Non-verbal interaction involvement as an indicator of prognosis in remitted depressed subjects

Psychiatry Research 113 (2002) 269–277

Non-verbal interaction involvement as an indicator of prognosis in remitted depressed subjects Elisabeth H. Bos*, Erwin Geerts, Antoinette L. Bouhuys Department of Psychiatry, University of Groningen, P.O. Box 30001, Groningen 9700 RB, The Netherlands Received 11 February 2002; received in revised form 3 October 2002; accepted 14 October 2002

Abstract Fifty-one remitted depressed inpatients and their interviewers were observed during a conversation. We investigated whether non-verbal behavioral elements indicative of involvement displayed by the remitted patients andyor their interviewers were predictive of depressive symptoms 6 months later. Involvement behavior of the patients appeared to be related to future complaints; the lower the level of involvement displayed, the more unfavorable the outcome. We interpret these results with reference to concepts of social support. 䊚 2002 Elsevier Science Ireland Ltd. All rights reserved. Keywords: Depression; Relapse; Non-verbal behavior; Engagement; Social functioning

1. Introduction Depression is frequently associated with interpersonal dysfunctioning. Several theories have been formulated linking interpersonal factors to depressive disorder (Lewinsohn, 1974; Coyne, 1976; Coyne et al., 1990; Gotlib and Hammen, 1992). Social support and well-developed social networks are supposed to protect against depression, to facilitate recovery and to prevent recurrence. Problems in interpersonal functioning and lack of social competence are thought to lead to interpersonal stress and withdrawal of the social environment and, via these, to increase risk of depression. *Corresponding author. Tel.: q31-50-361-2062; fax: q3150-361-9722. E-mail address: [email protected] (E.H. Bos).

This credible idea of a link between depression and the interpersonal is supported by a considerable amount of empirical research. Several studies show that depressives have problems in social functioning, and that these problems are associated with poor recovery and chronicity of the depression (for reviews, see Lara and Klein, 1999; Joiner, 2000). Most of these studies, however, merely tell us something about interpersonal problems of patients while they are depressed and about how these problems relate to episode duration. Little is known about social functioning of non-depressed or recovered subjects and its implications for depression onset and recurrence. Evidence that interpersonal factors are predictive of the development of depressive disorders and of relapse or recurrence is scarce (Paykel, 1994; Joiner, 2000).

0165-1781/02/$ - see front matter 䊚 2002 Elsevier Science Ireland Ltd. All rights reserved. PII: S 0 1 6 5 - 1 7 8 1 Ž 0 2 . 0 0 2 6 8 - 8

270

E.H. Bos et al. / Psychiatry Research 113 (2002) 269–277

It is not unreasonable to hypothesize that the factors responsible for interpersonal problems in depressed patients are also the ones that make subjects vulnerable for the development and recurrence of depression. Such would be consistent with the diathesis-stress model of depression, which presumes certain underlying traits (that interact with stressful events) predispose people to depression (e.g. Bebbington, 1987; Perris, 1987; Robins and Block, 1989). Reasoning within the lines of this model, one would expect interpersonal problems not to be confined to the depressive state, but to endure in remission, reflecting an underlying vulnerability. Some longitudinal studies with measurements both in depression and in remission support this idea. Marital discord and lack of close relationships have been shown to continue when the depression is over, and to predict relapse and recurrence (e.g. Rounsaville et al., 1980; Billings and Moos, 1985a,b). Other studies, however, could not establish that the social factors that impede recovery in depressed subjects are also the ones that increase risk of recurrence in recovered subjects (see Barnett and Gotlib, 1988). However, to date, too few longitudinal studies focusing on interpersonal functioning of remitted or recovered subjects have been done to be conclusive. In the present study, we examined interpersonal interactions of patients while the depression was in remission. We wanted to investigate whether there was anything observable in the behavior of remitted subjects andyor their conversation partners that differentiated subjects at high risk for relapse from others. Non-verbal, observable, behavior is an important, but often neglected aspect of interpersonal communication (cf. Burgoon, 1985; Cahn and Frey, 1992). Several nonverbal aspects of the behavior of depressives have been related to problems in social functioning (see Segrin, 1998, for a review). This appears to hold especially for non-verbal correlates of interaction involvement. Adequate levels of involvement behavior seem to be indispensable for the establishment and maintenance of satisfactory interpersonal relationships (cf. Cegala et al., 1982; Bell, 1987; Coker and Burgoon, 1987; Segrin and Abramson, 1994). Subjects showing very low levels of involvement behavior, for example, are

