Original Article
Journal of
CLINICAL NEUROMUSCULAR DISEASE
143
Volume 15, Number 4 June 2014
Higher Diagnostic Yield With the Combined Sensory Index in Mild Carpal Tunnel Syndrome Lawrence A. Zeidman, MD, Simran K. Singh, DO, and Dilip K. Pandey, MD, PhD Abstract Objectives: The combined sensory index (CSI), a sensitive composite score of 3 median sensory comparison studies, may still be underutilized in diagnosing mild cases of carpal tunnel syndrome (CTS). Our goal was to compare the effectiveness of the “standard” median digit 2 (D2) sensory study to a CSI algorithm in diagnosing mild CTS. Methods: We retrospectively identified patients with typical CTS symptoms and signs. Electrodiagnostically normal patients and those having mild CTS diagnosed by D2 or CSI algorithm were separated into groups. Results: Seventy-four patients were included, and 51 (68.9%) were diagnosed with mild CTS. Of the 51, 31 (60.8%) were diagnosed using the CSI algorithm, and 20 (39.2%) were diagnosed using D2 (P , 0.001). Conclusions: Our data suggest that the CSI algorithm is significantly more effective than the D2 to diagnose mild CTS. If mild CTS is diagnosed earlier, treatment can be initiated sooner and morbidity can likely prevented. Key Words: combined sensory index, carpal tunnel syndrome, EMG, median comparison studies, median nerve
( J Clin Neuromusc Dis 2014;15:143–146)
INTRODUCTION Carpal tunnel syndrome (CTS) is a common problem, estimated to occur in 3.8% of the population.1 Typical guidelines to diagnose CTS include the combination of clinical and electrodiagnostic findings, especially
considering that there is no “gold standard” for diagnosis, that nerve conduction studies are only 85%–90% sensitive for CTS overall, and that some patients may have asymptomatic median mononeuropathy at the wrist without having actual CTS.2 Sensory fibers are typically affected earlier in CTS, given that they have a higher content of large myelinated fibers with a high-energy requirement, and are more susceptible to ischemia.2 On nerve conduction studies, median motor abnormalities suggest moderate-to-severe CTS. If only sensory studies are abnormal, this suggests mild CTS.2 If studies are normal, but clinical suspicion remains, median sensory comparison tests are recommended. Comparison studies are believed to be more sensitive, which use the patient’s own nerves as references instead of population normals and they control for confounding factors such as temperature, age, systemic disease states (eg, diabetes), hand size, and sex.2,3 The combined sensory index (CSI) is the sum of 3 median comparison values to diagnose mild CTS, has higher sensitivity than the individual comparisons alone,4 and has a high test–retest reliability.5 An algorithmic approach starting with one of the CSI studies, and performing the entire only in uncertain cases, revealed an equivalent sensitivity and specificity, and minimizes time, cost, and patient discomfort.6 We have been using this algorithm in our laboratory in equivocal cases where there was a high clinical suspicion for mild CTS but the standard median digit 2 sensory study (D2) was normal. Despite guidelines,2 we have observed in many referrals that
From the Department of Neurology and Rehabilitation, University of Illinois at Chicago, Chicago, IL. Presented as a poster (P02.226) at the 65th annual American Academy of Neurology Meeting, March 19, 2013, San Diego, CA. The authors report no conflicts of interest. All authors had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. Supplemental digital content is available for this article. Direct URL citations appear in the printed text and are provided in the HTML and PDF versions this article on the journal’s Web site (www.jcnmd.com). Reprints: Lawrence A. Zeidman, MD, University of Illinois at Chicago Medical Center, Department of Neurology and Rehabilitation (M/C 796), Neuropsychiatric Institute, 912 S. Wood St, Chicago, IL 60612-7330 (e-mail: [email protected]). Copyright © 2014 by Lippincott Williams & Wilkins
144
Journal of
CLINICAL NEUROMUSCULAR DISEASE
Volume 15, Number 4 June 2014
Zeidman et al
median comparison studies are often not performed, and the aim of this study is to compare the effectiveness of the CSI algorithm with the standard technique in diagnosing mild CTS. We are unaware of a previous study that compared the D2 to the CSI or CSI component in diagnosing mild CTS.
