Guest, Maya BOHS, BMedSci (Hon); Attia, John R. PhD, MD; D'Este, Catherine A. PhD; Boggess, May M. PhD; Brown, Anthony M. MPH; Gibson, Richard E. GradDip (Med Stats); Tavener, Meredith A. PhD; Ross, James MPH; Gardner, Ian MPH; Harrex, Warren MSc (OccMed)
Peripheral neuropathies encompass a wide spectrum of clinical disorders associated with a large number of possible causes including the more well-known occupational neurotoxins: the hexacarbons n-hexane and methyl-n-butyl-ketone; the metals lead, mercury, arsenic and thallium; the organophosphate pesticides; and the gases carbon monoxide, carbon disulfide, and ethylene oxide.1
Two basic forms of damage to peripheral nerves can occur as a result of exposure to neurotoxins: segmental demyelination and axonal degeneration. Axonal degeneration may involve large fibers, with prominent vibratory sense impairment or small fibers with pain and temperature impairment. With remyelination, recovery is rapid and generally complete; however, axonal degeneration is very slow and often incomplete.1,2
In 1976, concern was raised about the possible chronic effect of exposure to jet fuel on the nervous system including peripheral neuropathy.3 More recently, studies have investigated the association between deficits in sense of vibratory perception, peripheral neuropathies, and exposure to solvent mixtures and solvent-based paints.4–7 Demers4 observed vibration perception deficits in commercial painters exposed to organic solvents compared with an unexposed group, measured as vibration perception thresholds (VPT). Broadwell6 observed an association between chronic low-dose solvent exposure and measured VPT in a population of solvent-exposed microcircuit production workers.
Whether a chemical elicits neurotoxic effects depends not only on the inherent characteristics and site specificity of the compound but also on a number of physiological and environmental factors.8 An individual's rate of metabolism can influence the response to chemicals that have to be activated to toxic metabolites after absorption in the body.9 Similarly, chemical and physicochemical interactions may alter the nature and/or the magnitude of toxic responses after exposure to combinations of pollutants. For example, methyl ethyl ketone, a compound devoid of neurotoxic potential, can potentiate the peripheral nerve damage caused by n-hexane.10,11
Several testing procedures can be applied to investigate the effect of occupational exposures on peripheral nerve function, including nerve conduction studies (NCS), VPT and two-point discrimination. NCSs are the gold standard in clinical practice for diagnosing neuropathy. The most commonly adopted method in occupational epidemiological studies is VPT as it is a painless, noninvasive, and detects early change.12 Numerous studies have used VPT for the early detection of vibration-induced neuropathy,13 and metal14 and chemical-related neuropathies.4,6,7,15
One occupational group that has been exposed to solvents, and thus potentially at increased risk of peripheral neuropathy, is F-111 fuel tank maintenance workers in the Royal Australian Air Force (RAAF). The F-111 aircraft does not have dedicated fuel bladders, with fuel instead occupying the empty spaces between other metal structures. Over time, the sealant between internal structures degraded resulting in fuel leaks and it was necessary for the RAAF to set up periodic sealant repair programs, which required dissolving or physically removing the original sealant (desealing) and then replacing it with new sealant (resealing).
In Australia, the RAAF performed four formal F-111 fuel tank Deseal/Reseal (DSRS) programs over more than two decades (1975–1999), each involving different processes and a range of approximately 60 hazardous substances, including jet fuel, a variety of organic solvents, epoxy resins, and various paint formulations.16,17 In 2001, the Department of Defence commissioned an epidemiological study of the health of personnel involved in DSRS activities: the Study of Health Outcomes in Aircraft Maintenance Personnel (SHOAMP). The study was conducted in three phases, between 2001 and 2004, the third of which was a General Health and Medical Study. Sensory and neuropsychological outcomes were a particular focus of the study, due mainly to evidence of their association with organic solvent exposures and the large number of anecdotal complaints about these outcomes in the DSRS workers.18,19
Previously examined as part of SHOAMP have been outcomes of cancer,20 mental health,21 neuropsychological health,22 hearing impairment,23 vestibular function,24 and color vision.25 This current article reports upon vibration sense of workers involved in F-111 fuel tank maintenance relative to appropriate comparison groups.
