Previous studies reported that patients with clinical depression walk more slowly and with less time spent in the swing phase of the gait cycle [6, 7]. The present findings lend support to those reports. Although the differences were not statistically significant in this study, subjects with MDD and subjects with bipolar disorder generally tended to walk more slowly and with a shortened swing phase compared to the control group. Here we report, for the first time, changes in the stride-to-stride variability of these two patient groups. In both MDD and bipolar disorder, significant increases in swing time variability were observed. This finding raises the intriguing possibility that dynamic instabilities in the walking pattern may be a heretofore unrecognized feature of gait in severe clinical depression and related affective disorders, including subjects with MDD and bipolar disorder. To our knowledge, this is the first quantitative investigation of gait in bipolar disorder patients.
Although not necessarily apparent to the clinical observer, the present quantitative study suggests that severe clinical depression and related major affective disorders may be associated with an unsteady gait, with certain features reminiscent of impairments in Parkinson's disease. There are, however, subtle but important differences. In advanced Parkinson's disease [18, 19], gait speed and swing time are typically reduced to a much greater extent than that seen in the present study (e.g., one study reported a mean gait speed of 1.0 m/sec and a mean % swing time of 33.5 % in subjects with Parkinson's disease [19]; compare with Table 1). In Parkinson's disease, increased stride-to-stride variability is seen in the swing time (as in the present study), but the magnitude of the increase in swing time variability is larger than that seen in MDD and bipolar disorder [19]. For example, a previous study found that the mean stride time variability and swing time variability were 4.4 % and 5.7 %, respectively, among patients with Parkinson's disease [19], values that are much higher than those observed in the present study (compare with Table 1). In addition, whereas in Parkinson's disease increased variability is observed in the stride time as well as in swing time, in MDD and bipolar disorder the gait impairment appears in swing time regulation and the alterations are smaller than those observed in Parkinson's disease.
In some previous studies that have examined both stride time variability and swing time variability, an increase in one measure was accompanied by an increase in the other measure [12, 19]. This observation is not surprising since swing time is a key component of gait cycle timing and a sub-phase of the stride time. An interesting question is why increased swing time variability was observed among the patients with MDD and bipolar disease in the present study, while stride time variability was not significantly increased? A number of explanations may be possible. Our findings suggest that in some circumstances swing time variability may be a more sensitive marker of subtle underlying changes in neural control and in the locomotor system's ability to repeatedly and precisely generate and execute neural commands. While stride time variability evaluates consistency of the loading response, as reflected in the variability in the time between consecutive heel-strikes, swing time variability reflects the stride-to-stride consistency of both the loading and unloading of body weight. Thus, when the regulation of gait is largely intact and stride-to-stride variability is relatively small, as in the present study, changes in swing time variability may be present, while stride time variability remains largely unchanged.
Another related explanation for the discrepancy been stride time variability and swing time variability has been suggested by Gabell and Nayack [24]. They proposed that stride time variability reflects gait patterning mechanisms and the rhythmicity of locomotor pacing while variability of stance and swing time more closely mirror dynamic equilibrium and postural control mechanisms. A complete explanation for the difference between swing time variability and stride time variability requires further investigation. We note, however, that such discrepant behavior among different measures of gait variability has also been observed in other studies [25–28]. For example in a study of patients with very mild Parkinson's disease [27], swing time variability was significantly increased while stride time variability was only slightly larger than control values.
In older adults, depression is associated with an increase in the risk of falls and hip fractures [1, 3]. The mechanisms responsible for this increase are not clear. The present findings suggest that depression and bipolar disorder may be associated with an increase in stride-to-stride variability which may, in turn, predispose to falls [10, 12, 20, 21]. Young and middle-aged adults who are depressed generally do not have an increased fall risk, probably because depression by itself does not increase stride variability enough to cause falls and because other physiologic systems that help maintain gait and balance stability are generally intact. In older adults, falls are often multifactorial [1]. If age-associated changes occur in multiple systems (e.g., vision, balance), depression may further contribute to fall risk by increasing stride variability and exacerbating instability. One possibility is that the changes in basal ganglia function in depression [15] increase stride variability, as occurs in Parkinson's disease; however, the mechanisms whereby depression and other affective disorders influence gait and impair the regulation of stride variability remain to be determined.
A previous study of gait in depression did not find any significant differences in the gait of subjects who were taking medications compared to those who were not [6]. Specifically, gait speed, stride time and swing time were apparently not related to medication usage. For these specific features of gait, our results support the previous findings. Gait speed and stride time were not related to medication usage in the subjects with bipolar disorder or MDD in the present study. Conversely, while swing time variability was increased relative to control values even in the subgroup of bipolar patients who were not taking any medications, among the patients with bipolar disease and the patients with MDD taking certain medications, variability of gait measures tended to be further increased. Although Lemke et al. did not find a medication effect on gait in patients with depression [6], the suggested relationship between gait and medication usage in the present study in the subjects with MDD is consistent with a recent study which reported that certain antidepressants may impair psychomotor abilities and integrative central nervous system function [29]. However, it is not apparent why such an effect was not observed among patients with bipolar disease and why antidepressants might affect the swing time variability of patients with MDD, but not of patients with bipolar disease.
This preliminary study has a number of limitations. As noted, subjects in the control group were generally younger than the two patient groups (mean differences of 4 and 7 years). However, a number of considerations suggest that the observed increase in swing time variability in the patient groups was not simply a function of aging. 1) Subgroup analysis demonstrates that the differences between the groups persist even when only the relatively young subjects (< 40 years of age) are studied. 2) Like the variability of gait speed and stride length, previous work has shown that the variability of stride time and swing time is similar in healthy young and healthy older adults, and that increases in older adults are a sign of pathology [24–26, 30–32]. For example, one study of the effects of physiologic aging on gait found stride time variability was essentially identical in a group of healthy young adults (mean age 24.6 years) and healthy older adults (mean age 75.7 yrs) [31], with values similar to those observed in the present study.
The present study has other limitations and raises additional questions that should be addressed in future studies. Among subjects with MDD, we could not evaluate if there were differences between subjects treated for acute illness and those with more chronic symptoms. Further, among patient with bipolar disorder, we did not control for the patients' acute mood status. The present study may have also been underpowered to detect small differences between the patients with bipolar disorder and MDD. A prospective study of a larger number of patients, perhaps in different age groups before and after they undergo different types of therapy, including pharmacologic interventions, should be helpful in leading to a more complete understanding of the relationships among stride variability, fall risk, and medication usage. Such studies may also be helpful in determining the mechanisms whereby depression and related mood disorders influence stride variability, in assessing how this relationship changes over time and with respect to disease state (i.e., acute or stable symptoms), and in evaluating if stride variability measures may prove clinically useful as an objective measure of major affective disorders. Stride variability is less affected by a conscious effort to alter gait and more closely reflects intrinsic underlying impairment [10]. Additional biomechanical studies may also be helpful in elaborating the mechanisms underlying gait changes in affective disorders and possible causal links between swing time variability, gait instability, and the risk of falls.