MATERIALS AND METHODS After obtaining institutional review board exemption, we retrospectively analyzed EMG reports of patients presenting to the University of Illinois at Chicago Neurology Clinic EMG laboratory with typical CTS symptoms between November 2010 and June 2012. In our laboratory, symptoms and signs leading to a high clinical suspicion of CTS are the following: nocturnal paresthesias, especially waking the patient from sleep, shaking the hands to relieve symptoms, hand pain or paresthesias with holding items, sensory disturbance in digits 1–4 (especially splitting the fourth digit), thenar weakness or atrophy, and a positive Phalen maneuver.3 Possible CTS is ascertained from the following symptoms and signs: pain anywhere in the affected upper limb, paresthesias in all digits, absence or presence of objective sensory loss in digits 1–4 (with or without splitting of the fourth digit), decreased manual dexterity, and a positive Tinel sign over the median nerve at the wrist.3 Only patients with an EMG performed by the lead author of this study were analyzed. Patients with alternative potentially confounding diagnoses on EMG test results, such as cervical radiculopathy, alternative mononeuropathy, or polyneuropathy, were excluded so that only patients with CTS were included. Patients with moderate-to-severe CTS, defined by having motor involvement,2 were also excluded so that only patients with mild CTS were included. Patients were evaluated with traditional sensory and motor nerve conduction studies of the affected limb. If the D2 latency was normal or equivocal in a patient with a possible or high clinical suspicion for CTS, the CSI algorithm was used. The CSI is composed of © 2014 Lippincott Williams & Wilkins
the following: (1) the median-ulnar ring finger antidromic latency difference at 14 cm (ringdiff); (2) the median-radial thumb antidromic latency difference at 10 cm (thumb-diff); and (3) the median-ulnar mid-palmar orthodromic latency difference at 8 cm (palm-diff).4 Robinson et al6 revealed that starting with the thumb-diff or palm-diff revealed a slightly higher sensitivity (83%) and specificity (95%) than the ring-diff (82% and 94%, respectively). Given the higher and equivalent sensitivity starting with the palm-diff or thumb-diff, we chose to start with the thumb-diff given its comparative ease of use, i.e., not having to move the recording electrodes. Consistent with the algorithm, if the thumb-diff was above 0.6 millisecond (CTS) or below 0.2 millisecond (normal), then further testing was considered unnecessary. If the value was between 0.2 and 0.6 milliseconds (uncertain range), then the entire CSI was calculated. A CSI $1.0 was considered abnormal. Patients diagnosed as normal, those having mild CTS by D2, and those with mild CTS by the CSI algorithm were separated into groups for analysis. We used STATA 11.0 software (StataCorp, college Station, TX) to run the statistical analysis. A x2 test was used to test the difference in categorical variables.
RESULTS In total, 74 patients (age, 13–70 years) (see Table, Supplemental Digital Content 1, http://links.lww.com/JCND/A5) with typical CTS symptoms were included, with 51 (68.9%) diagnosed with mild CTS using either D2 sensory latency (51% sensitive) or CSI algorithm (60.8% sensitive). Of these 51, 31 (60.8%) were diagnosed using the CSI algorithm, whereas only 20 (39.2%) were diagnosed using D2 sensory alone (P , 0.001). Furthermore, of the 31 patients diagnosed using the CSI algorithm, 17 (54.8%) were diagnosed with the thumb-diff alone, whereas 14 (45.2%) required the full CSI to be calculated (P , 0.001).