The SHOAMP was a retrospective cohort study, investigating the possible association between DSRS activities and adverse health status. The study involved a mailed postal questionnaire and a series of clinic assessments with consenting participants. The methods have been reported previously in detail21,22 and are briefly summarized here.
Ethical approval to conduct the study was granted from the following institutional ethics committees: the Human Research Ethics Committee of the University of Newcastle, the Australian Defence Human Research Ethics Committee, and the Department of Veterans’ Affairs Human Research Ethics Committee.
First, the study consisted of one population who reported involvement in one or more of the four formal DSRS activities and working with high levels of jet fuels, paints, solvents, and strippers as well as epoxy resins (the “exposed” cohort, from RAAF Base Amberley, Queensland).
Second, two comparison groups were obtained using stratified random sampling from the computerized Air Force Personnel Executive Management System, with stratification by gender, 5-year age group, posting category, and rank category:
1. Contemporaneous comparisons from the same base (RAAF Base Amberley) who worked in nonaircraft maintenance jobs (“unexposed, same Base, different job”); and
2. Contemporaneous comparisons from a different RAAF Base (Richmond, New South Wales), who did work in aircraft maintenance, but not on F-111 aircraft (“unexposed, different Base, similar job”).
A number of health assessments were performed by a physician, a nurse, and a clinical psychologist taking approximately 3 to 4 hours, at one of eight offices of Health Services Australia, a national provider of health assessments. Vibration perception threshold was assessed by the physician using biothesiometry with a biothesiometer (50–60 Hz; model PVD-LP; Bio-Medical Instrument Co., Newbury, Ohio).
Biothesiometry is a valid, noninvasive, and rapid method of evaluating persons at risk of sensory dysfunction. It is also an efficient tool for screening of workers with significant exposure to neurotoxins or with early sensory symptoms.26 The biothesiometer consists of a rheostat attached to a vibrating probe. As the voltage is increased, the amplitude of the vibration also increases. The intensity of stimulation ranged from a minimum of 0 V to a maximum of 50 V. The voltage is thus a measure of the vibration amplitude.
Participants were asked to remove their shoes, socks, or stockings and lie in a supine position on a bed with knees bent so that their feet were flat. Three sites on each lower limb were used: great toe (distal interphalangeal joint), ankle (medial malleolus), and knee (medial tibial plateau) as well as four sites on each arm: joint of middle finger (distal interphalangeal), hand (third metacarpophalangeal joint), wrist (radial styloid), and elbow (olecranon).
A standard protocol called “method of limits yes-no procedure” was used; this shows an acceptable measure of reproducibility and validity.27 First, a practical trial was given to ensure the participant understood the test. Then the amplitude of the vibrator button was increased by 1 V every 2–3 seconds from the minimum, until each participant reported being aware of a vibrating sensation. The test was conducted three times with the result of the second and third tests recorded.
Vibration Perception Outcomes
The results of VPT were analyzed in two ways: first, quantitatively, as voltage measured directly with the biothesiometer, and second, using a scale suggested by Cornblath et al,28 for the vibration sensation component of the Total Neuropathy Score (a scale of peripheral nerve function). As polyneuropathy occurs first in the nerve fibers, most distant from the brain (ie, the fingers and toes), we report descriptively all sites measured; however, we analyze further only the measures for middle fingers and great toes.
