Article Text

Download PDFPDF
CADS grading scale: towards better grading of ophthalmic involvement in facial nerve paralysis
  1. Raman Malhotra1,
  2. Kimia Ziahosseini1,
  3. Andre Litwin1,
  4. Charles Nduka2,
  5. Nora El-Shammah1
  1. 1Corneoplastic Unit, Queen Victoria Hospital NHS Foundation Trust, East Grinstead, UK
  2. 2Department of Plastic Surgery, Queen Victoria Hospital NHS Foundation Trust, East Grinstead, UK
  1. Correspondence to Raman Malhotra, Corneoplastic Unit, Queen Victoria Hospital NHS Foundation Trust, East Grinstead RH19 3DZ, UK; Raman.Malhotra{at}


Introduction Ophthalmologists lack a facial nerve grading instrument (FNGI) that comprehensively encompasses the ophthalmic sequelae of facial nerve paralysis (FNP). Assessment and management of ophthalmic sequelae remains inconsistent, and outcomes of clinical studies are incomparable. We have developed and successfully adopted an FNGI based on four aspects of periorbital involvement: cornea, asymmetry, dynamic function and synkinesis. This CADS classification is specific for periorbital involvement, with objective and subjective parameters, to standardise grading. We present this classification and the results of a validation study in clinical practice.

Methods A cross-sectional, validation study. Two clinicians independently assessed and graded each patient on the same day, blinded to each other's grading. Each grader assigned a score to each of four parameters: C (0–3, ±a), A (0–2), D (0–3), S (0–2).

Results Thirty patients (19 females, mean age 60, range 30–84 years) with unilateral facial paralysis were graded. A total of 60 assessments were conducted. CADS scores ranged from C0A0D1S1 to C3aA2D3S0. In the first 30 assessments (of the first 15 patients), the two assessors disagreed over the corneal grading in four patients. The last 30 assessments of 15 patients showed complete agreement in all four parameters of the grading scale. The overall inter-observer agreement was 86.7% for cornea, 93.3% for resting asymmetry, 93.3% for dynamic function and 86.7% for synkinesis. After the first six patients, Cohen's κ reached 1 for all but synkinesis that ranged between 0.9 and 1.

Conclusions We present a validation study of an FNGI specifically designed for ophthalmic involvement of FNP. Objective and subjective parameters helped standardise grading and management planning.

  • Rehabilitation
  • Eye Lids
  • Ocular surface
  • Cosmesis

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.


To paraphrase Lord Kelvin, ‘If you cannot measure it, you cannot improve it’.1 Facial nerve grading instruments (FNGI) aim to provide a uniform and accurate method for assessing facial nerve function, allowing communication and comparison between practitioners and evaluation of changes in clinical course. FNGI should document the clinical assessment as objectively as possible and should be sensitive enough to reflect signs of recovery or changes in function following interventions.2 In addition to being cost effective, fast, minimally invasive, sensitive, specific, objective and quantitative, the ideal characteristics of an FNGI have been suggested:3–6 It provides regional scoring of facial function, performs static and dynamic measures, examines secondary sequelae of facial palsy (eg, synkinesis), yields reproducible results with low inter observer and intraobserver variability, is sensitive to track changes over time and following interventions and convenient for clinical use.

Ophthalmologists lack such FNGI for managing ophthalmic sequelae of facial nerve paralysis (FNP).7 Currently available classifications fail to comprehensively encompass ophthalmic sequelae and complications.7 ,8 Our assessment and management of the functional and aesthetic ophthalmic complications of FNP remain inconsistent and the outcomes of clinical studies continue to be incomparable. It is time for ophthalmologists to reach consensus on a region-specific grading scale that is accurate, objective and reproducible.7

We have developed and successfully adopted an FNGI based upon four aspects of ophthalmic involvement in FNP. These comprise corneal involvement, resting or static asymmetry in the periorbital region, dynamic function of periorbital muscles and synkinesis involving this area. This CADS classification is specifically designed for periorbital involvement with a combination of objective and subjective parameters to facilitate a more standardised grading (table 1). In turn, this should aid in management planning.

