11.2 Automated Static Perimetry: HFA Strategies (SITA Standard, SITA Fast), Reliability Indices & Statpac Analysis

Automated Static Perimetry: HFA Strategies, Reliability Indices & Statpac Analysis

Core Clinical Mandate: Automated static perimetry is the clinical gold standard for the diagnosis, functional staging, and structural-functional correlation of glaucomatous optic neuropathy and neuro-ophthalmic pathways. The Certified Ophthalmic Medical Technologist (COMT) must possess mastery over the physical principles of bowl perimetry, the psychophysical algorithms governing threshold determination (SITA Standard, SITA Fast, SITA Faster), the critical thresholds of perimetric reliability, and the statistical deconstruction of Statpac single field printouts.

Physical Foundations & Optical Calibration of the Humphrey Field Analyzer

The Humphrey Field Analyzer (HFA) operates on the principles of static automated perimetry (SAP), wherein stationary light stimuli of variable luminance are projected onto a uniformly illuminated hemispherical bowl.

1. Bowl Dimensions & Background Luminance Calibration

• Bowl Radius: Standardized at 30 cm, establishing a fixed optical vergence requiring appropriate near refractive correction in the trial lens holder.
• Background Luminance: Calibrated strictly to 31.5 apostilbs (asb), equivalent to 10 candelas per square meter (cd/m²).
• Physiological Rationale (Weber's Law): A background luminance of 31.5 asb positions the patient's retinal adaptation at the lower end of the photopic (light-adapted) to upper mesopic range. Within this physiological zone, the retina operates under Weber's law of contrast sensitivity:

$\frac{\Delta I}{I} = \text{constant}$

Where $\Delta I$ is the differential luminance threshold of the target and $I$ is the background luminance. In this range, minor physiological fluctuations in pupil diameter or ambient illumination do not alter the differential threshold ratio, ensuring test reproducibility.

2. Stimulus Parameters: Duration & Goldmann Sizes

• Stimulus Exposure Duration: Fixed at 200 milliseconds (0.20 seconds). This duration is shorter than the human saccadic reaction latency (~250 ms). If a stimulus were presented for $\ge 250\text{ ms}$, the patient could initiate a refixation saccade to align their fovea with the peripheral target, invalidating peripheral threshold testing.
• Stimulus Dimensions (Goldmann Size Hierarchy):

Size III vs. Size V Clinical Transition: In advanced glaucoma where extensive absolute scotomas encompass the central 24° on a standard Size III stimulus, switching to a Goldmann Size V stimulus allows perimetrists to detect residual functioning retinal ganglion cell receptive fields that fall below the spatial summation threshold of Size III. Each step increase in Goldmann Roman numeral quadruples the target surface area ($4\times$).

3. The Decibel (dB) Photometric Scale

The HFA measures threshold sensitivity in decibels (dB), a logarithmic attenuation scale inversely related to physical stimulus brightness:

$\text{dB} = 10 \cdot \log_{10}\left(\frac{I_{\max}}{I}\right)$

• Maximum Stimulus Intensity (0 dB): Unattenuated projector output equal to 10,000 apostilbs (asb) (3,183 cd/m²).
• Decibel Attenuation: Each 10 dB represents a 1.0 log unit (10-fold) attenuation of light intensity.
• 30 dB Stimulus: Attenuated by $10^3$ (1,000-fold), equivalent to 10 asb.
• 40 dB Stimulus: Attenuated by $10^4$ (10,000-fold), equivalent to 1 asb.
• 0 dB (Absolute Scotoma): The brightest light the machine can project (10,000 asb) is not seen by the patient.
• Normal Foveal Sensitivity: Typically ranges between 34 dB and 38 dB in healthy young observers.

Perimetric Testing Strategies: SITA Architecture

Historically, perimetry relied on the Full Threshold 4-2 staircase bracketing strategy, in which each retinal location was crossed twice with 4 dB and 2 dB steps, requiring 12 to 18 minutes per eye. The advent of the Swedish Interactive Threshold Algorithm (SITA) transformed visual field testing.