DISCUSSION Our results suggest that the CSI, which is composed of multiple median sensory
Diagnosing Mild Carpal Tunnel Syndrome
comparison studies and relatively impervious to temperature effect on EMG, is more sensitive in diagnosing CTS compared with the traditional D2 sensory latency alone. In our study, a significantly higher portion of patients were diagnosed with mild CTS by the CSI algorithm than by D2 sensory latency alone. Additionally our results suggest that the thumb-diff can be used alone (as part of the algorithm) for diagnosis in a significantly high number of cases compared to the full CSI, which can spare patients from discomfort and cut down on time of the study, as well as costs. Our results suggest that use of the CSI algorithm in suspicious cases for mild CTS will improve diagnostic yield and allow patients to be treated earlier. Limitations of our study included the retrospective design and a small sample size. In our study, the use of the CSI algorithm yielded a sensitivity of 60.8%, which is lower than reported in the literature (83%).6 This difference may be attributed to the lower sample size (74 patients vs. 225 patients6) or to the demographic features of the patient population referred to our laboratory for evaluation. We plan to conduct a prospective randomized larger trial over a longer period comparing the CSI algorithm alone versus starting with D2 and using the CSI algorithm only when needed, which will provide more accurate sensitivity and specificity information and be less affected by selection bias. Though CTS is largely a clinical diagnosis, electrodiagnostic studies, including median comparison studies, are routinely recommended.7 Wrist splinting therapy can be effective in mild-to-moderate disease and is recommended before surgical treatment is considered.8 Many practitioners may consider recommending a splinting trial instead of proceeding with an EMG to confirm the diagnosis, but we believe that having a more confirmed diagnosis and excluding other treatable conditions is preferable. The additional sensitivity garnered by the CSI algorithm compared with a standard absolute latency, although more labor-intensive, costly, and potentially uncomfortable for the patient,
allows the diagnosis of more mild early cases. Introducing more median comparison studies from a statistical perspective may increase the false-positive rate, but this rate will be lower if those additional studies point to abnormalities, if the abnormalities are correlated clinically, and if the abnormalities are far from the normal values.9 At the mild early stage, wrist splinting or occupational/lifestyle modifications may be enough to treat CTS, instead of more invasive options including steroid injections or surgical treatments. One outcome study revealed that a higher CSI (2.5–4.6) correlates with better pain and paresthesiae resolution after carpal tunnel release surgery.10 We are currently conducting a retrospective outcome study with mild, moderate, and severe CTS cases, which will assess whether mild cases diagnosed by the CSI algorithm were false positives or not, and a prospective study will follow that. Early CTS diagnosis may overall reduce the costs to society including medical bills, time off of work, and disability and workers’ compensation claims.8 Additionally, an experienced EMG operator can perform the CSI studies quickly, and with the CSI algorithm patient discomfort may be lessened.
REFERENCES 1. Atroshi I, Gummesson C, Johnsson R, et al. Prevalence of carpal tunnel syndrome in a general population. JAMA. 1999;282:153–158. 2. Werner RA, Andary M. Electrodiagnostic evaluation of carpal tunnel syndrome. Muscle Nerve. 2011;44: 597–607. 3. Preston DC, Shapiro BE. Median neuropathy at the wrist. In: Preston DC, Shapiro BE, eds. Electromyography and Neuromuscular Disorders: ClinicalElectrophysiologic Correlations. 2nd ed. Philadelphia, PA: Elsevier; 2005:255–279. 4. Robinson LR, Micklesen PJ, Wang L. Strategies for analyzing nerve conduction data: superiority of a summary index over single tests. Muscle Nerve. 1998;21:1166–1171. 5. Lew HL, Wang L, Robinson LR. Test-retest reliability of combined sensory index: implications for diagnosing carpal tunnel syndrome. Muscle Nerve. 2000;23: 1261–1264. 6. Robinson LR, Micklesen PJ, Wang L. Optimizing the number of tests for carpal tunnel syndrome. Muscle Nerve. 2000;23:1880–1882. 7. Jablecki CKJ, Andary MT, Floeter MK, et al. Practice parameter: Electrodiagnostic studies in carpal tunnel www.jcnmd.com
Journal of
CLINICAL NEUROMUSCULAR DISEASE Volume 15, Number 4 June 2014
145
146
Journal of
CLINICAL NEUROMUSCULAR DISEASE
Volume 15, Number 4 June 2014
Zeidman et al
syndrome. Report of the American Association of Electrodiagnostic Medicine, American Academy of Neurology, and the American Academy of Physical Medicine and Rehabilitation. Neurology. 2002;58: 1589–1592. 8. Bickel KD. Carpal tunnel syndrome. J Hand Surg. 2010;35A:147–152.
© 2014 Lippincott Williams & Wilkins
9. Dorfman LJ, Robinson LR. AAEM minimonograph #47: normative data in electrodiagnostic medicine. Muscle Nerve. 1997;20:4–14. 10. Malladi N, Micklesen PJ, Hou J, et al. Correlation between the combined sensory index and clinical outcome after carpal tunnel decompression: a retrospective review. Muscle Nerve. 2010;41:453–457.
