The accuracy of clinical neurosensory testing for nerve injury diagnosis

J Orul 552-8

Maxillofac 1998

Surg

The Accuracy of Clinical Neurosensory Testing for Nerve Injury Diagnosis John R. Zuniga, Roger A. Meyer, Michael

Miloro,

DMD,

PHD, MS, *

DDS, MD, f John M. Gregg, DDS, PHD, MS, f DMD, MD,$

and Leon F. Davis,

DDS, MS, MD11

Purpose: The accuracy of the clinical neurosensory test to diagnose trigeminal nerve injuries has never been statistically evaluated. The purpose of this study was to determine the statistical efficacy of the clinical neurosensory test using surgical findings as the “gold” standard, and to determine whether a correlation existed between the sensory impairment score obtained by preoperative testing and the degree of nerve injury found at surgery. andMethods: A multisite, randomized, prospective, blinded, clinical trial was conducted on 130 patients with inferior alveolar nerve (IAN) and lingual nerve (LN) injuries. Preoperatively, patients were provided a sensory impairment score using a three-level drop-out clinical neurosensory test (NST), and blind comparisons were made with the surgical findings postoperatively. Materids

The positive predictive and negative predictive values for LN-injured patients were 95% and lOO%,respectively. The positive predictive and negative predictive values for LAN patients were 77% and 60% respectively. There were statistically significant differences in the distribution of age, duration of injury, cause of injury, presence of neuropathic pain, presence of trigger pain, and degree of injury between the IAN and LN patient populations. There was a statistically significant positive relationship found between the sensory impairment score and the degree of nerve injury. Resulfs:

Conclusions: The NST is a clinically useful method to diagnose IAN and LN injuries. However, the NST results are lessefficient for IAN injuries than LN injuries, and have a high incidence of false-positive (23%) and false-negative (40%) results when testing patients with IAN injuries. The different rates of statistical efficiency between the two groups of patients may be attributable to differences in prevalence and biologic covariates. The initial steps in the care and management of oral and maxillofacial surgery patients with neurosensory complaints involving the lingual or inferior alveolar nerve are based on the assessmentof general or special sensory impairment(s). The goal of sensory impairment assessmentis to acquire information to render a clinical diagnosis, to aid in determining meaningful prognosis, and to determine beneficial therapy for the nerve injury. The data required to provide a clinical diagnosis are obtained through investigative tests thought to evaluate sensory nerve function and distinguish different degrees of nerve injury. Although not conclusive, the list of publications of tests advocated for the diagnosisof Lingualand inferior alveolar nerve injury have included mapping schemes alone’; mapping with tactile, thermal, and pulp stimulation2; serial tactile and thermal testG5; clinical test algorithms with tactile and painful stimuli6,7; somatosensory evoked potential recording@; microelectroneurography9; electronic thermog-

*Associate Professor, Department of Oral and Maxillofacial Surgery, University of North Carolina at Chapel Hill, Chapel HilI, NC. tAssistant Emory

Clinical

University;

Surgery, Surgery,

Professor,

Plastic and Reconstructive

Consultant,

Georgia Department Northside Hospital,

Surgery,

Cleft Lip and Palate and Craniofacial of Human Resources, Atlanta, GA; in private

Department of practice, Mari-

ctta, GA. *Adjunct Technical

Professor, University,

Division Blacksburg,

of Veterinary VA; in private

Biology, practice,

Viiginia Blacks-

burg, VA. SAssistant Professor, Department of Oral and Maxillofacial gery, The Ohio State University, Columbus, OH. /Professor Department Medical

Sur-

and Chief, Section of Oral and Maxillofacial Surgery, of Surgery, College of Medicine, University of Nebraska

Center,

Omaha,

NE.

Address correspondence and reprint requests to Dr Zuniga: Associate Professor, Oral and Maxillofacial Surgery, UNC, School of Dentistry, Campus Box 7450, Chapel Hill, NC 275997450. 0 1998

American

Association

of Oral and Maxillofacial

Surgeons

0278.2391,‘98/5601-0002$3,00/O

2

ZUNIGA ET AI.

