Alcohol is a commonly used social drug and driving under influence is a well-established risk factor for traffic accidents [1]. To improve road safety, legal limits are set for blood alcohol concentration (BAC) and driving, usually at 0.05% (most European countries) or 0.08% (most US states, Canada and UK). For other tasks than driving, however, it remains unclear whether these BACs also reflect appropriate safety limits. Recent research showed that among working-aged people, ingestion of alcohol in the previous 6 hours is strongly and consistently related to falls at home resulting in admission to hospital or death, even with low levels of alcohol consumption [2]. The public health impact of falls is substantial and concomitant costs are growing [2, 3]. However, reducing alcohol intake is often not included in intervention strategies to prevent falls. Low alcohol intake is generally not deemed unsafe with regard to falls, but this was never investigated systematically. Accidental falls have been found to be associated with impaired obstacle avoidance skills [4]. To prevent tripping, accurate goal-directed reactions are required to avoid sudden obstacles in the travel path. In previous work we have observed that an increased percentage of obstacles that were hit ("obstacle avoidance failures") is associated with the presence of smaller and later EMG responses in the prime movers (such as the m. biceps femoris) involved in the obstacle avoidance reaction [5, 6]. From cat studies it is known that the parietal cortex and the cerebellum play an important role in this reaction [7, 8]. Imaging studies have shown that acute alcohol administration significantly reduces brain glucose metabolism in these areas that are important for obstacle avoidance [9–11]. Hence, one would predict obstacle avoidance reactions during gait to be disturbed by alcohol ingestion. Therefore we investigated the effect of low to moderate alcohol consumption on such reactions in healthy senior individuals by means of an obstacle avoidance task.

It was hypothesized that obstacle avoidance reactions are already impaired at low BACs, and that the increases in the percentage of obstacles that were hit after alcohol consumption will be accompanied by delayed and decreased muscle responses in the m. biceps femoris.

Before the start of the experiment, the BAC of each participant was 0.00%. After the final drink the BACs found ranged from 0.03% to 0.06%. Hence, we succeeded in our aim to reach BACs around or below common legal limits for driving (0.05% for most European countries or 0.08% for most US states, Canada and the UK).

The series of the obstacle avoidance trials were equally difficult in the three alcohol conditions (A0, A1, and A2), as there were no significant main effects of alcohol condition on the average ARTs (F _2,11 = 0.22, p = 0.81).

Failures

Figure 3A shows the effect of alcohol on avoidance failure rates. Overall, the failure rate increased significantly from 4.5% (A0) to 8.8% (p < 0.01) after consumption of two alcoholic drinks. Figure 3A also demonstrates that in each alcohol condition most failures were made in the MSw phase (p < 0.01, compared to late stance). Within this phase, the failure rate increased significantly with alcohol consumption, from 11.7% in A0, to 19.2% and 20.5% in A1 and A2, respectively (p < 0.01, compared to A0). Moreover, the probability of a failure not only increased with higher BACs, but also with lower ARTs, which corresponded to the MSw phase (Figure 3B). Compared to the placebo condition, chances of hitting the obstacle almost doubled after the final alcoholic drink (odds ratio (95%CI) = 1.93 (1.17,3.18); p = 0.01).

Response time

The results for BF response times, one of the proposed determinants of successful obstacle avoidance, are shown in Figure 4. Repeated measures MANOVA revealed a main effect of alcohol condition on overall BF response times (F _2,11 = 24.93, p < 0.001), as well as for each phase of obstacle release separately (Table 1). After ingestion of 2 alcoholic drinks (mean ± SD: 0.47 ± 0.04 g alcohol/kg bodyweight), BF response times were delayed by 12% compared to when participants were sober (179 ± 5.8 vs 160 ± 4.7 ms, F _1,12 = 53.42, p < 0.001) (Figure 4). From the data of the individual subjects (Figure 5) it can be seen that without exception, BF response times for all participants were delayed following A2. Furthermore, even after A1, 12 of the 13 participants already responded more slowly. The BF response times were significantly delayed from BACs of 0.035% upwards (t = 18.05; p = 0.003). There was a significant correlation between response time and level of BAC (r = 0.6; p < 0.001; Figure 5).