Journal of
CLINICAL NEUROMUSCULAR DISEASE
143
Volume 15, Number 4 June 2014
Higher Diagnostic Yield With the Combined Sensory Index in Mild Carpal Tunnel Syndrome Lawrence A. Zeidman, MD, Simran K. Singh, DO, and Dilip K. Pandey, MD, PhD Abstract Objectives: The combined sensory index (CSI), a sensitive composite score of 3 median sensory comparison studies, may still be underutilized in diagnosing mild cases of carpal tunnel syndrome (CTS). Our goal was to compare the effectiveness of the “standard” median digit 2 (D2) sensory study to a CSI algorithm in diagnosing mild CTS. Methods: We retrospectively identified patients with typical CTS symptoms and signs. Electrodiagnostically normal patients and those having mild CTS diagnosed by D2 or CSI algorithm were separated into groups. Results: Seventy-four patients were included, and 51 (68.9%) were diagnosed with mild CTS. Of the 51, 31 (60.8%) were diagnosed using the CSI algorithm, and 20 (39.2%) were diagnosed using D2 (P , 0.001). Conclusions: Our data suggest that the CSI algorithm is significantly more effective than the D2 to diagnose mild CTS. If mild CTS is diagnosed earlier, treatment can be initiated sooner and morbidity can likely prevented. Key Words: combined sensory index, carpal tunnel syndrome, EMG, median comparison studies, median nerve
( J Clin Neuromusc Dis 2014;15:143–146)
INTRODUCTION Carpal tunnel syndrome (CTS) is a common problem, estimated to occur in 3.8% of the population.1 Typical guidelines to diagnose CTS include the combination of clinical and electrodiagnostic findings, especially
considering that there is no “gold standard” for diagnosis, that nerve conduction studies are only 85%–90% sensitive for CTS overall, and that some patients may have asymptomatic median mononeuropathy at the wrist without having actual CTS.2 Sensory fibers are typically affected earlier in CTS, given that they have a higher content of large myelinated fibers with a high-energy requirement, and are more susceptible to ischemia.2 On nerve conduction studies, median motor abnormalities suggest moderate-to-severe CTS. If only sensory studies are abnormal, this suggests mild CTS.2 If studies are normal, but clinical suspicion remains, median sensory comparison tests are recommended. Comparison studies are believed to be more sensitive, which use the patient’s own nerves as references instead of population normals and they control for confounding factors such as temperature, age, systemic disease states (eg, diabetes), hand size, and sex.2,3 The combined sensory index (CSI) is the sum of 3 median comparison values to diagnose mild CTS, has higher sensitivity than the individual comparisons alone,4 and has a high test–retest reliability.5 An algorithmic approach starting with one of the CSI studies, and performing the entire only in uncertain cases, revealed an equivalent sensitivity and specificity, and minimizes time, cost, and patient discomfort.6 We have been using this algorithm in our laboratory in equivocal cases where there was a high clinical suspicion for mild CTS but the standard median digit 2 sensory study (D2) was normal. Despite guidelines,2 we have observed in many referrals that
From the Department of Neurology and Rehabilitation, University of Illinois at Chicago, Chicago, IL. Presented as a poster (P02.226) at the 65th annual American Academy of Neurology Meeting, March 19, 2013, San Diego, CA. The authors report no conflicts of interest. All authors had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. Supplemental digital content is available for this article. Direct URL citations appear in the printed text and are provided in the HTML and PDF versions this article on the journal’s Web site (www.jcnmd.com). Reprints: Lawrence A. Zeidman, MD, University of Illinois at Chicago Medical Center, Department of Neurology and Rehabilitation (M/C 796), Neuropsychiatric Institute, 912 S. Wood St, Chicago, IL 60612-7330 (e-mail: [email protected]). Copyright © 2014 by Lippincott Williams & Wilkins
144
Journal of
CLINICAL NEUROMUSCULAR DISEASE
Volume 15, Number 4 June 2014
Zeidman et al
median comparison studies are often not performed, and the aim of this study is to compare the effectiveness of the CSI algorithm with the standard technique in diagnosing mild CTS. We are unaware of a previous study that compared the D2 to the CSI or CSI component in diagnosing mild CTS.
