Reliability of the disk damage likelihood scale☆

Design 

This study was done as an observational case series.

Methods 

A retrospective review of the records of the Glaucoma Service Diagnostic Laboratory at the Wills Eye Hospital was performed to identify patients who had 35 mm stereo optic disk photographs and a Swedish Interactive Thresholding Algorithm (SITA) Standard or FASTPAC visual field examination performed within 6 months of each other. Patients were placed into one of four levels of visual field damage: no damage, mild, moderate, and severe. The criteria were as follows (adapted from Hodapp and associates24):

Records were excluded if stereo photographs of the optic disk were not available or the photographs were too blurred to permit interpretation.

Using a stand-mounted stereo viewer (Luminos Photo Corp., Yonkers, New York, USA) and light box, two glaucoma specialists staged the optic nerves according to the DDLS (Table 1) and Armaly cup/disk ratio. Forty-eight stereo pairs of optic nerve photos were selected according to the inclusion criteria and were presented to each masked investigator in a random fashion. A 1-cm crosshair, with a 1-mm scale, was taped to the light box and used to measure the size of the cup and disk by superimposing the left photo of the stereo pair. The quotient was used to determine a measured cup/disk ratio. For purposes of the DDLS, a 1.75-mm nerve diameter was assumed for all nerves (Table 1).25 Graders were asked to note the vertical, horizontal, and overall cup/disk ratio, the vertical, horizontal and overall DDLS stage, and the vertical and horizontal cup/disk ratio using the ruler. Each photograph was examined three times by each grader.

TABLE 1. The Disk Damage Likelihood Scale (DDLS)

	Stage	Significance	True Vertical Disk Diameter (mm)
			<1.5	1.5–2.0	>2.0
			Narrowest width of rim
	0a	No damage	0.5	>0.39	>0.29
	0b	Probably no damage	0.4–0.49	0.3–0.39	0.2–0.29
	1	Possible damage	0.3–0.39	0.2–0.29	0.1–0.19
	2	Probable damage	0.2–0.29	0.1–0.19	0.05–0.09
	3	Definite damage	0.1–0.19	0.01–0.1	0.01–0.05
	4	Definite damage	0.01–0.1	No rim <45°	No rim <45°
	5	Definite damage	No rim <45°	No rim 45–90°	No rim 45–90°
	6	Definite damage	No rim 45–90°	No rim 91–180°	No rim 91–180°
	7	Definite damage	No rim >90°	No rim >180°	No rim >180°

Statistical analysis was performed by SPSS for Windows version 10.1 (SPSS Inc., Chicago, Illinois, USA). Intraobserver and interobserver agreement was determined by the test–retest method for the DDLS, the Armaly cup/disk ratio, and the measured cup/disk ratio. Statistical significance was determined by the Fisher exact test.

Three observers examined 34 eyes of 24 consecutive glaucoma clinic patients during one morning of clinical practice. All patients seen during that morning were included. Eyes were excluded if there was no view of the optic nerve. Each observer made a single determination of the cup/disk ratio and DDLS stage using a Haag–Streit 900 slit lamp and a Volk 66-diopter lens according to the technique of Lim and associates.26 Each observer was masked to the readings of the fellow observers. The number of interobserver agreements was tabulated. Observers were considered to be in agreement if the two measurements were ≤ 1 DDLS stage and ≤ 0.1 cup/disk ratio.

Results 

Fifteen patients had no visual field damage, 12 had mild damage, ten had moderate damage, and 11 had severe damage. The intraobserver and interobserver agreement for selected disk parameters is given in Table 2. A summary of mean levels of agreement among the three trials for the remaining parameters is given in Table 3. Interobserver and intraobserver agreement for the DDLS is greater than the Armaly cup/disk ratio and the measured cup/disk ratio.