3

raphyl”; taste testing with vital staining and videomicroscopyll; and magnetic source imaging.12The clinical neurosensory test was the overwhelming choice of oral and maxillofacial surgeons belonging to the AAOMS Maxillofacial Neurologic Disorders Clinical Interest Group polled in 1995, because it is noninvasive, inexpensive, and can be performed easily at chairside in a short time. The relationship between the clinical neurosensory test results and the presence or absenceof a lingual or inferior alveolar nerve injury (ie, the diagnostic efficacy) has never been statistically examined. The efficacy of a diagnostic test is defined as the ability to indicate the presence or absence of disease,and it is usually calculated from statistical expressions of the diagnostic test’s sensitivity, specificity, and positive predictive and negative predictive accuracy acrossthe spectrum of the disease being studied. An ideal clinical diagnostic neurosensory test would always be positive in the presence of nerve injury and negative in the absence of nerve injury. Additionally, the test would accurately distinguish the degree of nerve injury. The purpose of this study was twofold: first, to determine the statistical efficacy of the clinical neurosensory test algorithm’ using a fourfold table with surgical findings as the “gold” standard, and second, to determine whether a correlation exists between the sensory impairment score obtained by testing and the degree of nerve injury found at surgery.

Patients

and

Methods

One hundred thirty patients were enrolled in a multisite, randomized, prospective, blinded, clinical trial conducted from March 1, 1995 to December 31, 1996. Inclusion criteria required that patients be of any gender and age, with a neurosensory complaint in the distribution of either the inferior alveolar nerve (IAN) or the lingual nerve (LN), and elect surgical intervention. The neurosensory complaint could be unacceptable altered sensation with or without painful dysesthesia. Before surgery, the following data were collected from each patient: 1) gender and age; 2) type of procedure or event associated with the injury, prompted by 10 options that included third molar, orthognathic, implant, dentoalveolar, preprosthetic, oncologic, and periodontal surgery, and endodontic therapy, external trauma, or needle injury; 3) time from injury to repair; 4) side of injury; 5) impairment level scored on the clinical neurosensory test algorithm (seelater discussion);6) presence or absenceof a trigger; and 7) presenceor absenceof neuropathic pain. There were five participating sites in this study.

These included three academic-basedand two private practices of oral and maxillofacial surgery that were geographically isolated (Blacksburg, VA; Marietta, GA; Chapel Hill, NC; Columbus, OH; and Omaha, NE). Evaluators at each of the participating sites were experienced in microneurosurgery of the IAN and LN and were trained and calibrated in the performance and recording of the neurosensory tests algorithm. The sensory impairment score in every patient was determined using a previously published three-level dropout clinical neurosensory test algorithm’ (Fig 1). Briefly, level A testing measured spatiotemporal sensory perception with brush-stroke directional sensitivity and static two-point discrimination; Level B testing measured contact detection with SemmesWeinstein monofilaments; and Level C testing measured pain threshold and tolerance using either an algometer, thermode, or sharp instrument. Testing for the IAN was performed over a l-cm area on the labiomental fold of the injured and uninjured sidesof the chin, and testing for the LN was performed over an area 1 cm distal to the outstretched tip of the tongue and 0.5 cm lateral to the midline of the injured and uninjured sidesof the tongue. Standard sensory procedures and psychophysical methods described by Essick13were used for testing. One of five scores of sensory impairment were assigned to each patient according to the following gradations: 1) Normal (the responseson the injured side and the uninjured side exhibited comparable values that were within published normative limits at all three levels of testing). “Normal” does not indicate that the patient was neurologically intact it refers to the patient’s ability to perform within normative limits and not on his or her report of altered sensation; 2) Mild (level A test results were abnormal but normal in B and C); 3) Moderate (Level A and B test results were abnormal but normal in C); 4) Severe (Level A and B test results were abnormal and elevated in C); 5) Complete (Level A and B test results were abnormal and absent in C). The presence or absence of a trigger was recorded for each patient. A trigger was defined as a painful or nonpainful electriclike sensation evoked by palpation of the nerve. When present, a trigger was characterized as radiating or nonradiating. The presence or absenceof neuropathic pain alsowas recorded. Neuropathic pain was defined asstimulus-induced or spontaneous pain associated with allodynia, hyperpathia, hyperalgesia, anesthesia dolorosa, or sympathetically mediated pain. There was no attempt to subcategorize the neuropathic pain state. The presurgical data were delivered to the University of North Carolina (UNC), which acted as the data collection site during the entire study. Each patient underwent microneurosurgery, and

ACCURACY OF NERVE TESTS

Right

Pain(tcr-) Tong”* Tkeshdd (1) Threshold (2) Tolera”c% (1) TdeTance (2)

_

_ 48

......... .........