MATERIALS AND METHODS After obtaining institutional review board exemption, we retrospectively analyzed EMG reports of patients presenting to the University of Illinois at Chicago Neurology Clinic EMG laboratory with typical CTS symptoms between November 2010 and June 2012. In our laboratory, symptoms and signs leading to a high clinical suspicion of CTS are the following: nocturnal paresthesias, especially waking the patient from sleep, shaking the hands to relieve symptoms, hand pain or paresthesias with holding items, sensory disturbance in digits 1–4 (especially splitting the fourth digit), thenar weakness or atrophy, and a positive Phalen maneuver.3 Possible CTS is ascertained from the following symptoms and signs: pain anywhere in the affected upper limb, paresthesias in all digits, absence or presence of objective sensory loss in digits 1–4 (with or without splitting of the fourth digit), decreased manual dexterity, and a positive Tinel sign over the median nerve at the wrist.3 Only patients with an EMG performed by the lead author of this study were analyzed. Patients with alternative potentially confounding diagnoses on EMG test results, such as cervical radiculopathy, alternative mononeuropathy, or polyneuropathy, were excluded so that only patients with CTS were included. Patients with moderate-to-severe CTS, defined by having motor involvement,2 were also excluded so that only patients with mild CTS were included. Patients were evaluated with traditional sensory and motor nerve conduction studies of the affected limb. If the D2 latency was normal or equivocal in a patient with a possible or high clinical suspicion for CTS, the CSI algorithm was used. The CSI is composed of © 2014 Lippincott Williams & Wilkins
the following: (1) the median-ulnar ring finger antidromic latency difference at 14 cm (ringdiff); (2) the median-radial thumb antidromic latency difference at 10 cm (thumb-diff); and (3) the median-ulnar mid-palmar orthodromic latency difference at 8 cm (palm-diff).4 Robinson et al6 revealed that starting with the thumb-diff or palm-diff revealed a slightly higher sensitivity (83%) and specificity (95%) than the ring-diff (82% and 94%, respectively). Given the higher and equivalent sensitivity starting with the palm-diff or thumb-diff, we chose to start with the thumb-diff given its comparative ease of use, i.e., not having to move the recording electrodes. Consistent with the algorithm, if the thumb-diff was above 0.6 millisecond (CTS) or below 0.2 millisecond (normal), then further testing was considered unnecessary. If the value was between 0.2 and 0.6 milliseconds (uncertain range), then the entire CSI was calculated. A CSI $1.0 was considered abnormal. Patients diagnosed as normal, those having mild CTS by D2, and those with mild CTS by the CSI algorithm were separated into groups for analysis. We used STATA 11.0 software (StataCorp, college Station, TX) to run the statistical analysis. A x2 test was used to test the difference in categorical variables.
RESULTS In total, 74 patients (age, 13–70 years) (see Table, Supplemental Digital Content 1, http://links.lww.com/JCND/A5) with typical CTS symptoms were included, with 51 (68.9%) diagnosed with mild CTS using either D2 sensory latency (51% sensitive) or CSI algorithm (60.8% sensitive). Of these 51, 31 (60.8%) were diagnosed using the CSI algorithm, whereas only 20 (39.2%) were diagnosed using D2 sensory alone (P , 0.001). Furthermore, of the 31 patients diagnosed using the CSI algorithm, 17 (54.8%) were diagnosed with the thumb-diff alone, whereas 14 (45.2%) required the full CSI to be calculated (P , 0.001).