Table 1

The CADS classification

We present this classification and the results of a validation study in clinical practice.


The CADS score is summarised in table 1.


This Section aims at recording the degree of exposure keratopathy and risk of sight-threatening decompensation, based upon horizontal corneal exposure zones.

Using slit lamp biomicroscopy and blue filter, standard tear film break-up time is measured. Corneal staining is graded into four categories based upon the extent of punctate epithelial erosions (PEEs), 0: no staining, 1: less than 5 inferior PEEs, 2: Greater than 5 PEEs affecting the inferior-half of the cornea but not beyond, 3: PEEs involving more than half of the cornea, extending beyond the horizontal corneal midline, or presence of a larger epithelial defect.

The presence, degree and direction of a Bell's phenomenon are recorded. A qualitative measurement for the presence of corneal sensation using a thin cotton wisp is carried out without topical anaesthesia. It is recorded as normal, reduced, or absent based upon the response to the contralateral eye or the patient's response. If the lower eyelid is in apposition to the globe, then a Schirmer’s test is performed with topical anaesthesia. Standardised strips of filter paper (Snostrips, Chauvin Pharmaceuticals, UK) are placed in the lateral canthus away from the cornea and left in place for 5 min with the eyes closed. Readings are reported in millimetres of wetting for 5 min.

Absence of an up-going Bell's phenomenon, absence or reduced corneal sensation, a reduced Schirmer's test of ≤5 mm, or if the affected eye is an only-eye will denote an additional ‘a’ to the corneal grade. This helps reinforce the priority of corneal protection and closure in an individual's care, over and above that of improving static asymmetry or treating synkinesis.

Static asymmetry

Resting asymmetry of the periorbital region grades the resting position of the involved brow and upper and lower eyelids. It is graded into 3 grades (0, 1 and 2) with increasing static involvement. Grading static parameters help in standardising and improving asymmetry of appearance. This often, but not always, would also reduce risk of corneal exposure. Instances where intervention for corneal exposure would certainly take precedence over that for asymmetry would be, for example in C3a severity, or when correcting brow ptosis may potentially compromise eyelid closure.

Grade 0 may be considered normal, but is largely subjective. Resting brow ptosis is subjectively graded as absent (0), mild (1) or severe (2) in comparison with the expected normal position for an individual and also to the contralateral brow.

Lower eyelid asymmetry refers to ectropion and retraction. Medial ectropion and/or mild lower eyelid retraction (≤2 mm inferior scleral show) is graded 1, whereas more significant ectropion and/or retraction (>2 mm inferior scleral show) is graded 2.

In addition to objective upper and lower eyelid positions in relation to the corneal limbus, the only other quantitative measurement is the lid margin-to-brow distance (LMBD; measured in mm). Using a standardised technique, LMBD represents the distance between the upper eyelid margin and the lower border of the brow hair in the mid-pupillary line. The assessor asks the patient to look straight ahead. She/he would imagine a vertical line from the pupil to the brow and would place their thumb on the brow where the two intersect. They ask the patient to look down and while elevating the brow, stretch the skin and smoothen all skin folds. A paper ruler, held against the eyelid measures the vertical distance between the lash line and the base of the brow hair. This is repeated on the contralateral side and compared. Ipsilateral contraction of upper eyelid skin is an observed sequelae of patients with FNP and may affect asymmetry and influence dynamic closure and lagophthalmos.8 It may also elude to possible coexisting contraction of lower eyelid skin.

Dynamic function

Dynamic function of orbicularis oculi and frontalis are the two most important indicators of functional periocular disability in FNP. Function may be clinically monitored indirectly, both quantitatively by lagophthalmos and qualitatively, by range of brow movement. Three aspects of lagophthalmos are measured: during spontaneous blink, gentle and forced closure.

