Temporal Patterns
Diurnal patterns in anxiety and depression scores in the highly and medium depressed groups were explored applying a linear regression technique to the entire sequence of beeps per subject and collapsed over subjects per group. For the moderately depressed group, no daily pattern was found for anxiety or depression symptoms, but for the highly depressed group, a within-day significant fluctuation was found for depression (t = -2.06, p < 0.04). This significant effect was produced by one subject who demonstrated a strong late in the day mood upswing (t = -3.93, p < 0.001). This finding was supported by a trend in the other four subjects. Interday differences per subject were also detected, but at the group level this effect disappeared. We next asked if we could separate the two groups on other temporal dimensions such as the recovery rate or decay time from high anxiety or low mood states. To demonstrate this, we tracked the flow of the symptom within the day from a point minimally one SD below that individual’s mean mental state on anxiety or depression. The "symptom event" was followed over subsequent signals that day to manufacture a recovery or decay curve of intense emotional anxiety episodes for each group. Figures 2 to 4 show these recovery curves in the course of the day in relation to the group mean. Figure 2 depicts temporal pattern of recovery from anxiety states in medium and highly depressed subjects. On this graph we see that the high depression group recovers from anxiety more slowly. We also note that at the end of the day, a rebound or a diurnal decrease in anxiety is reported by the high depression group. This finding is greater than may be expected from the regression to the mean shown by the moderately depressed group. In Fig. 3, the recovery from a depressive episode is depicted for both high and medium groups. In contrast to anxiety, a more rapid recovery is noted in both groups as well as a more erratic recovery pattern that does not distinguish the groups from each other. Since the two groups are differentiated by the degree of depression scores, it is interesting to note that the temporal moment-to-moment recovery from a depressive episode is not altered by increased severity. On the other hand, it is anxiety that recovers more slowly in the highly depressed group in addition to showing a late-day rebound. This suggests at least an interplay in daily life between the effects of anxiety and affect. Affect perhaps predisposes the individual to problems in the experience of and coping with anxiety. 
A second interesting aspect of the data is the counterintuitive finding that depression in both ill groups decays rapidly and anxiety more slowly. This rapid recovery of mood and the slower recovery of anxiety is at odds with the view that mood stability or slow recovery is generally a key diagnostic criterion in depressive states. In Fig. 4, we replicate the above graphs for 10 normal controls by comparing the high anxiety and depressed points reported by this group. Rather than a gradual decrease of anxiety, we observe here a precipitous return to the mean for anxiety, whereas depressive feelings returned to the mean at an expected slower rate. This suggests that the recovery of mood and anxiety states occurs in different ways and at different rates across groups varying in the type and severity of psychopathology, a not uninteresting finding for planning treatment. These data illustrate the utility of examining the temporal properties of mental states, but a larger number of subjects will be required to describe the processes more fully and reliably. We further sought to roughly determine if mood states or anxiety states predicted one another. To do this we carried out a cross-lag analysis to determine 
whether low mood or high anxiety scores had a temporal relationship. At lag-0 anxiety and depression are significantly correlated with each other for the high depression groups (r = 0.42; S.E., 0.131; p < 0.001). The analysis of further lags in the high group demonstrated a nonsignificant trend toward depression predicting anxiety. For the medium group, no trends were found. Life history data indicate that panic disorder precedes depressive symptoms (Cassano, Bellodi, this volume), hinting perhaps that anxiety would predict depression in the long run. In the moment-to-moment assessments, this seems not to be the case. The rough trend shown here of depression predicting anxiety is further supported by the temporal recovery curves in which anxiety plays a critical role in the highly depressed group. These data suggest again that depression may be a crucial factor influencing or altering an individual’s capacity to experience or cope with anxiety.