Cornblath et al.28 recommends five categories of result, based on normative values. For ease of interpretation, we added the descriptors. The categories are:
1. Normal to 125% above normal,
2. Slightly reduced sensation—126% to 150% above normal,
3. Moderately reduced sensation—151% to 200% above normal,
4. Severely reduced sensation—201% to 300% above normal, and
5. Extremely reduced sensation—greater than 300% above normal.
Laidlaw and Hamilton29 report normative data for vibratory thresholds in young adults for various dermatomic body areas. The VPT for the great toe is 9 V and the middle finger is 6 V. Using the Laidlaw and Hamilton normative thresholds and the categorization scheme proposed by Cornblath et al,28 categories used for classification of measurement of the great toe were then:
1. 0 to 11.25 V
2. 11.26 to 13.5 V
3. 13.51 to 18 V
4. 18.01 to 27 V
5. 27 V
The categories for the middle finger measurements were:
1. 0 to 7.5V
2. 7.51 to 9V
3. 9.01 to 12V
4. 12.01 to 18 V
5. 18 V
Other Risk Factors
During the clinical examination, a number of other risk factors were assessed. The existence of anxiety and depression was assessed by a clinical psychologist using the Composite International Diagnostic Interview.30 The methods have been reported previously in detail.30 In addition, participants were screened using the Rey-15 item test31 to screen for inadequate effort. Participants with a score of 8 or less were excluded from analyses.32,33 Finally, height and weight were measured by a trained occupational health nurse, as these participant characteristics have previously been reported to have a relationship with VPT of the great toe in particular.34,35
Participants also completed a postal questionnaire, which included self-reported information on general health and well-being, alcohol intake, smoking history, self-report of physician-diagnosed diabetes, current medications, military postings, and a civilian job history calendar. Alcohol intake was coded to four categories according to The Australian Alcohol Guidelines:36 teetotaler and safe drinker, moderate drinker, hazardous chronic drinker and hazardous binge drinker. Smoking behavior was grouped into one of three cate-gories: never smoked, exsmoker, and current smoker. Participants provided a list of current medications, which were later coded by a research pharmacist according to the World Health Organisation's Anatomical Therapeutic Chemical Classification. Of particular interest to this current study was the use of antidepressants; in previous reported papers for this study, population anxiety, depression, and the use of antidepressant medication have been significant risk factors for sensory outcomes.23–25 A civilian job history calendar was used to estimate civilian exposure to organic solvents and lead. These data were obtained by initially classifying jobs to the Australian Standard Classification of Occupations37,38 and translating this into the Finnish occupation codes used in the Finnish Job Exposure Matrix.39 Partici-pants were classified as having an exposure if they had reported a civilian job for which the Finnish Job Exposure Matrix probability of exposure was more than 20%.
Sociodemographic characteristics and potential confounders were compared across the one exposure and two comparison groups using Pearson's chi-square test and analysis of variance. The two outcomes of interest for this study were (1) VPT and (2) clinical categories. VPT was compared across the three groups using linear regression with a natural log transform on the response to ensure normality of the residuals and robust standard errors, clustered on the participant (four observations per participant: two for each of the left and right sides), using an F-test for significance. Clinical categories were compared across the groups using multinomial regression with robust standard errors, clustered on the participant, using a chi-square Wald test for significance.
Linear regression models were used to examine the association between F-111 DSRS exposure and VPT of the great toe and middle finger separately, adjusting for potential confounders. Since this approach uses four observations per participant, robust, clustered standard errors were obtained that adjust for possible correlation among the observations of the same participant.40 A natural log transformation of the VPT was used in both cases to ensure normality of the residuals, thus predicted values are exponentiated to make them comparable to the scale of the original observations. Logistic regression models, with robust clustered standard errors, were used to examine the association of clinical categories to exposure, adjusting for possible confounding variables. The normal versus nonnormal approach was used because of small numbers in some of the nonnormal categories. All variables of interest were initially included in the models, and backward stepwise elimination used to exclude variables, which were not significant at the P = 0.1 level based on the Wald test. Variables with more than two categories, such as smoking and group, were collapsed to two groups if the difference between two of the three was found to be not significant. Odds ratios with 95% confidence intervals (CI) are reported for logistic models. Coefficients with 95% CI are reported for linear regression models. The statistical analysis package STATA v11.1 was used for analysis.41
The response to recruitment and final participation figures have previously been reported,21 thus only a summary is provided here. A total of 872 exposed individuals, 1251 Amberley compari-sons, and 1264 Richmond comparisons were eligible for inclusion in the SHOAMP General Health and Medical Study, of whom a total of 1538 (45%) had a health examination. Four participants did not undertake VPT testing, and 17 participants did not successfully complete the Rey 15-item test and were excluded from the analysis. A small number of participants in each group did not complete all questions in the postal questionnaire; therefore, totals in each analysis vary accordingly. Table 1 presents distribution of characteristics across the three groups.