To observe the subtlety of spontaneous blink lagophthalmos, the patient is silently observed for a minute with no instruction given regarding eyelid closure. Either a millimetre ruler is placed just lateral to the eyelids to record lagophthalmos occurring during blink or the patient is observed on the slit lamp with a thin vertical light beam adjusted to the height of blink lagophthalmos. The graticule height of the vertical beam (in mm) can then be recorded. Frontalis function is assessed qualitatively by the degree of brow movement. In comparison with the normal side, reduced brow elevation is recorded as either reduced (but present), or absent (or twitch movement only).


Synkinesis is not fully understood but is postulated to be due to misdirected regenerating axons.9 Beurskens et al10 reviewed the frequency and location of synkinesis and found that the most common movements that caused involuntary eye closure were lip puckering (82%), smiling (59%) and brow elevation (50%). Attempted eye closure is therefore accompanied by a paradoxical contraction of the frontalis, which elevates the brow, antagonising eye closure.11 Some patients are exquisitely sensitive to eyelid synkinesis, and a frequent strategy employed by many is to downgrade their facial movements in order to minimise the degree of eyelid asymmetry.12 Periocular synkinesis is a particularly distressing complication of FNP as it is occurs during social interactions such as smiling, eating and drinking. Gustatory epiphora (or crocodile tears) is another bothersome ophthalmic sequelae due to ipsilateral lacrimation when eating, drinking or salivating. This occurs following misdirection of regenerating gustatory facial nerve fibres, through the greater superficial petrosal nerve innervating the lacrimal gland. Patients request treatment as this interferes with talking, eating and drinking. Assessment of synkinesis in clinical practice is currently largely qualitative and subjective. Although validated facial synkinesis questionnaires may be useful when evaluating therapies,13 for the purposes of weighting the severity and significance of synkinesis to a patient's overall FNP we have simplified grading into three categories: absent, mild and severe. We have found this helpful in deciding the appropriateness and planning of botulinum toxin (BoNT) treatment. For example, a CADS score of S1 (mild eye closure when smiling/speaking/eating or gustatory epiphora, but not bothersome) would not usually warrant treatment. However, in order not to risk corneal exposure, a synkinesis score of S2 (significant eye closure when smiling/speaking/eating or bothersome gustatory epiphora) may not automatically be an indication for BoNT if the CADS cornea score was C1 (superficial punctate staining but less than 5) or above, or CADS dynamic function score was D2 (≥5 mm lagophthalmos on blink, or ≤5 mm on gentle closure) or above.

Validation study

We conducted a cross-sectional validation study over a 3-month period (October–December 2014). Three clinicians (one consultant (RM) and two fellows) participated in grading. The CADS table was available as a reference during assessment and grading. When grading each parameter, the highest applicable score was assigned. Two clinicians independently assessed and graded each patient attending clinic on the same day, blinded to each other's grading. The Schirmer's test and corneal sensation were only assessed once. A Cohen's κ test was used to measure agreement between two raters.


Thirty patients (19 females, mean age 60, range 30–84 years) with unilateral FNP were graded. A total of 60 assessments were conducted. The aetiology of FNP was acoustic schwannoma surgery (9), Bell's palsy (10), Ramsey Hunt Syndrome (4), parotid tumour (1), facial nerve neuroma (1), Lyme disease (1), paraganglioma (1), vagal schwannoma (1), facial laceration (1) and congenital (1).

Table 2 details the CADS scores and hence, severity of FNP for all 30 patients. These ranged from C0A0D1S1 to C3aA2D3S0. Of the majority (19, 63%), 15 and 4 were graded as having a CADS corneal score of C2 (punctate epithelial erosions ≤½ cornea) or C3 (punctate epithelial erosions >½ cornea or epithelial defect), respectively (table 2). Of those graded C2, 73% (11) and 20% (3) had dynamic function of D2 (lagophthalmos on blink5 mm or lagophthalmos on gentle closure5 mm) or D3 (lagophthalmos on gentle closure>5 mm or lagophthalmos on forced closure>2 mm), respectively. Overall, of the 30 patients in this series, 80% (24) had a CADS dynamic function score of D2 (18, 60%) or D3 (6, 20%).