easily experienced as unresponsive, unconcerned and lacking initiative, (e.g. Davis and Holtgraves, 1984; Burgoon et al., 1989). On the other hand, very high levels of involvement are likely to be experienced as a sign of dependency and excessive reassurance seeking (e.g. Harlow and Cantor, 1994). In either case, the reaction of the social environment may be aversion and rejection. Displaying adequate involvement in interactions thus seems to be a matter of a good balance. In line with this, both extremes in involvement behavior of depressives have been related to unfavorable depression outcome. High levels of involvement (Troisi et al., 1989; Bouhuys and Albersnagel, 1992; Geerts et al., 1995), as well as low levels (Zeiss and Lewinsohn, 1988), have been shown to be associated with poor recovery. These findings do not settle the question of how involvement behavior relates to depression, however. The way involvement was operationalized varied in these studies. Moreover, the association between involvement behavior and outcome was not consistently found. For instance, Hale et al. (1997b) could not find such an association. Geerts et al. (2000) did find a relationship between involvement and outcome, but only with regard to the behavior of the conversation partners of the depressives. Most importantly, the above studies again concern the behavior of depressed subjects and its consequences for episode duration. Evidence that the same applies to the behavior of remitted subjects having implications for depression relapse is lacking. Hale et al. (1997a) studied involvement behavior in remitted subjects, but in too small a sample to draw any firm conclusions (ns34). The present study is an extension of the abovementioned study by Hale et al. (1997a). In a larger sample, we investigated non-verbal involvement behavior of remitted patients in a dyadic interaction. The behavior of the patients’ interviewers and the interplay between patients and interviewers were studied as well, since these might also contain prognostic information. Possibly, the influence of interaction partners on each other would reveal the level of involvement, and the degree to which this occurred might relate to depression outcome. Such ‘interviewer’ and ‘interaction effects’ were found

E.H. Bos et al. / Psychiatry Research 113 (2002) 269–277

in previous studies of our research group (Bouhuys and Van den Hoofdakker, 1993; Geerts et al., 1995, 1996, 1997, 2000; Geerts and Bouhuys, 1998). The behavioral elements we focused on in this study were gesticulations, head movements, eye contact, verbal backchannel (i.e. small supportive sounds like ‘hm, hm’ and ‘yes, yes’ emitted during listening) and speech. These behaviors are considered to be indicative of involvement (see Cappella, 1983; Patterson, 1983; Coker and Burgoon, 1987; Bouhuys et al., 1991; Bouhuys and Van den Hoofdakker, 1991). We hypothesized that involvement as reflected in these behaviors would be related to severity of depressive symptoms at 6 months after remission. As we did not know whether this relationship would be a negative or a positive one, we tested this hypothesis bilaterally. 2. Methods 2.1. Subjects and design We studied 51 inpatients recovered from a depressive episode: major depressive disorder, ns 41, dysthymic disorder, ns1, and bipolar disorder (depressed phase), ns9, with psychotic features, ns25 (DSM-IV, American Psychiatric Association, 1994). The diagnosis was made at hospital admission by experienced psychiatrists. We assessed severity of depression by means of the Hamilton Rating Scale for Depression (HRSD, 21item version; mean of two independent raters; Hamilton, 1967). This HRSD interview was done thrice: at hospital admission (T0), at remission (T1) and 6 months after remission (T2). The interviews were conducted between 09.00 and 12.00 h. Interviewers were six researchers, three male and three female, with a mean age of 36.8 years (S.D.s10.8; range 25–50). Patients were included in the present study if their HRSD score was at least 16 at admission and at most 8 at remission. All subjects gave their written informed consent to participate. The group consisted of 35 women and 16 men. Mean age at hospital admission was 46.6 (S.D.s13.1, range 18–71). Average length of hospital stay was 141.7 days (S.D.s70.6, range

