Tremor and hand-arm vibration syndrome (HAVS) in road maintenance workers

The objectives of this study were to evaluate postural and rest tremor among workers exposed to hand-held vibrating tools, taking into account the possible effect from lifestyle habits such as alcohol and tobacco consumption. A further aim was to study tremor parameters in workers who were diagnosed with HAVS at the time of the examination and to investigate factors that might increase the likelihood of being diagnosed with HAVS.

In order to validate the r.m.s. levels of acceleration according to the values in the “Hand-arm vibration guide to good practice” given as 17 ms⁻² r.m.s. for rock drills and 7 ms⁻² r.m.s. for impact wrenches (Griffin et al. 2006), 44 field measurements of the acceleration levels of rock drills and impact wrenches were performed within the present study.

The vibration exposure level during rock drill operation varies with several factors. Among these are the need to use a fixed hand grip to control and press the device toward the drill hole, the angle with the gravity line at which the drill operates, the use of a jack leg, hand jack or iron weight to generate force toward the drill hole, and the hardness of the rock (Table 1).

Measurements of representative acceleration levels during the operation of rock drills across the variable conditions mentioned above were performed according to ISO 5349-2, using recording accelerometer equipment with adaptors interposed between each hand of the operator and the tool, within the glove. Measurements on different rock drill equipment were taken, and average exposure time was estimated based on observations and interviews with the workers. This was done in order to record the actual contact time and integrate the acceleration level that was transferred through the adaptor to the hand during operation (Table 2).

The measurements were taken with two different triaxial accelerometers. The SEN021F accelerometer with a T-adaptor fixed between fingers 2 and 3 connected to a Larson Davis HVM 100 logging vibration level meter with Blaze software (Larson Davis, Depew, NY 140,243, USA) was used for this purpose. Further the accelerometer SV105A with a palm adaptor connected to a logging vibration level meter Svantek SV106 with Swan software (Svantek, 04-0872 Warszawa, Poland) was used. The r.m.s. values of acceleration were computed by integrating software in the measurement equipment, using the area under the curve of frequency-weighted acceleration levels during the time of contact with the tool as the variable of interest.

For the two main exposure categories, the rock drill and the impact wrench, a cumulative dose of acceleration level times lifetime exposure time for each individual was estimated, using the formula Σa _wi ·t _i where a _wi  = weighted acceleration for each tool (i) in ms⁻² r.m.s. and t _i  = the lifetime exposure to vibration for each tool in hours (h) (Bovenzi et al. 1995; Edlund et al. 2014).

The number of workdays was set to 180 per year according to the company’s work schedule. The cumulative exposure was used for exposure in models of associations with outcomes. Information about work with other handheld vibrating tools in present and past employments and in leisure time was also obtained from questionnaires. For these tools, we could not assess the actual exposure which was represented by self-reported data only. The use of other tools was therefore represented by self-reported number of cumulated hours in exposed work (without vibration level) in the models. A dummy variable, based on the use of other hand-held vibrating tools >100 h (yes/no) was made, and this variable was used as an indicator variable in the modeling of exposure-outcome associations.

The subjects were examined using international consensus criteria (Brammer and Lundström 1995; Olsen and Hagberg1995; Griffin and Bovenzi 2007; Negro et al. 2008). A vascular HAVS diagnosis was based on a history of finger whiteness reported by the patient and a clinical examination with standardized color chart and finger mapping of vibration-induced white fingers. A neuro-sensorial HAVS diagnosis was based on a neurological examination, including finger mapping of numbness and tingling, vibration perception thresholds on the index and little fingers, and examination of hand function based on a questionnaire. The questionnaire included questions about easily loosing objects, difficulty in fastening of buttons, opening a tight jar lid, picking up coins from a flat surface, pouring liquid from a jug, turning a door knob lever, and putting on a jacket or pullover. As a supplement, a test for manual dexterity (Grooved Pegboard) was applied. The cut-off limit for reduced dexterity measured with the Pegboard test was set to 1 SD below mean (Grooved Pegboard Instruction Manual, Lafayette Instrument Company). Other potential causes for a vascular or neurological diagnosis, e.g., primary Raynaud syndrome, nerve entrapment, injuries or musculoskeletal diseases, were also considered. At the time of the clinical examination, the clinician made a temporary assessment of exposure. The diagnosis was mainly based on the clinical picture of HAVS, as described above.