Table 2 describes the VPT results and shows the median and first and third quartile of VPT for the three exposures of interest. In addition, the number and percentage of measurements in the five categories of clinical severity of VPT for each exposure group are given. Note, totals are four times greater than the number of participants since there are four measurements per participant, two per site on both the left and right sides. Using a linear regression model, a significant difference was only observed between study groups for the hand. The clinical categories for great toe show that 28% of exposed, 26% of the nontechnical comparison group, and 24% of the technical comparison group had 126% and greater above normal VPT. For the middle finger, the proportions were not quite so large (14% of exposed, 12% of both the nontechnical and technical comparison groups). Using a multinomial regression, significant differences were observed between groups for the categories of clinical severity for the middle finger.
Table 3 shows the results of the linear regression model on the VPT of the great toe. Table 4 is the analogous model for the middle finger. The comparison groups, Amberley and Richmond, were not significantly different to each other (great toe –P = 0.4, middle finger P = 0.9), thus the two groups were collapsed and the final model contained one indicator variable for participation in an F-111 DSRS program. Nonsmokers and exsmokers were not significantly different from each other (great toe P = 0.1, middle finger P = 0.2), so the final model contained one indicator variable for current smoker. In the great toe, VPT varied significantly among groups and in the middle finger VPT we observed a difference, which nears statistical significance (P = 0.07). As would be expected, smoking, diagnosed diabetic, age, and height are all significantly associated with the VPT.
The predicted average VPT in the great toe for a 45-year-old nonsmoking male in the unexposed group, 180-cm tall and weighing 90 kg, is 9.3 V (95% CI: 9.0 to 9.6 V). This voltage is estimated at
1. 9.9 V (95% CI: 9.5 to 10.3 V) if involved in the DSRS program,
2. 9.7 V (95% CI: 9.2 to 10.2 V) if a current smoker, and
3. 12.4 V (95% CI: 10.9 to 13.9 V) if diabetic.
It is noticeable that this is approximately a 1-V increase for involvement in a DSRS program and current smokers, and approximately a 3-V increase for a diabetic.
The predicted average VPT in the middle finger for a 45-year-old nonsmoking male in the unexposed group, 180-cm tall and weighing 90 kg, is 4.8 V (95% CI: 4.7 to 4.9 V). This voltage is estimated at
1. 5.0 V (95% CI: 4.8 to 5.1V) if involved in the DSRS program,
2. 5.1 V (95% CI: 4.9 to 5.3V) if a current smoker, and
3. 5.4 V (95% CI: 4.8 to 6.0 V) if diabetic.
The VPT difference associated with involvement in a DSRS program and current smoking is less than one third of a volt, but over half a volt for a diabetic.
Table 5 shows the results of the logistic regression model on the normal versus nonnormal VPT of the great toe. Table 6 is the analogous model for the middle finger. As in the VPT case, the two comparison and one exposure groups were not significantly different to each other (great toe P = 0.2, middle finger P = 0.6) and the nonsmoker and exsmoker categories not different significantly different to each other (great toe P = 0.3, middle finger P = 0.2). We observed significant differences between groups in the odds of abnormal VPT for the great toe but not the middle finger. Once again, smoking, age and height are all remained in the model; however, a diagnosis of diabetes was only significant in the great toe model.
The predicted probability of some loss of vibration sensation in the great toe for a 45-year-old nonsmoking male in the unexposed group, 180-cm tall and weighing 90 kg, is 25% (95% CI: 21% to 28%). This probability is estimated to be
1. 30% (95% CI: 26% to 35%) if involved in the DSRS program,
2. 33% (95% CI: 27% to 39%) if he is a current smoker, and
3. 53% (95% CI: 35% to 71%) if he is diabetic.