271

48–323). At T0, the mean group HRSD score was 24.1 (S.D.s5.8, range 16–38). At T1, the mean HRSD score was 4.3 (S.D.s2.3, range 0.5–8.0). All patients received treatment according to their clinical needs. Most subjects used psychotropic medication at remission: antidepressants; benzodiazepines; neuroleptics; antiparkinson treatment; lithium; or combinations of these. 2.2. Analysis of behavior HRSD remission interviews were videotaped for later analysis of non-verbal behavior. The first 15 min of the videotaped interview were analyzed. Seven trained scorers registered various behavioral elements of the patient and the interviewer by means of an event-recording system. The mean interrater reliability (k; Cohen, 1968) was 0.89 (range 0.68–0.98). Different categories of behavioral elements were recorded in different runs of the videotape. Durations and frequencies of behaviors were registered. The behaviors were analyzed relative to a subject’s speaking or listening, as the occurrence of these behaviors depends on whether one is doing one of the two. We recorded different sets of behavioral elements for patients vs. interviewers. With respect to the patient, we registered gesticulations (all sorts), general head movements, eye contact and speech. With respect to the interviewer, we measured yes-nodding, verbal backchannel and speech. The choice of these specific sets of behavioral elements was based on a factor-analytic study of patient and interviewer behavior done earlier by Bouhuys and associates (Bouhuys et al., 1991; Bouhuys and Van den Hoofdakker, 1991; Geerts et al., 1995). The difference in the collections of behaviors for patient vs. interviewer is a result of the outcome of this factor analysis. This outcome can be seen as a reflection of the different roles interviewers and interviewees have in a conversation, interviewers mainly listening and inquiring, interviewees mainly answering and telling. These two sets of behavioral elements proved to have predictive qualities in several studies of our group (Bouhuys and Albersnagel, 1992; Bouhuys and Van den Hoofdakker, 1993; Geerts et al., 1995, 1996, 2000; Geerts and Bouhuys, 1998). More-

272

E.H. Bos et al. / Psychiatry Research 113 (2002) 269–277

over, the two sets appeared to be (causally) related to one another (Bouhuys and Van den Hoofdakker, 1991; Geerts et al., 1997). In addition, in other studies these behavioral elements are also considered as indicative of involvement (e.g. Cappella, 1983; Patterson, 1983; Coker and Burgoon, 1987). For these reasons, we chose to retain them in the present study. The behavior scores were normalized over the interviews, to enable pooling of the behavioral elements into two behavioral factors: ‘patient involvement’ and ‘interviewer involvement’. The factor ‘patient involvement’ consisted of duration and frequency of gesticulations, duration and frequency of head movements, duration and frequency of eye contact (all during speech), and duration of eye contact during listening. This factor can be seen as a reflection of the amount of effort one puts into speaking (the factor has also been called ‘speaking effort’). The factor ‘interviewer involvement’ consisted of duration and frequency of verbal backchannel and duration and frequency of yes-nodding (all during listening). This factor is considered to reflect the degree one is stimulating the other in speaking by showing one is attending and understanding (the factor has also been called ‘encouragement’) (for a detailed description of the composition of the factors, see Geerts et al., 1995). 2.3. Statistical analysis We employed partial correlation and linear multiple regression to detect relationships between behavioral predictor variables measured in remission and depressive symptom scores 6 months later. In all equations with higher order terms, we used centered versions of first order variables (Aiken and West, 1991). The upper level of significance to reject H0 was set at 0.05.