Biological samples were collected on the day of the neurobehavioral examinations. Whole blood was collected from the cubital vein with 8 mL Vacutainer tubes without additives (BD Vacutainer, Belliver Industrial Estate, Plymouth, UK). The serum was separated by centrifugation at 2000 g for 10 min. Four samples were pipetted into 4.0 mL NUNC^® polypropylene cryotubes (Sigma-Aldrich, St. Louis, Missouri, US) and frozen for storage at the National Institute of Occupational Health, Oslo, Norway at −80 °C until analysis. Due to time pressure and capacity problems on the day of examination, blood samples were available for only 95 subjects.

Carbohydrate-deficient transferrin in the serum was measured at Fürst Medical Laboratory (Oslo, Norway) by capillary electrophoresis using CapillarysTM (Sebia Inc., Georgia, USA). The method’s limit of detection (DL) was 0.4%. A level of <1.7% is considered normal by the laboratory (Ellingsen et al. 2014). Samples for the glycated hemoglobin (HbA1c) measurement were collected in EDTA tubes, and these samples were also analyzed at Fürst Medical Laboratory (Oslo, Norway). A level of 4.0–<6.1% is considered normal by the laboratory.

The sample preparation of cotinine, caffeine and nicotine in serum has previously been described (Ellingsen et al. 2006). However, in this study, two internal standards were used rather than one. To 0.5 mL serum aliquots, we added 100 µl 0.0025 mg mL⁻¹ internal standard solution, containing caffeine-¹³C₃ and cotinine-(methyl-d₃). A Waters CapLC System (Milford, MA, USA) was used to separate the analytes. A Quattro LC tandem quadrupole MS with positive electrospray ionization (ESI, Micromass, Manchester, UK) was used to separate the analytes and mass spectrometric (MS) detection. Nicotine, cotinine, cotinine-(methyl-d₃), caffeine and caffeine-¹³C₃ were monitored as product ions of their respective [M+H]⁺ molecular ions with m/z transitions of 163–132, 177–80, 180–80, 195–138 and 198–140, respectively. The mass spectrometric settings have been described (Ellingsen et al. 2006).

Cotinine, caffeine and nicotine were quantified by adding internal standards and relative comparisons to spiked serum blank samples that were prepared identically. The methods were evaluated over the concentration ranges of 30–20,000 µg nicotine L⁻¹ serum, 7.5–15,000 µg caffeine L⁻¹ serum and 1.5 to 1000 µg cotinine L⁻¹ serum, displaying a coefficient of correlation >0.996. The within-assay (n = 6) and between-assay (n = 6) precision for nicotine, caffeine and cotinine were <27, <36 and <11%, respectively.

The DL for nicotine was 31 µg nicotine L⁻¹ serum; for caffeine, it was 2.1 µg caffeine L⁻¹ serum; and for cotinine, it was 1.9 µg cotinine L⁻¹ serum. The DL was defined as 2 × standard deviation of the blank.

Continuous variables with a skewed distribution (skewness >2) were log₁₀-transformed. The log-transformed values were used in the statistical analysis, and the arithmetic means with standard deviations (SD) for these variables (cumulative dose of exposure, Tremor Intensity of postural tremor for both hands and sCDT) are also presented. Cumulative exposure was log-transformed and treated as a continuous variable. In order to log-transform the values that are equal to zero exposure, their exposure values were set to 1 ms⁻² r.m.s. h.

Exposure to vibrating tools other than rock drills and impact wrenches was treated as dummy variables in the analysis to assess whether these exposure indicators, even though they were not precisely classified using quantified exposure time, were associated with tremor or, alternatively, confounded the associations between exposure and tremor. A covariate was not regarded as a confounder if the influence on the association between exposure and possible effect was <10%.

Multiple regression analysis was used to assess the age-adjusted associations between cumulative exposure and tremor parameters, all subjects being stratified into users and nonusers of tobacco products, based on the individual concentrations of cotinine in serum (≥5 µg L⁻¹). The effects are shown as beta-coefficients (Table 3). A paired t test was used to compare postural tremor variables with rest tremor variables. ANOVA was used to compare the subjects who were diagnosed with HAVS with those who were not. Three subjects unexposed to the main tools, and with uncertain information about the underlying exposure, but with a clinical picture resembling HAVS, were not included in the analysis of exposure-outcome associations with HAVS. Logistic regression was used to analyze association between exposure and a diagnosis of HAVS, adjusted for covariates.

The statistical analyses were performed with IBM SPSS^®, version 22.0, (IBM Corporation, New York). The level of significance was set at p ≤ 0.05.