The predicted probability of some loss of vibration sensation in the middle finger for a 45-year-old nonsmoking male in the unexposed group, 180-cm tall and weighing 90 kg, is 9% (95% CI: 7% to 11%). This probability is estimated to be
1. 11% (95% CI: 8% to 13%) if involved in the DSRS program,
2. 12% (95% CI: 8% to 15%) if a current smoker, and
3. 13% (95% CI: 6% to 19%) if diabetic.
The aim of this research was to examine possible chronic and persistent peripheral neuropathy in RAAF personnel involved in F-111 fuel tank maintenance, relative to two comparison groups. We did not expect to find clinically relevant and manifest disease as there is a general agreement that studies in occupational populations are not suitable for detecting manifest occupational illness.5 However, our hypothesis was that a solvent-related disease develops gradually from subtle or minor neurotoxic effects. We therefore used biothesiometry to detect abnormalities that had not yet led to functional impairment and would thus remain mostly unnoticed by the affected individuals and examined clinical categories to describe the extent of impairment.
We observed that greater than 26% of all study participants had persistent increased VPT in the great toe (ie, poorer sensation). Statistically, the DSRS-exposed group relative to the two comparison groups differed significantly; however, the magnitude of the difference in voltage was small. Using the clinical categories, the odds of abnormal sensation (including slight, moderate, severe, and extreme neuropathy) was higher in the DSRS-exposed group, relative to the comparison groups. We did not observe any statistically significant relationship between DSRS exposure and VPT of the middle finger. This is consistent with the current biological understanding that neurotoxic effects are generally symmetric with distal nerve involvement1 and the two published studies of shipyard painters exposed to solvents that identified peripheral neuropathy in the feet, but not the hands.42,43
In this study, we report persisting, chronic poorer sensation and long-term effects of solvent exposure. While the majority of the published occupational studies report findings in currently exposed populations, our study population had past exposures and it was these past exposures that were the focus of this article. Our analysis controlled for current occupational exposures to neurotoxins. The effects we observe are persistent and perhaps suggest axonal degeneration, which in severe cases does not regenerate completely.
In all our models age, height, weight, and diabetes were significantly associated with VPT; peripheral neuropathy is a well-known complication of diabetes. In 1996, Cheng et al44 claimed that there was “growing, yet still unappreciated evidence” that height may be an important and predictor of peripheral neuropathy. Our findings add to this body of knowledge. The mechanism of this effect is not completely understood. While some authors have postulated that it takes longer for the signal to reach the brain from the distal extremities because of distance, Cheng suggests the reason could be increased axon surface exposure to toxins.
Study Strengths and Weaknesses
This large cohort study has described the level of peripheral neuropathy in a selected RAAF population within Australia. It is the first Australian study of a military population to be published, which has examined vibration perception in the hands and feet. The strength of this study is the sample size of 1538, being one of the largest studies investigating the relationship between solvent exposure and persistent peripheral neuropathy. In addition, our approach to the statistical analysis of using all data (collected from left and right of the body), rather than the more usual approaches of avera-ging or using the worst observation, has increased the number of observations. The findings relating to height add to the current body of knowledge in this field. A weakness in the study is that it was carried out retrospectively without real-time exposure data for those solvents used in the DSRS programs.
In conclusion, this research demonstrated a statistically significant higher VPT in the great toe in the DSRS population compared with the two comparison groups, although the magnitude of the difference was small. Using the clinical categories, the odds of abnormal VPT (including slight, moderate, severe, and extreme neuropathy) was higher in the DSRS group relative to the comparison groups.
This study reported in this article was funded by the Australian Department of Defence and administered by the Australian Department of Veterans’ Affairs. This article has been reviewed by the Department of Defence and the Department of Veterans’ Affairs prior to publication, and the views expressed are not necessarily those of the Australian Government. The authors declare that they have no conflict of interest.
1. Filley CM, Kelly JP. Sullivan JB, Krieger GR, eds. Clinical neurotoxicity and neurobehavioural toxicology. In: Clinical Environmental Health and Toxic Exposures. 2nded
. Philadelphia: Lippincott Williams & Wilkins; 2001:247–259.