TABLE 2. The Interobserver and Intraobserver Agreement (≤0.1 c/d or ≤ one Stage) for Selected Disk Measurements Using Stereo Optic Disk Photographs

	Interobserver	Field Damage	Armaly Vertical c/d Ratio			Overall DDLS Stage
			Reading 1	Reading 2	Reading 3	Reading 1	Reading 2	Reading 3
	Grader 1 + 2	Normal	9/15	11/15	12/15	13/15	14/15	13/15
		Mild	9/12	9/12	8/12	10/12	10/12	10/12
		Moderate	7/10	7/10	4/10	10/10	10/10	10/10
		Severe	7/11	6/11	8/11	8/11	8/11	9/11
		Total	32/48	33/48	32/48	41/48	42/48	42/48
		P value*	0.9	0.7	0.2	0.4	0.2	0.6
	Intraobserver		Reading 1–Reading 2	Reading 2–Reading 3	Reading 1–Reading 3	Reading 1–Reading 2	Reading 2–Reading 3	Reading 1–Reading 3
	Grader 1	Normal	13/15	14/15	13/15	15/15	15/15	15/15
	Grader 2	Normal	14/15	13/15	14/15	15/15	15/15	15/15
	Grader 1	Mild	9/12	10/12	10/12	12/12	12/12	12/12
	Grader 2	Mild	11/12	12/12	11/12	12/12	11/12	11/12
	Grader 1	Moderate	10/10	8/10	8/10	10/10	10/10	10/10
	Grader 2	Moderate	8/10	10/10	10/10	10/10	10/10	10/10
	Grader 1	Severe	11/11	11/11	11/11	10/11	10/11	10/11
	Grader 2	Severe	9/11	10/11	11/11	11/11	11/11	11/11
	Grader 1	Total	43/48	43/48	42/48	47/48	47/48	47/48
			P = .2*	P = .4*	P = .5*	P = .3*	P = .3*	P = .3*
	Grader 2	Total	42/48	45/48	46/48	48/48	47/48	47/48
			P = .7*	P = .4*	P = .6*	P = N/A	P = .4*	P = .4*

c/d = cup/disk; DDLS = Disk Damage Likelihood Scale.

* Fisher exact test.

TABLE 3. A Summary of the Mean Interobserver and Intraobserver Agreement (≤1 DDLS Stage or ≤0.1 c/d Ratio) for the Remaining Disk Criteria (Percent Agreement and SD) Using Stereo Optic Disk Photographs

		Interobserver	Intraobserver
			Grader 1	Grader 2
	Horizontal c/d ratio	69 (0.02)	89 (0.07)	92 (0)
	Vertical c/d ratio	68 (0.02)	89 (0.01)	95 (0.05)
	Overall c/d ratio	74 (0.04)	90 (0.04)	90 (0.06)
	DDLS horizontal stage	84 (0.03)	97 (0.01)	99 (0.01)
	DDLS vertical stage	85 (0.01)	97 (0.01)	99 (0.01)
	DDLS overall stage	85 (0.02)	99 (0.01)	99 (0.01)
	Measured horizontal c/d ratio	65 (0.04)	79 (0.08)	93 (0.01)
	Measured vertical c/d ratio	73 (0.1)	80 (0.05)	89 (0.05)
	Armaly horizontal c/d ratio vs measured horizontal c/d ratio		56 (0.14)	60 (0.02)
	Armaly vertical c/d ratio vs measured vertical c/d ratio		72 (0.06)	77 (0.06)

c/d = cup/disk; DDLS = Disk Damge Likelihood Scale; SD = standard deviation.

Of the 34 eyes, 4 were a stage 0a, 2 were at stage 0b, 7 were at stage 1, 10 were at stage 2, 5 were at stage 3, none was at stage 4, 1 was at stage 5, 2 were at stage 6 and 3 were at stage 7, as determined by one of us (G.L.S.). For both the DDLS and the cup/disk ratio determination, all three observers were in agreement to a similar amount (Table 4). The number of times two of the three observers were in agreement was similar using both systems.