initial remission criterion of 8. The HRSD scores of 10 subjects had considerably increased at T2. These 10 subjects could be labeled ‘relapsed’ according to common criteria (HRSD score G12 and 100% increased relative to score at remission). Four subjects had an intermediate HRSD score (between 8 and 12). Thus, the distinguished subgroups are presented here only to describe the sample, and will not be used in further analyses. HRSD remission scores (T1) appeared to be positively related to HRSD scores at T2 (Pearson’s rs0.30, Ps0.035). The amount of change in depression score from T1 to T2 was not related to age, length of antecedent hospitalization or severity of depression at admission (partial correlations, correction for HRSD score at T1). Neither was change in depression score related to subtype of depression (unipolar or bipolar, with or without psychotic features; ANCOVA, correction for HRSD score at T1). Women, however, showed a higher increase in depression score than men (ANCOVA, correction for HRSD score at T1; F(1,48)s4.35, Ps0.042). This gender difference was to be expected given the generally higher risk of depression in women relative to men (Weissman and Klerman, 1977; Paykel, 1991). Baseline HRSD scores (T1) did not differ for the two sexes. Male and female interviewers were evenly distributed over interviews with male and female patients (x2-test). No difference in change in depression score was found between patients with an interviewer of the same sex and patients with an interviewer of the opposite sex (ANCOVA, correction for HRSD score at T1). The HRSD score at T1 was not significantly correlated with either patient involvement scores (Pearson’s rs0.087, n.s.) or interviewer involvement scores (Pearson’s rs0.204, n.s.). 3.2. Involvement behavior

3. Results 3.1. Patient characteristics At T2, 6 months after remission, the mean HRSD score of the sample was 6.9 (S.D.s6.6, range 0.5–30.5). The depression scores of 37 of the 51 remitted patients had remained below the

We applied a linear multiple regression design to determine whether the involvement behavior of the patient measured at T1 contributed to the prediction of severity of depressive symptoms at T2. As we knew that depression score at T1 and patient’s gender accounted for some of the variance in depression score at T2, we controlled for

E.H. Bos et al. / Psychiatry Research 113 (2002) 269–277

273

Fig. 1. Regression of residual HRSD scores at T2 (corrected for HRSD scores at T1 and gender) on patient involvement scores at T1 (see also Table 1).

these factors by forcing them into the equation in the first blocks. The factor patient involvement indeed appeared to be a significant predictor of later depression score. Lower levels of patient involvement were associated with worse prognosis (Fig. 1 and Table 1a). Since in our hypothesis we took account of the possibility that both low and high involvement may be related to a negative outcome, we also tested for a curvilinear effect of patient involvement. This curvilinear model, however, yielded a slightly worse fit than the linear model wF(4,46)s 3.55, Ps0.013x, and the higher order term (patient involvement squared) did not explain any variance at all (DR 2s0.000, DF(1,47)s0.97, n.s.). To check whether the effect of patient involvement was different for the two sexes, we added an interaction term ‘gender = patient involvement’ to the linear regression equation. No significant interaction effect was found, however wDR 2s 0.023, DF(1,47)s1.41, n.s.x. Given the relatively high proportion of subjects

in our sample diagnosed as having a disorder with psychotic features, we were able to investigate whether this subgroup behaved differently from the subgroup without psychotic features. We found out that the ‘psychotic features’ group tended to be lower in patient involvement than the ‘no psychotic features’ group wANOVA; F(1,49)s3.22, Ps0.079x. When we controlled for the presence of psychotic features in the equation of Table 1a, however, the factor patient involvement was still predictive of outcome wDR 2s0.083, DF(1,46)s 5.11, Ps0.029x. The psychotic features factor did not contribute significantly to the equation. In a similar linear regression design (controlling for HRSD score at T1 and gender), we investigated whether the behavior of the interviewer was related to outcome as well. It appeared that the factor interviewer involvement was not predictive of later depression score at all wDR 2s0.001, DF(1,47)s 0.03, n.s.x. The six interviewers, however, appeared to differ in their individual levels of involvement behavior

E.H. Bos et al. / Psychiatry Research 113 (2002) 269–277

274

Table 1 Summary of regression analyses examining the relationship between involvement behavior at T1 and HRSD score at T2, controlling for HRSD score at T1 and gender Model

DR 2

b-value

P-value

F(d.f.)