2. Baker EL, Fielder NL. Levy BS, Wegman DH, Baron SL, Sokas RK, eds. Neurologic and psychiatric disorders. In: Occupational and Environmental Health. 5th ed. Philadelphia: Lippincott, Williams & Wilkins; 2006:570–586.
3. Knave B, Persson HE, Goldberg JM, Westerholm P. Long-term exposure to jet fuel: an investigation on occupationally exposed workers with special reference to the nervous system. Scand J Work Environ Health. 1976;2:152–164.
4. Demers RY, Markell BL, Wabeke R. Peripheral vibratory sense deficits in solvent-exposed painters. J Occup Med. 1991;33:1051–1054.
5. Nasterlack M, Dietz MC, Frank KH, et al.. A multidisciplinary cross-sectional study on solvent-related health effects in painters compared with construction workers. Int Arch Occup Environ Health. 1999;72:205–214.
6. Broadwell DK, Darcey DJ, Hudnell HK, Otto DA, Boyes WK. Work-site clinical and neurobehavioral assessment of solvent-exposed microelectronics workers. Am J Ind Med. 1995;27:677–698.
7. Myers JE, Nell V, Colvin M, Rees D, Thompson ML. Neuropsychological function in solvent-exposed South African paint makers. J Occup Environ Med. 1999;41:1011–1018.
8. Schaumburg HH, Spencer PS. Recognizing neurotoxic disease. Neurology. 1987;37:276–278.
9. Perucca E, Manzo L. Galli CL, Manzo L, Spencer PS, eds. Metabolic activation of neurotoxicants. In: Recent Advances in Nervous System Toxicology. New York: Plenum Press; 1988:67.
10. Ralston WH, Hilderbrand RL, Uddin DE, Andersen ME, Gardier RW. Potentiation of 2,5-hexanedione neurotoxicity by methyl ethyl ketone. Toxicol Appl Pharmacol. 1985;81:319–327.
11. Bos MJ, de Mik G, Bragt PC. Critical review of the toxicity of methyl n-butyl ketone: risk from occupational exposure. Am J Ind Med. 1991;20:175–194.
12. Gobba F. Occupational exposure to chemicals and sensory organs: a neglected research field. Neurotoxicology. 2003;24:675–691.
13. Cock N, Piette A, Malchaire J. Can a battery of functional and sensory tests corroborate the sensorineural complaints of subjects working with vibrating tools? Int Arch Occup Environ Health. 2000;73:316–322.
14. Chuang HY, Schwartz J, Tsai SY, Lee ML, Wang JD, Hu H. Vibration perception thresholds in workers with long term exposure to lead. Occup Environ Med. 2000;57:588–594.
15. London L, Nell V, Thompson ML, Myers JE. Effects of long-term organophosphate exposures on neurological symptoms, vibration sense and tremor among South African farm workers. Scand J Work Environ Health. 1998;24:18–29.
16. D'Este C, Attia J, Brown A, et al.. Study of Health Outcomes in Aircraft Maintenance Personnel, Volume 2—Mortality and Cancer Incidence Study Interim Report. Newcastle, Australia: The University of Newcastle Research Associates (TUNRA) and Hunter Medical Research Institute; 2003.
18. Donaldson E. Nature and Extent of Health Complaints, Report to the F111 Deseal/Reseal Board of Inquiry vol. 2, Part 1, Appendix B. Amberley, Australia: RAAF; 2000.
19. Ross J. 501 F111 Fuel Tank Spray Sealing Investigation. Interim Occupational Medicine Report. Canberra, Australia: Department of Defence; 2000.
20. D'Este C, Attia JR, Brown AM, et al.. Cancer incidence and mortality in aircraft maintenance workers. Am J Ind Med. 2008;51:16–23.
21. Attia JR, D'Este C, Schofield PW, et al.. Mental health in F-111 maintenance workers: the study of Health Outcomes in Aircraft Maintenance Personnel (SHOAMP) general health and medical study. J Occup Environ Med. 2006;48:682–691.
22. Schofield PW, Gibson R, Tavener M, et al.. Neuropsychological health in F-111 aircraft maintenance workers. Neurotoxicology. 2006;27:852–860.