48

Trigger Painful _ Non paintul _ radiating _ nor radiating _ none _ Comments:

FIGURE 1, nerve trigger

Clinical neurosensory test in forms used for reporting sensory impairment scores injuries. Each form was designed for recording the necessary demographic information response, recording test outcomes at Levels A, 8, and C, and determining the sensory

the surgical specimen report or the surgeon’s operative findings were used to determine the definitive nerve injury diagnosis. One of the five following categories of surgical findings was assigned by each surgeon: 1) normal/intact (N/I); 2) compressed/intact (C/f); 3) neuroma-in-continuity (NIC); 4) Partial transection (PT); 5) complete transection (CT). Each patient had a surgical procedure(s) report, which included: 1) external neurolysis; 2) internal neurolysis; 3) direct neurorraphy; 4) graft neurorraphy; 5) no repair. The presence or absence of a neuroma was recorded. The surgical findings reports were delivered to the data collection site. The surgical findings served as the “gold” standard diagnosisof nerve injury for each patient in this study, and blind comparisons were made between the surgical findings and the presurgical neurosensory test (NST) diagnosis. The NST results and the surgical findings for the LAN and LN patient populations were grouped and analyzed separately. The statistical efficacy of the NST diagnosis was calculated using the fourfold table described by Sackettl* to obtain computations for sensitivity, specificity, positive prediction value, negative prediction value, accuracy, and prevalence. The distribution of patient age, gender, side of

in patients with A, inferior alveolar nerve and 5, lingual about the patient, recording the location and type of impairment score based on test outcomes.

injury, and duration of injury were described by mean, median, standard deviation and range values. A twotailed paired t-test was used to analyze the differences in age, gender, side of injury, and duration of injury between the LAN and LN patient pools. A Fisher’s exact square test was used to analyze the difference in neuropathic pain and triggers between the IAN and LN patient pools. A multiple regressions analysis was performed to evaluate the effect of degree of nerve injury on the sensory impairment score. A Spearman nonparametric rank correlation analysis was performed to evaluate the association between sensory impairment score and the degree of nerve injury on a continuous scale. Level of significance was set at .05.

Results There were 130 patients enrolled in this study. There were 60 L4N and 70 LN injuries. The distribution of patient gender, age, side of injury, and duration of injury among the LAN and LN groups are provided in Table 1. Within the pool of IAN-injured patients, there were 21 males and 39 females, with 30 rightsided and 30 left-sided injuries. Within the pool of LN-injured patients, there were 25 males and 45

ZUNIGA ET AL

5

Inferior Alveolar Nerve Patients Gender M F Age WI Mean Range Median Side of injury R L Duration of injury (mo) Mean Raw NOTE.

Standard

deviations

60

70

21 39

25 45

42.08 (12.1) 18-70 43

32.15 (10.6) 14-56 30

30 30

38 32

14.06 (15.84) 2-120 11

Median

Lingual Nerve

8.23 (9.74) 1-72 5.5

of the mean are given in parentheses

females, with 38 right-sided and 32 left-sided injuries. There was no statistical difference in the side and gender between and within groups, but there was a prediliction for female patients in the study. The mean age and standard deviation was 42.08 k 12.1 years for IAN-injured patients and 32.15 t 10.58 years for LN-injured patients. The median age was 43 years for L4N patients and 30 years for LN patients. The difference between groups was statistically significant (P < .OOOl). The mean duration of time and standard deviation from injury to repair was 14.06 -+ 15.84months LAN in patients and 8.23 t 9.74 months in LN patients. The medianduration time was 11months for IAN patients and 5.5 months for LN patients. The difference between groupswas statisticallysignificant (P = .043). Table 2 lists the various procedures associated with