Evolution of Automated Perimetry Strategies:
Full Threshold (4-2 dB Staircase; 15 min/eye)
       │
       ▼
SITA Standard (Bayesian Likelihood Models; 5-7 min/eye)
       │
       ▼
SITA Fast (Wider Confidence Limits; 3-4 min/eye)
       │
       ▼
SITA Faster (No Blind Spot/FN Catch Trials; 2 min/eye)

1. SITA Standard

• Mathematical Engine: Utilizes dynamic Bayesian probability models based on extensive normative and glaucomatous visual field databases. The algorithm maintains a continuously updated probability distribution of threshold sensitivity for each test point.
• Cross-Information Processing: Information gained at one test location immediately updates the prior probability distributions of neighboring points based on known receptive field clustering.
• Real-Time Patient Pacing: Adjusts inter-stimulus intervals dynamically according to the individual patient's average reaction time.
• Clinical Profile: Reduces test time to 4 to 7 minutes per eye while achieving diagnostic reproducibility identical or superior to the original Full Threshold strategy.

2. SITA Fast

• Modifications: Utilizes wider confidence intervals for stopping rules and begins thresholding with fewer stimulus presentations per location.
• Clinical Profile: Cuts test time to 2.5 to 4 minutes per eye. However, it exhibits slightly higher test-retest variability and slightly less sensitivity for early, shallow glaucomatous scotomas.
• Indications: Pediatric patients, elderly individuals with poor endurance, physically debilitated patients, or those with severe fatigue artifacts on SITA Standard.

3. SITA Faster

• Modifications (Introduced on HFA3):
  • Eliminates False Negative Catch Trials: Recognizes that false negative trials waste perimetric time and cause fatigue; instead, assesses response variability from the primary data stream.
  • Eliminates Heijl-Krakau Blind Spot Checks: Relies entirely on real-time infrared gaze tracking rather than flashing stimuli into the blind spot.
  • Optimized Starting Thresholds: Initiates testing at age-corrected normal sensitivity rather than 4 dB above threshold.
• Clinical Profile: Reduces test time by another 30% (averaging 2 minutes per eye). Highly effective for high-volume clinics, though baseline shifts must be accounted for when transitioning patients from SITA Standard to SITA Faster on Guided Progression Analysis (GPA).

Perimetric Test Grid Patterns

Selecting the correct test grid is critical to mapping specific visual pathway pathology:

1. The 30-2 Grid

• Architecture: 76 test locations arranged in a 6° square grid covering the central 30° of the visual field.
• The '-2' Configuration: Test points are offset by 3° from the horizontal and vertical meridians. This design ensures points straddle the meridians (e.g., at coordinates $(3, 3), (3, -3)$), allowing clean detection of steps across the horizontal raphe and hemianopic vertical demarcations.
• Limitation: The peripheral ring between 24° and 30° frequently displays false defects from the trial lens rim, ptotic upper eyelids, or prominent facial brow arches.

2. The 24-2 Grid

• Architecture: 54 test locations spaced 6° apart. It completely eliminates the outermost ring of the 30-2 grid except on the nasal border, where two additional points extending to 30° are preserved.
• Clinical Mandate: Preserving nasal points out to 30° is critical because early glaucomatous nerve fiber bundle loss preferentially produces a nasal step across the horizontal raphe. Eliminating the superior and temporal outer points removes 90% of trial lens rim artifacts and eyelid droop artifacts.

3. The 24-2C Grid

• Architecture: Retains the 54 points of the 24-2 grid and adds 10 strategically placed central points within the central 10°.
• Anatomical Rationale: Glaucoma is now recognized to cause early, focal macular ganglion cell loss in up to 16% of patients who appear normal or borderline on traditional 24-2 grids. The 24-2C detects these central defects without requiring a separate 10-2 test.

4. The 10-2 Grid

• Architecture: 68 test points arranged in a dense 2° grid within the central 10° of fixation (straddling meridians by 1°).
• Indications:
  • Advanced end-stage glaucoma to monitor central foveal-sparing islands.
  • Glaucomatous defects splitting fixation.
  • Hydroxychloroquine (Plaquenil) and chloroquine retinal toxicity surveillance (essential for detecting parafoveal scotomas).
  • Maculopathies (e.g., Stargardt disease, macular telangiectasia type 2, central serous chorioretinopathy).

Perimetric Reliability Indices: Identification & Artifacts

Before interpreting any visual field scotoma, the technologist must rigorously audit the reliability indices located in the upper-left quadrant of the printout.