Cumulative exposure was calculated from acceleration levels of the tools. The levels used were as given by Griffin et al. (2006), i.e., 17 ms⁻² r.m.s. as the typical level for rock drills and 7 ms⁻² r.m.s. for impact wrenches. These levels were confirmed using field measurements of vibration magnitude and questionnaire-based assessment of time used to operate the main handheld tools. Cumulative exposure was used as the exposure measure. The validation of vibration exposure levels using field measurements is considered a strength of this study.

The reliability of exposure–response relationships in studies of vibration-exposed workers has been questioned (Burström et al. 1998) in that studies have shown both over- and underestimation of the predicted risk (Burström et al. 2006). One study reported a large discrepancy in reported exposure time when comparing the answers on a questionnaire with a structured interview that included questions regarding estimated hand-held vibration exposure (Gerhardsson et al. 2005).

The CATSYS test system has been used in a number of studies of nervous system effects of exposure to neurotoxins but in only one published study of tremor in hand-arm vibration-exposed subjects (Edlund et al. 2015). In the present study, the testing time was set to 16.4, 2 s to stabilize and 14.4 s for recording, which is equal to the recording time applied in a number of studies (Edlund et al. 2015; Ellingsen et al. 2014; Wastensson et al. 2012).

Cumulative exposure to hand-arm vibration was statistically significantly and positively associated with increased postural and rest tremor among smokers and users of smokeless tobacco. The Center Frequency of the postural tremor, but not the rest tremor, was also significantly associated with cumulative exposure and with consumption of tobacco products (Table 3). These finding are in accordance with a study by Edlund et al. (2015) in which nicotine use was associated with increased postural Tremor Intensity and with higher tremor frequency, although the latter was found only for the dominant hand. Studies of manganese-exposed workers have shown increased tremor associated with the self-reported consumption of tobacco products (Bast-Pettersen et al. 2004, 2005) and with self-reported tobacco consumption as well as with cotinine levels in serum and urine (Ellingsen et al. 2006). Whether the observed associations between tobacco consumption and tremor parameters are related to a nicotine effect or to withdrawal from nicotine cannot be decided in the present study. However, we can document an association between tobacco consumption and tremor parameters.

Cotinine has a half-life of approximately 16 h compared with nicotine’s 2 h half-life, and therefore, cotinine values are better estimates of present nicotine use in the final hours prior to the serum sampling (Davis et al. 2015; Hukkanen et al. 2005). To our knowledge, this is the first study of tremor in vibration-exposed workers in which consumption of alcohol, nicotine and caffeine are assessed using biomarkers of exposure rather than by self-report. Edlund et al. (2015) used an interview to assess smoking habits and divided the subjects into nicotine users and nonusers. Self-report of alcohol and tobacco consumption has a number of weaknesses compared with biomarkers of this consumption. Gorber et al. (2009) found that self-report underestimated the true smoking prevalence in most studies included in a systematic review.

Because none of the participants in the present study had sCDT values exceeding the upper reference limit of the laboratory (>1.7%), there was no indication of excessive alcohol consumption at a level of daily consumption of 60–80 g of ethanol or more in any of the participants (Bortolotti et al. 2006). Thus, an influence of heavy drinking on the results is not suspected in the present study.

The results confirm that activation conditions have a large impact on tremor parameters. When comparing postural and rest tremor parameters, we found, in addition to the expected lower rest Tremor Intensity, a higher Center Frequency for the rest tremor. Further, the rest tremor had a narrower Frequency Dispersion, indicating that the power was concentrated at a narrower range of frequencies (Table 5). It has been suggested that rest tremor represents “a central tremor” (Deuschl et al. 2001) that to a larger extent is dominated by the 8–12 Hz oscillator. It is also possible that the recorded tremor is more affected by a component of finger tremor than hand tremor in the resting position, and finger tremor has been reported to have a higher frequency than hand tremor (Deuschl et al. 2001).

Because rest tremor has been proposed to represent a more “central tremor” (Deuschl et al. 2001), our findings of a strong association between cumulative exposure and increased postural tremor with a higher frequency and a weaker association between rest tremor and cumulative exposure dose could indicate that the increased tremor in the exposed group represents a “peripheral tremor.” Further, the lack of associations between tobacco consumption, cumulative exposure and rest tremor Center Frequency could indicate that the increased postural tremor Center Frequency among the smokers and users of smokeless tobacco was also due to a “peripheral tremor” effect from nicotine consumption among hand-arm vibration-exposed subjects.