injury to the LAN and LN according to 10 categories. Third molar surgery was the most common procedure reported in both groups, representing the procedure in 53 (76%) of LN-injured cases,but only 24 (40%) of IAN-injured cases.Other differences between the two groups based on procedure were noted in the categories of implant surgery (IAN = 8, LN = 0), trauma (IAN = 8, LN = 0), endodontics (lAN = 3, LN = 0), and needle injury (L-N = 7, L4N = 0). There were nearly equal numbers of LN (n = 8) and IAN (n = 9) injuries associatedwith orthognathic surgery. There were 36 casesof neuropathic pain reported in this study (overall % = 27.7); 34 caseswere associated with IAN injury (57% of all IAN injuries), and two caseswere associated with LN injury (2.8% of all LN injuries). The difference between groups was significant (P < .OOOl). There were 71 casesof identifiable triggers on presurgical examination (overall % = 55%); 53 of these were LN injuries (76% of all LN injuries). Of the 53 cases,the pain in 46 (87%) radiated to the tongue tip on the affected side, and seven (13%) had nonradiating pain. There were 18 casesof IAN injuries associatedwith a trigger (30% of all IAN injuries). Of the 18, in 15 (83%) the pain radiated to the lip/chin on the affected side and in three (17%) it was nonradiating. The difference between groups was significant (P < .OOOl). The LAN surgical findings included four N/l, 17 C/l, 14 NIC, 10 FT, and 15 CT; the prevelance of a severe ClasslV (PT) or ClassV (CT) IAN injury, asdefined by Sunderland,15was 41.6% in this study. The LN surgical findings included 0 N/l, 10 C/I, 3 NIC, 16 PT, and 41 CT; the prevalence of a severe ClassIV or V LN injury was 81.4%. The distribution of surgical procedures is shown in Figure 2. Direct or graft neurorraphy was required in 83% of the LN surgeries performed but was required

8070-

Inferior Alveolar”

Nerve Third molar odontectomy Orthognathic surgery Implant surgery Needleinjury External trauma Endodonticstherapy Dentoalveolarsurgery Preprostheticsurgery Periodontalsurgery Oncologic surgery

24 (40) 9(15) 8 (14) 0 (0) 8 (14) 3 (5) 5 G9 l(l.5) l(l.5) l(l.5)

*Actual number of cases and overall nerve in parentheses.

percentage

60-

Lingual* Nerve 53 (76) 8 (11) 0 (0) 7 (10) 0 (0) 0 (0) l(l.5) 0 (0) l(l.5) 0m for the type

of

8

50-

t

40-

r; a

30-

$ Q

20-

External

ilJ .

Internal

Microneurosurgical FIGURE 2. Percent based on the nerve (WI = lingual nerve)

distribution injured (IAN

Direct . Neurorrapny Procedures

Graft Neurorraphy

NO Repair

Performed

of surgical procedures (0) = inferior alveolar

performed nerve; 1N

ACCURACY

6

in only 36.6% of the LAN surgeries. Conversely, an external or internal neurolysis was required in 48.3% of the IAN surgeries performed, but was required in only 12.8% of the LN surgeries. A neuroma was found and removed in 86% of LN surgeries and in 43% of LAN surgeries. The results of comparing graded neurosensory test results (the diagnostic test) and surgical findings (the gold standard) in a fourfold table are shown in Table 3. There was a statistically significant correlation between the surgical findings and the NST scores for both IAN (P = .002) and LN (P < .OOOl>.The ability of the NST to positively predict the results of LN microneurosurgery was 95%, and its negative predictive value was 100%. For LN injuries, the sensitivity was 100%and the specificity was 62.5%. The accuracy of the NST among the LN-injured patient pool was 96%, and the prevalence of LN injury in this study population was 88.6%. The ability of the NST to predict the results of IAN microsurgery was less efficient than for LN microsurgery. The NST’s positive predictive value for IAN microsurgery was 77%, and its negative predictive value was 60%; its sensitivity was 85%, and its specificity was 47%. The accuracy of the NST among the IAN-injured patient pool was 73%, and the prevalence of LAN injury in this study population was 68%. To better assessthe relationship between the clinical NST diagnostic efficiency and the surgical findings, the correlation between the five graded NST impairment scores and the five graded degrees of nerve injuries in each group were graphed in Figure 3. Spearman’s correlation coefficient for LN-injured patients was 0.61 and 0.56 for IAN-injured patients, and both were significantly greater than 0 (P < 0.0001). For both nerve injuries, at higher sensory impairment scores (ie, severe and complete), the NST tended to overestimate the degree of nerve injury. In addition, at lower sensory impairment scores (ie, “normal” and mild), the NST tended to underestimate the degree of IAN nerve injury.