TABLE 4. Level of Interobserver Agreement for In Vivo Patient Measurements Using the DDLS Stage and c/d Ratio(≤1 DDLS Stage and ≤0.1 c/d Ratio) for the Three Observers

	DDLS Stage	Agreement of All Three Observers (%) for DDLS Stage	Agreement Between Two of Three Observers (%) for DDLS Stage	Agreement of All Three Observers (%) for Armaly c/d Ratio	Agreement Between Two of Three Observers (%) for Armaly c/d Ratio
	0a	4/4 (100)	4/4 (100)	4/4 (100)	4/4 (100)
	0b	1/2 (50)	2/2 (100)	2/2 (100)	2/2 (100)
	1	5/7 (71)	7/7 (100)	4/6 (67)	6/6 (100)
	2	8/10 (80)	10/10 (100)	7/10 (70)	10/10 (100)
	3	2/5 (40)	5/5 (100)	2/5 (40)	4/5 (80)
	5	1/1 (100)	1/1 (100)	1/2 (50)	2/2 (100)
	6	2/2 (100)	2/2 (100)	1/2 (50)	2/2 (100)
	7	2/3 (67)	3/3 (100)	2/3 (67)	3/3 (100)
	Total	25/34 (74)	34/34 (100)	23/34 (68)	33/34 (97)
	P value*	.5	N/A	.6	.5

c/d = cup/disk; DDLS = Disk Damage Likelihood Scale.

* Fisher exact test.

We also measured the level of agreement for the in vivo patients a different way, by lumping groups together. When the stages 0a to 3 and 4 to 7 are combined to create two groups of lesser and greater damage, there was no difference in the level of agreement between the groups. For the DDLS, 100% of the time two thirds of the observers agreed for both stages 0a to 3 and stages 4 to 7, 78% of the time 3/3 observers agreed for stages 0a to 3 and 64% of the time 3/3 observers agreed for stages 4 to 7 (P = .4, Fisher exact test). For the Armaly cup/disk ratio, 100% of the time two thirds of the observers agreed for stages 0a to 3 while 92% of the time two thirds agreed for stages 4 to 7 (P = .4, Fisher exact test). Seventy-seven percent of the time 3/3 observers agreed for the cup/disk ratio for stages 0a to 3 and 50% of the time for stages 4 to 7 (P = .1, Fisher exact test).

Discussion 

An examination of the optic nerve is a crucial part of an evaluation for glaucoma. Traditionally, physicians have used the Armaly cup/disk ratio to describe the disk. This is a useful technique, but it is limited by issues of focal rim narrowing and physiologic variability based on disk size. The new DDLS system described here directs the attention of the clinician to the optic disk size and to the region of the optic nerve that is actually being damaged, the neuroretinal rim, by using a scale of the narrowest rim/disk ratio that simultaneously accounts for the overall size of the optic nerve. Such a scale is readily usable, communicable, and overcomes two of the problems (a narrow neuroretinal rim that may not contribute to overall cup/disk ratio measurements and the problem of physiologic cupping) inherent in the Armaly cup/disk ratio.

The DDLS has limitations. It is not easy to apply to tilted optic nerves or to those with sloped temporal rims, who may have an advanced DDLS stage yet have less severe disease as measured by visual field criteria. It does not address the case in which new narrowing of the rim develops in a second area, while already existing damage in a different area remains unchanged.

This study demonstrates that the DDLS is reliable using both disk photos and in vivo patient measurements. Additionally, there is no proof that the scale does not “overcall” optic nerves that may have physiologically narrow rims (myopic tilted nerves and normal nerves with physiologically narrow temporal rims but thick superior and inferior rims) or is useful to detect change over time.

This study is limited by the fact that we did not know the disk size of the photographs that we viewed and we assumed a 1.75 mm size. We believe this is not a problem for a study of intraobserver and interobserver agreement, although it may be a significant limitation as we seek to correlate disk and field damage using this scale.

In conclusion, when examining stereo optic disk photographs, the interobserver and intraobserver agreement for the DDLS is greater than both the Armaly cup/disk system and the optic nerve measurement system. For in vivo patient measurements, the agreement between the DDLS and the Armaly cup/disk ratio is the same.