1. Fixation Losses (FL)

• Heijl-Krakau Blind Spot Technique: At the beginning of the test, the machine plots the coordinates of the patient's physiological blind spot (typically centered at 15° temporal and 1.5° inferior to fixation). Throughout testing, the instrument projects a suprathreshold (Size III, 4 dB brighter than expected, or 0 dB) stimulus directly into this blind spot.
• Criterion: If the patient presses the response button, the eye has drifted from fixation. A fixation loss rate > 20% is flagged with an 'x' on the printout and signals compromised reliability.
• False Fixation Losses: Occur if the blind spot was improperly located at the start of the test (e.g., patient tilted head, or blind spot expanded by peripapillary atrophy or optic disc edema).
• Real-Time Gaze Tracking: Displays an continuous infrared corneal-reflection tracing at the bottom of the printout:
  • Upward deflections: Represent gaze deviations (saccades away from the fixation target). The height of the deflection corresponds to the angular magnitude of the drift.
  • Downward deflections: Represent pupil occlusion (eyelid blinks, blepharoptosis, or head drooping).

2. False Positive (FP) Errors: The "Trigger-Happy" Patient

• Mechanism: The instrument pauses stimulus projection or activates the mechanical projection shutter sound without illuminating a light target. If the patient presses the buzzer, a false positive error is recorded.
• Diagnostic Threshold: False positive rates > 15% completely invalidate the visual field.
• Clinical Manifestation:
  • The patient anticipates the rhythmic presentation of targets ("trigger-happy").
  • Generates abnormally elevated raw threshold values, often exceeding 45 to 50 dB (well beyond physiological human photoreceptor capacity).
  • Creates an artificially elevated Mean Deviation (MD can become positive, e.g., $+3.00\text{ dB}$).
  • Generates the classic "white scotoma" artifact on Pattern Deviation plots, where the software overcompensates for the hyper-sensitive points and suppresses normal peripheral points.

3. False Negative (FN) Errors: Fatigue vs. Disease

• Mechanism: The instrument projects a stimulus significantly brighter (typically 9 dB brighter) than the previously established threshold at a location that has already demonstrated seeing ability.
• Diagnostic Threshold: A rate > 25% to 33% is flagged on the printout.
• Clinical Dual Etiology:
  1. Patient Fatigue / Inattention: If FN is high in an otherwise normal visual field, it reflects drowsiness, drifting attention, or cognitive lapse. This typically generates a "cloverleaf" visual field artifact, in which the initial central quadrants tested have normal sensitivities, while the peripheral points tested later in the session are completely missed.
  2. Severe Glaucomatous Damage: In eyes with moderate to advanced glaucomatous field loss, damaged retinal ganglion cells exhibit extreme biological threshold variability. A point may see a 12 dB stimulus on one pass and miss a 21 dB stimulus on the next pass. In an eye with an established dense arcuate defect, high FN is a marker of disease severity rather than poor testing effort.

4. Technical Artifacts

• Trial Lens Rim Artifact: Occurs when the trial lens is placed too far from the eye (>15 mm) or decentered, creating a dense ring of peripheral depression resembling a 360° ring scotoma. Always verify trial lens centering.
• Uncorrected Refractive Blur: Every 1.00 diopter of uncorrected spherical or cylindrical error causes approximately 1.25 dB of generalized sensitivity reduction.
• Ptosis / Dermatochalasis: Superior drooping eyelids create pseudo-superior depression. If suspected, the test must be repeated with the upper eyelid taped securely to the orbital rim.

Statpac Single Field Analysis & Probability Maps

Statpac is the proprietary expert statistical package integrated into the Humphrey Field Analyzer that compares the patient's individual thresholds against an age-corrected normative database.

1. Raw Numerical & Grey Scale Displays

• Raw Sensitivity Map: Shows the exact threshold measured in decibels (dB) at each retinal coordinate.
• Grey Scale Map: Converts numerical dB values into shaded pixel patterns (from clear white for >35 dB to solid black for 0 dB). Clinical Rule: Never base a diagnostic decision on the grey scale alone! It exaggerates shallow depressions and masks early focal defects.

2. Total Deviation (TD) vs. Pattern Deviation (PD)

The core diagnostic engine of Statpac lies in the side-by-side comparison of Total Deviation and Pattern Deviation probability maps:

Statpac Probability Map Comparison:
┌──────────────────────────────────────┐     ┌──────────────────────────────────────┐
│         TOTAL DEVIATION (TD)         │     │        PATTERN DEVIATION (PD)        │
│                                      │     │                                      │
│  Measures total loss relative to     │     │  Filters out generalized depression  │
│  age-matched normative database.     │     │  (cataract, miosis, corneal haze).   │
│  Depressed by:                       │     │  Adjusts by 85th percentile shift    │
│  - Cataract / Media Opacity          │     │  to isolate TRUE FOCAL DEFECTS.      │
│  - Uncorrected Refractive Error      │     │                                      │
│  - Glaucoma & Neurological Defects   │     │  *PRIMARY DIAGNOSTIC PLOT*           │
└──────────────────────────────────────┘     └──────────────────────────────────────┘