Discussion This study showed the statistical efficacy of clinical neurosensory tests to diagnose IAN and LN injuries and their ability to predict the results found at surgery,

alveolar nerve (n = nerve (n = 70)

Abbreviations:

NST, neurosensory

making the NST clinically useful. The results also pointed out the differences in statistical rates of efficacy between the two populations of nerve injury. Using the fourfold table, the value of sensitivity acts as an index of the NST’s ability to detect nerve injury when present and answers the question: “Lf patients have a nerve injury, how likely are they to have a positive test?“. Specificity acts asan index of the NST’s ability to correctly identify the absence of a nerve injury and answers the question: “If patients do not have a nerve injury, how likely are they to have a negative test?“. However, in clinical practice, clinician’s concerns are directed to two different questions during diagnostic testing: “If patients have a positive NST, how likely are they to have a nerve injury?” (“rule in”), and “If patients have a negative NST, how likely are they to not have a nerve injury?” (“rule out”). The indices that answer these questions are the accuracy for a positive predictive value and negative predictive value, respectively. The latter two indices depend not only on sensitivity and specificity but alsoon the prevalence of nerve injury in the study population. The results of applying the fourfold test to LNinjured patients in this study suggested that the diagnostic efficacy of the NST was strong. In 95% of the time, the NST was correct in “ruling in” the presence of a LN injury and very good at predicting the degree of nerve injury (r = .61). However, there was a 5% incidence of a false-positive result using the NST. The major limitation was that the NST score tended to overestimate the injury at higher scores. However, the negative predictive value was perfect, meaning that 100% of the time the test accurately “ruled out” the presence of LN injury when the patient performed normally. There were no instances of a false-negative test found in this study. However, the fourfold test results showed that the NST was statistically lessefficient for LAN-injured patients. Seventy-seven percent of the time, the NST was correct in “ruling in” the presence of a IAN injury and was good at predicting the degree of nerve injury (r = .56). However, the NST underestimated the degree of nerve injury when the sensory impairment scores were low and overestimated the degree of nerve injury when the scoreswere high. The resultsalsoindicated that the rate of obtaining afalse-positiveNSTscorein this study

SF = Injury

NST = Injury SF = No injury

NST = No injury SF = Injury

NST = No injury SF = No injury

2:

10

6 0

9 5

NST = Injury

Inferior Lingual

OF NERVE TESTS

60)

3 test; SF, surgical

finding.

ZUNIGA ET AL

7

. .

FIGURE 3. Scatter-plot represcntatinns of the relationship between the sensory impairment score and the degree of nerve injury found at surgery {N/I = normal/Intact; C/I = compressed/intact; NIC = neuroma-incontinuity; Pi = partial transection; CT = complete transection]. A, Inferior alveolar nerve-iniured patients (n = 60). B, lingual nerveiniured patients in = 70).

.

..

.*. l/ ?*

..

/ /

x’ 2.

. . . .+:e 7.0 / J’ 2: / .. 2.

CT/ g .g e

.*.

3 P z 2 3 E g

::

PT/

/

A

I r&ml

.::: ::::

.

NC-

/ f

C/I / N/I -

I

/

&::: ::x:: x::::

/

/

/ .‘.

::

/

8’