Table 2

CADS scores for 30 patients grouped according to cornea grades

In the first 30 assessments (of the first 15 patients), two assessors disagreed over corneal grading in four patients by a maximum of two grades, one grade in two patients and two grades in the other two. This was identified as being due to staining following tonometry which one of the two assessors attributed to corneal exposure. This discrepancy resolved in the subsequent assessments. The last 30 assessments of 15 patients showed complete agreement in all four parameters of the grading scale. Taking into account all assessments, the overall interobserver agreement for corneal grading was 86.7%. κ varied from 0.306 to 1, suggesting fair to perfect agreement. Fair agreement was observed in the first six assessments (between KZ and NE-S) and improved to good and perfect for the rest. Between NE-S and RM, κ was 0.759 and between KZ and RM, κ was 1.

The interobserver agreement for the grading of static asymmetry was 93.3% with one grade disagreement in two patients. κ varied from 0.720 to 1, good to perfect agreement that improved in the latter assessments. Between KZ and NE-S, κ was 0.739. There was complete agreement between KZ and RM (κ=1) and between NE-S and RM κ was 0.720.

The interobserver agreement for dynamic function was also 93.3% with one grade disagreement in two patients. κ agreement was poor for the first few assessments, but then improved to moderate (0.588) and perfect agreement in the latter assessments. Between KZ and RM, κ was 1. Between NE-S and RM, κ was 0.588.

The observers disagreed by one grade over the grading of synkinesis in four patients. This gave an interobserver agreement of 86.7%. This was due to interobserver disagreement on the presence or severity of gustatory epiphora in the first few assessments of one fellow. κ statistics were 0.182 for the first six patients and improved significantly for the remainder to 0.90 and 1. Between KZ and NE-S, κ was 0.182. Between KZ and RM, κ was 0.903, and between NE-S and RM, κ was 1.


To our knowledge, the CADS classification is the first FNGI specific to the ophthalmologist, addressing ophthalmic involvement. This is a step towards reaching a universal agreement in assessing and managing ophthalmic manifestations of FNP.

The Sunnybrook grading system,14 introduced in 1996, has been recommend as the current standard in reporting outcomes of facial nerve disorders.2 This subjective, regional scale assigns a weighted score to facial regions, evaluating symmetry at rest and during voluntary movements as well as scoring synkinesis to produce a single score. Such a score on a continuous scale can be manipulated mathematically. It is arguably a more comprehensive and accurate description of facial nerve function than the 1983 House-Brackmann grading system (HBGS),15 ,16 which has less sensitivity,17–20 repeatability21 or interobserver/intraobserver reliability when used by novice or expert users.17 ,21 However, important functional and aesthetic periorbital aspects are not addressed. The resting position of eyebrow does not count towards the total score. The four options for assessment of resting symmetry of the periorbital region are simply normal=0, narrow=1, wide=1 and eyelid surgery=1.‘Narrow’ and ‘wide’ palpebral fissures score the same despite a widened palpebral fissure carrying a higher risk of incomplete closure and corneal exposure. The symmetry of voluntary movement is graded in comparison with the other side (assuming it to be normal and therefore compromising grading of bilateral FNP). Although gentle eye closure is given a score of 1–5, the efficiency of blink and degree of blink lagophthalmos is ignored. The protective Bell's phenomenon, corneal sensation, tear secretion and corneal slit lamp evaluation is not assessed either.