• Total Deviation (TD) Plot: Demonstrates the decibel difference between the patient's threshold at each point and the median value for age-matched normal individuals. Below the numerical grid, the probability plot assigns statistical significance symbols ($p < 5%$, $p < 2%$, $p < 1%$, $p < 0.5%$).
• Pattern Deviation (PD) Plot: The most critical plot in automated perimetry. Statpac evaluates the overall height of the visual field by analyzing the 85th percentile of sensitivity (the 7th highest sensitivity point). It then mathematically raises or lowers the entire field profile to correct for generalized depression (such as nuclear cataracts, posterior capsular opacification, corneal scars, or small pupils < 2.5 mm).
• Diagnostic Rule of Thumb:
  • If the TD plot is heavily depressed but the PD plot is completely clear, the patient has pure generalized depression (e.g., dense cataract), not localized glaucomatous damage.
  • If the PD plot shows a cluster of significant points ($p < 1%$ or $p < 0.5%$) respecting the horizontal raphe, the patient has true localized glaucomatous nerve fiber bundle loss, regardless of the appearance of the TD plot.

Global Indices & Longitudinal Progression Metrics

Statpac summarizes overall visual field performance into quantitative global indices:

1. Mean Deviation (MD)

• Definition: The weighted average of all points on the Total Deviation plot compared to age-matched normal controls. Points closer to fixation are weighted slightly more heavily.
• Polarity & Units: Expressed in decibels (dB). Normal values range from 0.00 dB to -2.00 dB. A negative value indicates overall visual field depression below normal:
  • Mild Glaucoma: $\text{MD} > -6.00\text{ dB}$
  • Moderate Glaucoma: $\text{MD}$ between $-6.00\text{ dB}$ and $-12.00\text{ dB}$
  • Advanced / Severe Glaucoma: $\text{MD} < -12.00\text{ dB}$
• Limitation: MD is non-specific; a patient with a mature cataract and zero glaucoma will exhibit a severely depressed MD of $-15.00\text{ dB}$.

2. Pattern Standard Deviation (PSD)

• Definition: A measurement of the standard deviation of points from the overall hill of vision; it measures focal irregularity and contour non-uniformity across the visual field.
• Clinical Significance: High PSD indicates localized scotomas (characteristic of early to moderate glaucoma, branch retinal vein occlusions, or localized neurological infarcts). A normal field or a field with pure dense cataract has a low PSD because the field is uniformly smooth.
• The Terminal Glaucoma PSD Paradox: In end-stage glaucoma (MD $<-25\text{ dB}$), PSD paradoxically decreases back toward normal levels (approaching 0 to 2 dB). This occurs because the entire visual field is extinguished; when all test points drop to 0 dB, there is zero focal variance across the field!

3. Glaucoma Hemifield Test (GHT)

• Algorithm: Compares visual sensitivity across the horizontal raphe by evaluating five mirrored sector pairs in the superior versus inferior visual hemifields (nasal, central arcuate, peripheral arcuate, paracentral, and central).
• Diagnostic Output Categories:
  1. Outside Normal Limits (ONL): Significant difference between superior and inferior sector pairs at the $p < 1%$ level. Highly specific for glaucomatous nerve fiber loss.
  2. Borderline: Sector differences significant at the $p < 3%$ level.
  3. General Reduction of Sensitivity: Overall field depressed without significant asymmetry between hemifields ($p < 0.5%$).
  4. Abnormally High Sensitivity: Field sensitivities significantly higher than age-matched norms ($p < 0.5%$), almost invariably secondary to high false positives (>15%).
  5. Within Normal Limits (WNL): No statistically significant asymmetry or depression.

4. Visual Field Index (VFI)

• Definition: Introduced by Bengtsson and Heijl, VFI expresses the percentage of remaining visual field function adjusted for age, ranging from 100% (completely normal) to 0% (perimetrically blind).
• Key Advantages:
  • Calculated primarily from the Pattern Deviation plot, making it highly resistant to the confounding effects of progressive cataracts.
  • Centrally weighted: foveal and parafoveal points receive significantly higher mathematical weighting than peripheral points, matching the physiological density of retinal ganglion cells.
  • Used in Guided Progression Analysis (GPA) to calculate a linear regression rate of progression (% loss per year) and project future visual field loss over a 5-year timeline.