	LSt			ESw			MSw			Total
	Diff(ms)	F 1,12	p	Diff(ms)	F 1,12	p	Diff(ms)	F 1,12	p	Diff(ms)	F 1,12	p
A0 vs A1	8.3	11.0	0.006	9.2	5.6	0.036	5.4	2.0	0.182	8.7	15.0	0.002
A1 vs A2	10.8	12.0	0.005	5.0	1.2	0.287	11.0	18.4	0.001	10.8	22.7	0.000
A2 vs A0	19.2	26.8	0.000	14.2	10.0	0.008	16.4	36.3	0.000	19.4	53.4	0.000

A0 = placebo, A1 = first alcohol, A2 = second alcohol, LSt = Late Stance, ESw = Early Swing, MSw = Mid Swing, Total = all trials, Diff = difference. Bold numbers indicate significance.

A significant effect of obstacle release phase on BF response times was also found. The LSt responses were significantly slower than those in ESw (F _1,12 = 52.65, p < 0.001). In turn, ESw responses were again significantly slower than those in MSw (F _1,12 = 49.86, p < 0.001).

Response amplitude

Figure 6 shows the results of the normalized response amplitudes. Repeated measures MANOVA revealed a main effect of alcohol condition on overall BF response amplitudes (F _2,11 = 4.83, p = 0.03). Post-hoc analyses yielded a significant effect of alcohol condition on response time in LSt, MSw, and in total (Table 2). A trend for decreasing amplitude with increasing BACs can be noted in all step cycle phases (Figure 6).

	LSt			ESw			MSw			Total
	Diff(%)	F 1,12	P	Diff(%)	F 1,12	p	Diff(%)	F 1,12	p	Diff(%)	F 1,12	p
A0 vs A1	18.4	3.5	0.087	15.8	4.6	0.054	17.1	4.7	0.050	16.5	9.1	0.011
A1 vs A2	14.7	11.2	0.006	4.2	2.3	0.158	1.5	2.5	0.142	10.1	9.1	0.011
A2 vs A0	33.1	8.3	0.014	20	4.7	0.052	18.6	5.7	0.035	26.5	10.3	0.007

A0 = placebo, A1 = first alcohol, A2 = second alcohol, LSt = Late Stance, ESw = Early Swing, MSw = Mid Swing, Total = all trials, Diff = difference. Bold numbers indicate significance.

To rule out the possible effect of background activity on amplitude, the background activity was analyzed as well. The normalized background activity did not change significantly with increasing BACs (F _2,11 = 0.24, p = 0.79).

The present study investigated the effect of low to moderate levels of alcohol consumption on obstacle avoidance reactions during gait. The results clearly show that even with low BACs ( < 0.06%), reactions to sudden gait perturbations are seriously affected. After ingestion of 2 alcoholic drinks, obstacles were hit more often, BF response times were delayed and response amplitudes were reduced. These changes were most obvious in situations with little available response time.

This is the first study to investigate the effect of alcohol on responses to sudden gait perturbations (as a relevant task related of falls). Previous studies have concentrated on the effects on posture [16, 17]. Low doses usually increased body sway but in some subjects the inverse was seen, indicating that these doses may have an beneficial effect in some cases [17]. However, it is questionable whether these data are actually relevant for falls since falls rarely occur during standing. Locomotion studies seem to be more relevant in this respect. Mallinson et al.[18] found that it may be possible to detect subtle dynamic imbalance induced by alcohol ingestion (89 ml of 40% alcohol) during tandem walking with eyes closed. In the present study participants had to walk on a treadmill with eyes open after consuming an average of 125 ml of 40% alcohol. Because the background activity was not significantly different between the alcohol conditions, and the obstacle was only released when a regular walking pattern was observed and after at least five unperturbed strides had been taken from the start of the trial, we feel confident that any differences found in the failure rate and any changes in BF activity reflect the effect of the increased BAC.