The nine subjects exposed to the main tools who were diagnosed with HAVS had a Tremor Intensity of 0.24 ms⁻² (dominant hand) and 0.23 (non-dominant hand), while the subjects without HAVS had a Tremor Intensity of 0.13–0.15, a difference that was statistically significant (Table 6). Edlund et al. (2015) found tremor intensities of 0.14 (dominant hand) and 0.12 (non-dominant hand) among 139 workers exposed to HAV, but these workers had not been diagnosed with HAVS. When comparing the Tremor Intensity of the subjects diagnosed with HAVS with subjects from other studies of working populations, the magnitude of the tremor was much higher than in subjects exposed to neurotoxins such as mercury (Bast-Pettersen et al. 2005; Wastensson et al. 2006) or manganese (Bast-Pettersen et al. 2004; Blond and Netterstrøm 2007; Ellingsen et al. 2008, 2015; Wastensson et al. 2012). In the present study, the subjects without HAVS had a Tremor Intensity similar to the values in the above-mentioned studies. Thus, a Tremor Intensity of 0.23–0.24 among the HAVS subjects is of clinical interest. The findings suggest that hand tremor may be a part of the clinical picture of HAVS, at least for hand-arm vibration-exposed subjects who are also tobacco users.

The subjects who were diagnosed with HAVS had a higher cumulative exposure to handheld vibrating tools than did the subjects without HAVS. However, there was considerable variation in cumulative exposure (Table 6). The concentrations of CDT found in the present study did not indicate heavy alcohol consumption in any subject, and this factor therefore probably did not influence the likelihood of a diagnosis of HAVS in the present study.

The vibration-exposed workers’ concentrations of cotinine and nicotine in serum were considerably higher than among the unexposed, indicating a higher consumption of tobacco products. One explanation for this could be that nicotine acts as a vasoconstrictor on small blood vessels (Powell 1998). A HAVS diagnosis, which to a certain degree is based on a clinical examination with a color chart and finger mapping (of vibration-induced white fingers), could therefore be affected by the subjects’ consumption of tobacco.

The likelihood of being diagnosed with HAVS was statistically significantly associated with cumulative exposure (Table 7). Higher cotinine concentrations as a biomarker of smoking and use of smokeless tobacco, although nonsignificantly associated with the likelihood of a HAVS diagnosis, were not confounding the associations between cumulative exposure and the likelihood of being diagnosed with HAVS.

The high participation rate (97%) indicates that the findings in the present study are representative for male road maintenance workers with similar work tasks and exposure duration.

Other strengths of the study were that tremor was measured with a standardized tremor test and that potential confounders such as alcohol consumption, use of nicotine products and consumption of caffeine-containing drinks were assessed using biomarkers rather than assessment by self-report.

Because all of the participants were men, there were no potential gender effects on the tremor results, such as differences in tolerance to nicotine or alcohol that could confound associations between exposure and outcome. However, caution is warranted if these findings are to be generalized to female subjects.

Some cautions must be made when interpreting the exposure parameters. We used the same main tool vibration level for all subjects who used the respective tools. It is, however, possible that individual factors such as working technique and grip force has an influence on the vibration exposure on the individual level. Such differences could impose a misclassification of exposure. It has also been debated whether the frequency weighting according to ISO 5349-2 that was used in the present study is the most appropriate measure of exposure to hand-arm vibration (Brammer and Pitts 2012).

We did not have measurement data to assess exposure to hand-held vibrating tools other than rock drills and impact wrenches. Instead, such exposure was introduced into models by indicator variables. These did not show any associations with the outcome. The fact that indicators that represent such exposures in the regression models were not associated with tremor parameters may be a spurious finding, resulting from exposure misclassification that is known to dilute a true association, if present (Rothman et al. 2008).

As with all cross-sectional studies, inference should be made with caution when drawing conclusions about associations between exposure and possible effects. For instance, we cannot rule out the possibility that that there has been a selection out from high-exposed jobs of workers who have experienced untoward effects from exposure to hand-held vibrating tools. Such selection, equal to a healthy worker effect, due to exclusion from exposure, may dilute the true associations between exposure and outcome. Future research using a longitudinal design with better control over such selection phenomena may validate the findings of the present study.

Despite the limitations that a cross-sectional design entails, our findings indicate that exposure to vibrating hand tools was related to the differences in hand tremor between the exposed and the referents.