaucoma and the relation to filtering bleb development. Graefes Arch Clin Exp Ophthalmol 2001;239:199–207. 13 Wimmer I, Welge-Luessen U, Picht G, Grehn F: Influence of argon laser trabeculoplasty on transforming growth factor-beta 2 concentration and bleb scarring following trabeculectomy. Graefes Arch Clin Exp Ophthalmol 2003;241:631–636. 14 Landis JR, Koch GG: The measurement of observer agreement for categorical data. Biometrics 1977;33:159–174. 15 Lopes JF, Moster MR, Wilson RP, Altangerel U, Alvim HS, Tong MG, Fontanarosa J, Steinmann WC: Subconjunctival sodium hyaluronate 2.3% in trabeculectomy: a prospective randomized clinical trial. Ophthalmology 2006; 113:756–760. 16 Wells AP, Ashraff NN, Hall RC, Purdie G: Comparison of two clinical bleb grading systems. Ophthalmology 2006;113:77–83. Ophthalmologica 2008;222:408–413 413 Downloaded by: Siriraj Medical Library, Mahidol University 18.104.22.168 - 1/27/2016 6:18:16 AM 5 Picht G, Grehn F: Classification of filtering blebs in trabeculectomy: biomicroscopy and functionality. Curr Opin Ophthalmol 1998; 9:2–8. 6 Shingleton BJ: Management of the failing glaucoma filter. Ophthalmic Surg Lasers 1996;27:445–451. 7 Vesti E: Filtering blebs: follow-up of trabeculectomy. Ophthalmic Surg 1993; 24: 249– 255. 8 Wells AP, Crowston JG, Marks J, Kirwan JF, Smith G, Clarke JC, Shah R, Vieira J, Bunce C, Murdoch I, Khaw PT: A pilot study of a system for grading of drainage blebs after glaucoma surgery. J Glaucoma 2004;13:454– 460. 9 Guthoff R, Klink T, Schlunck G, Grehn F: In vivo confocal microscopy of failing and functioning filtering blebs: results and clinical correlations. J Glaucoma 2006; 15: 552– 558.