The HBGS is a ranked, subjective and qualitative FNGI that gives an overall impression of FNP by classifying it into six grades; grade I is normal and grade VI is complete paralysis. Being simple, it is still widely used. However, it cannot distinguish finer differences in facial nerve dysfunction due to the broad definitions of each grade. It therefore cannot accurately evaluate patients with differing extent of weakness or recovery in each facial region. It is not effective in detecting change following therapeutic intervention and does not accurately address secondary defects such as synkinesis or contracture.22 ,23 Reported interobserver reliability is fair at approximately 63% with a range of 50%–87%, but dependent upon the statistical method used in comparison.19 ,21 Others have found interobserver reliability to be poor at 33%.24 Furthermore, the HBGS only uses eye closure to describe ophthalmic involvement and fails to describe the static position of the upper and lower eyelid or the completeness of blink or other subtleties of lagophthalmos.

The four parameters and descriptive style of the CADS score is inspired by the American Joint Committee on Cancer (AJCC) tumour, node, metastasis (TNM) classification.25 ,26 Just as tumour (T), lymph node (N) and metastasis (M) in the AJCC classification are graded in parallel, the CADS score grades each of the four aspects of ophthalmic involvement in FNP in parallel: corneal involvement, resting or static asymmetry in the periorbital region, dynamic function of periorbital muscles and synkinesis. Each assessment can be conducted within minutes in the outpatient clinic but usually requires the CADS table to be available at hand for reference. The system applies to any aetiology of FNP, or for that matter any myopathic cause for orbicularis oculi weakness. We observed a short learning curve for clinicians introduced to CADS-style grading. Interobserver agreement reached 100% after the first seven to eight assessments for every observer. It was therefore easy to learn and detailed enough to take into consideration ocular involvement ranging from the subtleties of lagophthalmos on blink, upper eyelid retraction to synkinesis. The high interobserver agreement in static asymmetry and dynamic function reflects the more objective nature of the grading criteria. It is obvious that tonometry may falsely contribute to a higher grading of corneal staining with fluorescein. This became apparent when evaluating interobserver agreement of corneal grading in our study. Once this was recognised and addressed, interobserver agreement for corneal staining reached 100% in our study. It is also important to allow sufficient time to elapse after instillation of Fluorescein for corneal staining to take place in order to avoid a false low grade. In the cornea category, vision was not included as it is dependent on many variables, some of which may be unrelated to FNP. We avoided including reduction in vision as this again may be unrelated or due to factors such epiphora due to reflex tearing. We deliberately chose not to use standard ocular surface scoring systems, such as the van Bijsterveld scoring system.27 Although these are accepted fluorescein dye (or 1% rose bengal) based ocular surface grading scales, they divide the ocular surface into three vertical zones—nasal conjunctival, corneal and temporal conjunctival, with staining scores (0–3) attributed to each zone. Vertical zones are not so relevant to the distribution of staining and exposure keratopathy in FNP, where the exposure zones are horizontal and related to lagophthalmos. Resting brow ptosis is difficult to record objectively and quantitatively. We deliberately avoided the category of moderate brow ptosis as both variability and subjectivity of grading increases and also, in our experience, the decision as to whether or not to surgically treat brow ptosis is more reproducible with gradings of absent, mild and severe. We did not include frequency of blink as a component of dynamic function. Although we recognise its clinical importance, it is difficult to assign reliable normative values for the purposes of the CADS scale that would apply at all times of the day and be independent of activities.