Earlier research has shown that many falls are primarily due to stumbling and tripping [19]. In order to avoid falls due to hitting an obstacle, one needs to be able to respond adequately to both unseen obstacles causing a stumble [20, 21], and to obstacles suddenly appearing in the travel path [6]. The muscle that shows the first major response in all these reactions is the m. biceps femoris [6, 15, 20, 21]. When compared to young adults, both an increase in response latency and a decrease in response amplitudes of this muscle were found in older adults [6, 22]. These longer onset latencies and smaller amplitudes were associated with lower success rates [6]. The underlying mechanism for the decreased amplitudes during the stumbling and obstacle avoidance reactions in older adults may involve various age-related physiological changes, both in the CNS (e.g. fewer motoneurons) [23] and in skeletal muscle properties (fewer type II muscle fibers and overall muscle atrophy) [23, 24]. In contrast, in the present study the delay and decrease in response are more likely to be due to CNS changes. A possible explanation for the increased failure rate could be that the pathways used in the avoidance reactions have been altered by alcohol consumption. Obstacle avoidance reactions are often very fast; this has led to the proposal that fast supraspinal pathways may be involved that bypass the primary motor cortex [6, 25]. These pathways may involve the parietal cortex and/or the cerebellum. For example in studies on cats, Drew [26] showed that the parietal cortex is involved in obstacle avoidance. They also proposed that both a fast directly descending pathway originating from the parietal cortex may exist along with a slower one involving the motor cortex.

Another possible explanation for the impaired obstacle avoidance skills after alcohol consumption lies at the level of neurotransmitters. For example, previous research has shown that the endorphinergic system [27] and GABA (Gamma-AminoButyric Acid) [28] are intimately involved in the actions of alcohol. The sedative, tranquilizing and/or anaesthetic properties of alcohol have been related to the enhancement of the flow of chlorine ions across neural membranes due to GABA [28]. Yet alcohol does not have this effect on all GABA receptors. Motor incoordination due to ethanol is caused by potentiation of GABA_A-associated adenosine A_2A receptors in the striatum (caudate nucleus & putamen) [28]. Moreover, it is suggested that alcohol-induced deterioration in motor function is linked to changes in patterns of brain activity rather than changes in specific brain regions. Specifically, changed activity in the cerebellum as well as in the frontal and parietal cortices are involved in the motor-incoordinating effects of alcohol [29].

Studies on the effects of alcohol on brain activity with leg movements are lacking. However, for arm movements Van Horn et al. [30] found that the human cerebellum and PPC (Posterior Parietal Cortex) are involved in goal-oriented limb movements and that this role is compromised by alcohol. In particular, alcohol may cause a disturbance in the ability of these brain regions to compute appropriate corrective behavioral responses [30]. In this context it is reasonable to suggest that the presently observed deficits in obstacle avoidance skills may be due to the effect of alcohol on information processing in the PPC and the cerebellum. Experiments involving techniques to record brain activity during obstacle avoidance should be performed to test this hypothesis.

To limit discomfort to the subjects we used a handheld breath analyzer instead of blood samples. In contrast to blood analysis, these breath analyzers are easy and quick to use, do not require additional hard or software, and are not a burden to the participants. The readings of such portable devices are in good agreement with the results of confirmatory analyses performed by stationary devices (r = 0.978) [31]. Furthermore, the correlation with blood analysis is quite high for both the readings of the handheld (r = 0.940) as well as the stationary devices(r = 0.936) [31].

To the best of our knowledge, the effect of alcohol or other substances on obstacle avoidance during gait has never been studied before in healthy senior individuals. Therefore, it is not possible to make a direct comparison with results from similar studies. However, the obstacle avoidance task used in the present study has proven to be sensitive enough to detect significant age-related deficits [32]. A possible limitation is the relatively small sample size. However, in this type of motor control studies it is quite usual to have similar group sizes (because of the extensive data analysis involved). Furthermore, even with the small number the study yielded unequivocal outcomes. Hence, a larger sample size will mostly accentuate the significance of the present results.

In conclusion, the present results clearly show that alcohol levels, considered to be safe for driving, seriously hamper the ability to successfully avoid sudden obstacles in the travel path. It is suggested that many of the alcohol-related falls are the result of the disruptive effects of alcohol on the online corrections of the ongoing gait pattern when walking under challenging conditions. In general the use of alcohol is primarily seen as a risk factor for driving [1, 33]. However, Kool et al. [2] estimated that approximately 20% of unintentional falls at home in a working-aged population may be attributable to the consumption of two or more standard alcoholic drinks in the preceding 6 h. Moreover, accidents can also occur while walking, particularly under challenging conditions such as when negotiating suddenly appearing obstacles. The present data show that the required skills for obstacle avoidance frequently fail even after consumption of a low dose of alcohol.