DISCUSSION Our results suggest that the CSI, which is composed of multiple median sensory
Diagnosing Mild Carpal Tunnel Syndrome
comparison studies and relatively impervious to temperature effect on EMG, is more sensitive in diagnosing CTS compared with the traditional D2 sensory latency alone. In our study, a significantly higher portion of patients were diagnosed with mild CTS by the CSI algorithm than by D2 sensory latency alone. Additionally our results suggest that the thumb-diff can be used alone (as part of the algorithm) for diagnosis in a significantly high number of cases compared to the full CSI, which can spare patients from discomfort and cut down on time of the study, as well as costs. Our results suggest that use of the CSI algorithm in suspicious cases for mild CTS will improve diagnostic yield and allow patients to be treated earlier. Limitations of our study included the retrospective design and a small sample size. In our study, the use of the CSI algorithm yielded a sensitivity of 60.8%, which is lower than reported in the literature (83%).6 This difference may be attributed to the lower sample size (74 patients vs. 225 patients6) or to the demographic features of the patient population referred to our laboratory for evaluation. We plan to conduct a prospective randomized larger trial over a longer period comparing the CSI algorithm alone versus starting with D2 and using the CSI algorithm only when needed, which will provide more accurate sensitivity and specificity information and be less affected by selection bias. Though CTS is largely a clinical diagnosis, electrodiagnostic studies, including median comparison studies, are routinely recommended.7 Wrist splinting therapy can be effective in mild-to-moderate disease and is recommended before surgical treatment is considered.8 Many practitioners may consider recommending a splinting trial instead of proceeding with an EMG to confirm the diagnosis, but we believe that having a more confirmed diagnosis and excluding other treatable conditions is preferable. The additional sensitivity garnered by the CSI algorithm compared with a standard absolute latency, although more labor-intensive, costly, and potentially uncomfortable for the patient,
allows the diagnosis of more mild early cases. Introducing more median comparison studies from a statistical perspective may increase the false-positive rate, but this rate will be lower if those additional studies point to abnormalities, if the abnormalities are correlated clinically, and if the abnormalities are far from the normal values.9 At the mild early stage, wrist splinting or occupational/lifestyle modifications may be enough to treat CTS, instead of more invasive options including steroid injections or surgical treatments. One outcome study revealed that a higher CSI (2.5–4.6) correlates with better pain and paresthesiae resolution after carpal tunnel release surgery.10 We are currently conducting a retrospective outcome study with mild, moderate, and severe CTS cases, which will assess whether mild cases diagnosed by the CSI algorithm were false positives or not, and a prospective study will follow that. Early CTS diagnosis may overall reduce the costs to society including medical bills, time off of work, and disability and workers’ compensation claims.8 Additionally, an experienced EMG operator can perform the CSI studies quickly, and with the CSI algorithm patient discomfort may be lessened.
REFERENCES 1. Atroshi I, Gummesson C, Johnsson R, et al. Prevalence of carpal tunnel syndrome in a general population. JAMA. 1999;282:153–158. 2. Werner RA, Andary M. Electrodiagnostic evaluation of carpal tunnel syndrome. Muscle Nerve. 2011;44: 597–607. 3. Preston DC, Shapiro BE. Median neuropathy at the wrist. In: Preston DC, Shapiro BE, eds. Electromyography and Neuromuscular Disorders: ClinicalElectrophysiologic Correlations. 2nd ed. Philadelphia, PA: Elsevier; 2005:255–279. 4. Robinson LR, Micklesen PJ, Wang L. Strategies for analyzing nerve conduction data: superiority of a summary index over single tests. Muscle Nerve. 1998;21:1166–1171. 5. Lew HL, Wang L, Robinson LR. Test-retest reliability of combined sensory index: implications for diagnosing carpal tunnel syndrome. Muscle Nerve. 2000;23: 1261–1264. 6. Robinson LR, Micklesen PJ, Wang L. Optimizing the number of tests for carpal tunnel syndrome. Muscle Nerve. 2000;23:1880–1882. 7. Jablecki CKJ, Andary MT, Floeter MK, et al. Practice parameter: Electrodiagnostic studies in carpal tunnel www.jcnmd.com
Journal of
CLINICAL NEUROMUSCULAR DISEASE Volume 15, Number 4 June 2014
145
146
Journal of
CLINICAL NEUROMUSCULAR DISEASE
Volume 15, Number 4 June 2014
Zeidman et al
syndrome. Report of the American Association of Electrodiagnostic Medicine, American Academy of Neurology, and the American Academy of Physical Medicine and Rehabilitation. Neurology. 2002;58: 1589–1592. 8. Bickel KD. Carpal tunnel syndrome. J Hand Surg. 2010;35A:147–152.
© 2014 Lippincott Williams & Wilkins
9. Dorfman LJ, Robinson LR. AAEM minimonograph #47: normative data in electrodiagnostic medicine. Muscle Nerve. 1997;20:4–14. 10. Malladi N, Micklesen PJ, Hou J, et al. Correlation between the combined sensory index and clinical outcome after carpal tunnel decompression: a retrospective review. Muscle Nerve. 2010;41:453–457.