When using the CADS score in clinical practice, we prioritised corneal findings as a major factor in clinical decision-making. In aiming to improve the ocular surface and reduce lagophthalmos, this may result in a significantly reduced palpebral aperture, particularly where FN function is poor. During follow-up, if the C score consistently remained 0, however, then this would provide justification to cautiously reverse such changes, dependent on the patient's desire to improve static appearance. The CADS score therefore does not provide a classification that aims towards a normal appearance. We used it to aim for the lowest C score and also to grade and monitor severity of FNP sequelae. An example of this is in the care of a 71-year-old male with a 4-month history of right Ramsey–Hunt facial palsy. He has severe right facial weakness with severe static asymmetry, minimal FN function and no sign of synkinesis. He has severe brow ptosis and no frontalis function. The upper eyelid is retracted (1 mm of scleral show) but obscured by brow ptosis. The lower eyelid is retracted (4 mm scleral show) with obvious paralytic ectropion. Lagophthalmos is 12 mm, 10 mm and 8 mm on blink, gentle and forced closure, respectively. He has previously undergone upper eyelid blepharoplasty; however, upper eyelid margin-to-brow distance on stretch is 2 cm on the right in contrast to 3 cm on the contralateral side, suggestive of ipsilateral skin contracture. He has marked corneal punctate staining affecting the inferior half of the right cornea. Corneal sensation is present and Bell's phenomenon is up-going. His CADS score is therefore C2A2D3S0. He undergoes insertion of a 1.4 g platinum segment chain, levator recession, lower eyelid transconjunctival retractor recession with lateral horn lysis, transcaruncular medial canthal tendon plication, suture sling and lateral tarsal strip. Four weeks postoperatively, lagophthalmos is 5 mm and 0 mm on blink and gentle closure, respectively. He has brow ptosis and blepharoptosis, no lower eyelid retraction and no corneal staining. His CADS score is C0A2D2S0 (A2 due severe brow ptosis still present). Two months following a direct browlift, he remains comfortable, and has reduced his frequency of using lubricants to once a day only. He has improved static symmetry of brows, upper and lower eyelids. Lagophthalmos is 6 mm, 2 mm and 0 mm on blink, gentle and forced closure, respectively. He has more than five inferior corneal punctate stains, but limited to his inferior half of the cornea. A right Schirmer's test (globe-lid apposition is now present) is 10 mm. His CADS score is C2A1D2S0. He may look better, with improved closure; however, the cornea has signs of increased exposure. Increased lubricants and facial therapy (including eyelid stretching) is advised. Seven months later, his blepharoptosis has improved (uMRD 3 mm), lagophthalmos is 5 mm on blink, resolving on gentle closure. One month following fat grafting to his right brow temple, lateral canthus and lower eyelid and cheek region, he is comfortable on twice daily lubricants. Upper face static symmetry is excellent. Lagophthalmos is 4 mm on blink, resolving on gentle closure. His CADS score is C0A0D1S0.

The CADS score may also be of value is determining a management plan. For example, a 42-year-old female with a 9-month history of a partially recovered Bell's palsy and marked synkinesis as well as gustatory epiphora is seeking BoNT therapy. She uses lubricants twice daily for exposure-related symptoms. She has excellent upper face static symmetry. Lagophthalmos is 6 mm and 1 mm, on blink and gentle closure, respectively. She has up to five inferior corneal punctate stains. Her CADS score is C1A0D2S2. She is therefore advised to either increase the use of lubricants if she is to undergo BoNT therapy (either to eyelids or lacrimal gland), or consider upper eyelid loading with a 0.8 g platinum segment chain (based upon a loading trial) with the aim of achieving C0 and ideally, D1 prior to BoNT.

In conclusion, we present a validation study and our experience of an FNGI based upon four aspects of ophthalmic involvement in FNP. It is specifically designed for periorbital involvement with a combination of objective and subjective parameters to facilitate standardised grading and aid management planning. This is a step towards reaching a universal agreement in assessing and managing ophthalmic manifestations of FNP.


View Abstract


  • Contributors All authors have made a substantial contribution to the conception or design of the work; or the acquisition, analysis or interpretation of data for the work; drafting the work or revising it critically for important intellectual content; final approval of the version to be published.

  • All authors agree to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved.

  • Competing interests None declared.

  • Ethics approval Queen Victoria Hospital R+D committee.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Linked Articles

  • At a glance
    Keith Barton James Chodosh Jost Jonas