I /’ I

Yld

Sensory

I

I

m#.?rate

sev&re

Impairment

Score

was 23%. The negative predictive value in this study was GO%,meaning that the NST was lesseflicient in “ruling out” an IAN nerve injury, and this resulted in a falsenegative rate of 40%in the study population. The differences in statistical efficacy of the NST between the two groups of nerve injuries may be due in part to differences in the prevalence covariates. For example, there were statistically significant differences in age, duration of injury, and mechanism of injury (ie, procedure associated with injury). Lingual nerve injuries were seen in younger patients, were treated earlier, and were predominantly the result of a singleprocedure (eg, third molar odontectomy), which would place the injury at the same location along the nerve axis in most patients tested. In contrast, IAN injuries occured in older patients, were treated later, and occurred as a result of multiple causes(ie, trauma, endodontics, implants, etc.), and therefore the injuries probably occurred at different locations along the nerve axis. Thus, unlike IAN injuries, LN injuries occur in a more homogenous group of patients, which may account for better test efficacy amongLN-testedpatients. If the effects of age, duration of injury, location of injury, and type of injury on nerve injury response and outcome occur in humans in the same way as they occur in animalsi one would expect to find differences between the two groups. However, the differences between the two groups may also be due, in part, to differences in biologic covariates. For example, there was a statistically significant difference in the presence of neuropathic pain, the presence of a neuroma and its detection (ie, triggers), and the degree of nerve injury found at surgery. The results of this study concur with Robinson’s2 earlier work, in which he noted that patients with different degrees of IAN injuries may show similar clinical neurosensory test values. The occurrence of neuropathic pain in the IAN-injured population in this study was nearly ubiquitous, and was relatively rare in LN-injured patients. Gregg” has pointed out that neuropathic pain states occur in intact and injured nerves in a large sample of

”

I

carplete

B

I NormSI

I Mid

Sensory

, Modera(e

lmpainnent

I Severe

I GX,“ete

Score

patients. Additionally, neuropathic pain is frequently characterized by an abnormal reduction in stimulusresponse threshold for painful and nonpainful stimuli. Thus, the greater number of neuropathic patients in the L4N group may have resulted in less efficient diagnosis of the nerve injury, because the tests were stimulus-detection-based tests. Finally, although not proven, it is tempting to speculate that the IAN will show differences in regenerative properties, when compared to the lingual nerve, due to its anatomic position within a bony canal. The, lack of a detectable neuroma (ie, trigger) among IAN injuries may be due, in part, to its intrabony location. This study showed that the clinical NST algorithm is a reliable diagnostic test to “rule in” and “rule out” LN injuries. Becausethe test is easy, noninvasive, inexpensive, and can be performed chairside in a short time, its routine use should be encouraged for LN-injured patients. The clinical NST algorithm is good to “rule in” IAN injuries and therefore should be considered a clinically useful test. However, it is less reliable in “ruling out” L4N injury and may result in a high incidence of false-negative findings. The clinician may need to take into account the age of the patient, duration of the injury, mechanism of injury, and the presence of neuropathic pain as potential confounding factors. The clinical NST algorithm will need to be carefully looked at in the future in light of better testing methods for IAN injuries.8~9~10~12 Acknowledgment The authors thank the members Neurologic Disorders Clinical Interest 1995 phase 1 study poll.

of the AAOMS Maxillofacial Group for participating in the

References 1. Robinson R, Williams C: Documentation method for inferior alveolar and lingual nerve paresthesias. Oral Surg 62:128, 1986 2. Robinson PP: Observations on the recovery of sensation following inferior alveolar nerve injuries. Br J Oral Maxillofac Surg 261177,

1988

8

DISCUSSION

3. Campbell RL, Shamaskin RG, Harkins SW: Assessment of recovery from injury to inferior alveolar and mental nerves. Oral Surg 64:519,1987 4. Ghali GE, Epker BN: Clinical neurosensory testing: practical applications. J Oral Maxillofac Surg 47:1074, 1989 5. Kesenwani A, Antonyshyn 0, MacKinnon S, et al: Facial sensibility testing in the normal and posttraumatic population. Ann Plast Surg 22:416, 1989 6. LaBanc J: Trigeminal nerve injuries and repair. Selected Readings Oral Maxillofac Surg 1:1, 1991 7. Zuniga JR, Essick GK: A contemporary approach to the clinical evaluation of trigeminal nerve injuries. Oral Maxlllofac Surg Clin North Am 4:353,1992 8. Jones DL, Wolford LM, Hartog JM: Comparison of methods to assessneurosensory alterations following orthognatbic surgery. Int J Adult Orthod Orthognath Surg 5:35, 1990 9. Colin W: Conduction velocity of the human inferior alveolar nerve: A preliminary report. J Oral Maxillofac Surg 51:1018, 1993 10. Gratt B, Shetty V, Saiar M, et al: Electronic thermography for the

J Oral

Maxillofac

11, 12.