23. Guest M, Boggess M, Attia J, et al.. Hearing impairment in F-111 maintenance workers: the study of health outcomes in aircraft maintenance personnel (SHOAMP) general health and medical study. Am J Ind Med. 2010;53:1159–1169.
24. Guest M, Boggess MM, Attia J, D'Este C, Brown A. An observed relationship between vestibular function and hearing thresholds in aircraft maintenance workers. J Occup Environ Med. 2011;53:146–152.
25. Guest M, Boggess M, Attia J, et al.. Colour vision deficiencies in F-111 maintenance workers: the study of health outcomes in aircraft maintenance personnel (SHOAMP) general health and medical study. Int Arch Occup Environ Health. 2011; DOI 10.1007/s00420-010-0600-9.
26. Manzo L, Cosi V. Costa LG, Manzo L, eds. Evaluation and management of neurotoxicity in occupational illness. In: Occupational Neurotoxicology. Boca Raton, FL.: CRC Press; 1998:124.
27. Gerr F, Letz R. Vibrotactile threshold testing in occupational health: a review of current issues and limitations. Environ Res. 1993;60:145–159.
28. Cornblath DR, Chaudhry V, Carter K, et al.. Total neuropathy score: validation and reliability study. Neurology. 1999;53:1660–1664.
29. Laidlaw RW, Hamilton MA. Thresholds of vibratory sensibility as determined by the pallesthesiometer: a study of sixty normal subjects. Bull Neur Instit NY. 1937;6:494–503.
30. World Health Organization. Composite International Diagnostic Interview (CIDI); Core Version 2.1 Interviewer's Manual. Geneva, Switzerland: World Health Organization; 1997.
31. Rey A. L'Examen Clinique en Psychologie. Paris: Presses Universitaire de France; 1964.
32. Goldberg JO, Miller HR. Performance of psychiatric inpatients and intellectually deficient individuals on a task that assesses the validity of memory complaints. J Clin Psychol. 1986;42:792–795.
33. Hiscock CK, Branham JD, Hiscock M. Detection of feigned cognitive impairment: The two-alternative forced-choice method compared with selected conventional tests. J Psychopathol Behav. 1994;16:95–110.
34. Letz R, Gerr F. Covariates of human peripheral nerve function. 1. Nerve conduction celocity and amplitude. Neurotoxicol Teratol. 1994;16:95–104.
35. Skov T, Steenland K, Deddens J. Effect of age and height on vibrotactile threshold among 1,663 U.S. workers. Am J Ind Med. 1998;34:438–444.
36. National Health & Medical Research Council. Australian alcohol guidelines: health risks and benefits. Canberra, Australia: NH&MRC; 2001.
37. McLennan W. Australian Standard Classification of Occupations. Canberra, Australia: Australian Bureau of Statistics; 1997.
38. Australian Bureau of Statistics. ASCO Coder. Version 4.6.2 ed. Canberra, Australia: Australian Bureau of Statistics; 1997.
39. Kauppinen T. From cross-tabulations to multipurpose exposure information systems: a new job-exposure matrix. Am J Ind Med. 1998;33:409–417.
40. Williams RL. A note on robust variance estimation for cluster-correlated data. Biometrics. 2000;56:645–656.
41. StataCorp. Stata Statistical Software: Version 11.0. College Station, TX; 2009.
42. Ruijten MW, Hooisma J, Brons JT, Habets CE, Emmen HH, Muijser H. Neurobehavioral effects of long-term exposure to xylene and mixed organic solvents in shipyard spray painters. Neurotoxicology. 1994;15:613–620.
43. Halonen P, Halonen JP, Lang HA, Karskela V. Vibratory perception thresholds in shipyard workers exposed to solvents. Acta Neurol Scand. 1986;73:561–565.
44. Cheng YJ, Gregg EW, Kahn HS, Williams DE, De Rekeneire N, Narayan KM. Peripheral insensate neuropathy–a tall problem for US adults? Am J Epidemiol. 2006;164:873–880.