13. 14. 15. 16. 17.

assessment of inferior alveolar nerve deficit. Oral Surg 80:153, 1995 Zuniga JR, Chen N, Phillips CL: Chemosensory and somatosensory regeneration after lingual nerve repair in humans. J Oral Maxillofac Surg 552, 1997 McDonald AR, Roberts TPL, Rowley HA, et al: Noninvasive somatosensory monitoring of the injured inferior alveolar nerve using magnetic source imaging. J Oral Maxillofac Surg 54:1068, 1996 Essick GK: Comprehensive clinical evaluation of perioral sensory function. Oral Maxillofac Surg Clin North Am 4:503, 1992 Sackett DL: How to read clinical journals. II: To learn about a diagnostic test. Can Med Assoc J 124:703, 1981 Sunderland S: A classification of peripheral nerve injuries producing loss of function. Brain 74:491, 1951 Lundborg G: Nerve regeneration, in Lundborg C (ed): Nerve Injury and Repair. London, Churchill Livingstone, 1988, p 149 Gregg JM: Studies of traumatic neuralgias in the maxillofacial region: Surgical pathology and neural mechanisms. J Oral Maxillofac Surg 48:228, 1990

Surg

56:8, 1998

Discussion The Accuracy of Clinical Neurosensory Testing for Nerve Injury Diagnosis Alex R. McDonald, DDS, PhD Universiv

of the Pacific,

San Francisco,

California

When a patient presents with postoperative alteration in sensory nerve function it is essential that the severity of the injury be accurately documented. This study attempts to correlate sensory nerve function, assessed using a clinical neurosensory test (CNT) protocol, with the physical condition of the nerve established by surgical exploration. It is surprising how little data exist supporting the accuracy of CNT. This was a large, prospective, multicenter study examining many forms of nerve injury. The strength of this study resides in the careful calibration of a group of very experienced examiners. This need for calibration can be a problem when less experienced examiners attempt to use these protocols for neurosensory examinations. We have noted discrepancies within our own institution when different examiners evaluate the same patient. Although details are given of the time between injury and surgical exploration, no mention is made of the time between CNT and surgical exploration. This is extremely important because, when examining neuropraxic injuries, large improvements in CNT can occur in a relatively short time. This could greatly affect the confidence in the experimental findings and can be significant when comparing CNT in both lingual and inferior alveolar nerve injuries. The majority of lingual nerve injuries examined were severe in nature and, as stated by the authors, such injuries usually represent nerve discontinuities, rarely undergoing spontaneous recovery. In contrast, because of the bone canal associated with the inferior alveolar nerve, spontaneous recovery can take place. A delay of several months between CNT and surgery could result in significant improvement in

the neurosensory results and this could account for some of the discrepancy between CNT and status of the nerve shown in their Figure 3. Of surprise to me is the authors’ decision to surgically explore patients who presented with a “normal examination.” Most surgeons will operate on patients who fall into the severely impaired category or those patients with peripheral nerve dysesthesia. I presume that in this study the patients who had a normal neurosensory examination had episodes of pain justifying surgical exploration. After CNT, seven patients were considered to have complete sensory impairment. Subsequent surgical exploration of this group of patients revealed only crush injuries with intact nerves. Had a longitudinal examination been performed on these seven patients, perhaps they would have been spared surgery. Because surgical exploration of nerve injuries is not trivial, often requiring an extraoral approach and use of general anesthesia, we and others have tried to develop objective methods of better evaluating nerve injuries. The authors note that the cost and inconvenience of objective testing, such as magnetic source imaging (MSI), is prohibitive and it is true that the CNT method used in this study is both inexpensive and convenient. However, it is conceivable that as objective test modalities become more available, it would be both cost-effective and in the patient’s best interests to have additional tests, such as MSI, performed before surgery. Overall, this is an excellent study by a group of very experienced investigators showing both the strengths and weaknesses of an established testing modality. The authors show that the CNT algorithm adopted by the AAOMS clinical interest group is a fairly accurate predictor of nerve injury. I hope that further advances in objective testing modalities will complement this work and that, in the future, we will be better equipped to manage nerve injuries.