R2

P-value

s4.84

0.236

0.005

s3.18

0.261

0.015

1a

HRSD score at T1 Gender Patient involvement

0.087 0.076 0.073

0.331 y0.303 y0.272

0.013 0.022 0.040

F(3,

47)

1b

HRSD score at T1 Gender Interviewer involvement Patient involvement Patient involvment=interviewer involvment

0.087 0.076 0.001 0.072 0.025

0.342 y0.309 0.007 y0.276 y0.161

0.012 0.021 0.955 0.039 0.222

F(5,

45)

wANOVA, F(5,45)s3,73, Ps0.007x. Aware of the fact that this could have had repercussions on the levels of their partners’ involvement behavior (as much as patient levels of involvement could have influenced interviewer levels), we checked whether these differences in interviewer style had confounded our results. Entering interviewer dummy variables into the equation of Table 1a revealed that this was not the case. The factor patient involvement was still significantly related to outcome wDR 2s0.087, DF(1,42)s4.95, Ps0.031x. Interviewer variables did not contribute significantly to the equation (DR 2s0.008, DF(5,43)s 0.084, n.s.x. 3.3. Interaction Besides investigating whether the behaviors of patient and interviewer separately had predictive qualities, we examined whether there were any interaction effects between the behaviors of the conversation partners. For that purpose, we regressed the HRSD score at T2 on interviewer involvement, patient involvement and the patient involvement=interviewer involvement interaction (controlling for HRSD score at T1 and gender; see Table 1b). As can be seen from the table, this analysis did not yield any new insights. No significant interaction effects were found. 4. Discussion In this study, we investigated non-verbal behavioral elements indicative of involvement in remit-

ted depressed patients and their conversation partners. We examined whether this non-verbal involvement behavior was predictive of return of depressive symptoms 6 months later. We found that involvement behavior of the remitted patients was related to outcome. The lower the level of involvement displayed, the more unfavorable the prognosis. Involvement behavior of the interviewers was not predictive of later depressive complaints, and neither was the interaction between the conversation partners’ behavior. The finding that low patient levels of involvement behavior were related to return of depressive symptoms can be seen as an indication of the importance of gratifying social interactions and the role the patient plays in these. Fruitful social interactions may need some mutual investment. Social support, considered to be so important in protecting against depression, is something that subjects may receive, but that they also largely have to acquire. The literature on social support predominantly focuses on the support given by the social environment, disregarding the active role the needing party itself can play. Displaying some involvement in social interactions may be indispensable for the establishment and maintenance of social supportive relationships. Subjects displaying too little involvement may thus run an increased risk of depression as a result of a decreased social buffering capacity. This idea is in line with more general perspectives that interpersonal strategies involving approach—as opposed to avoidance— can buffer against the depressogenic effects of

E.H. Bos et al. / Psychiatry Research 113 (2002) 269–277

negative life events by enhancing social resources (cf. Dill and Anderson, 1999; Holahan et al., 1999). It is instructive to compare the present results with those of previous studies of our research group (Bouhuys and Albersnagel, 1992; Bouhuys and Van den Hoofdakker, 1993; Geerts et al., 1995, 1996, 2000; Geerts and Bouhuys, 1998). In these studies, involvement was operationalized in exactly the same way as in the present study, but was measured when subjects were in a depressed state and used to predict subsequent improvement. When patients’ involvement behavior was related to outcome in these studies, it was the subjects displaying the lowest levels of involvement who showed the greatest improvement (Bouhuys and Albersnagel, 1992; Geerts et al., 1995). Low involvement on the part of the interviewers was related to patients’ improvement as well (Bouhuys and Van den Hoofdakker 1993; Geerts et al., 1995). Also, measures of the interaction between patients’ and interviewers’ involvement behavior were related to the outcome; the degree to which patients and interviewers adjusted their levels of involvement to each other predicted subsequent improvement (Geerts et al., 1996, 2000; Geerts and Bouhuys, 1998). The present results clearly deviate from these earlier findings. A favorable outcome was related to higher levels of involvement here, and no interviewer or interaction effects were found. This may indicate that involvement behavior differently relates to depression maintenance and depression relapse, which would be in line with Barnett and Gotlib’s (1988) observation that factors predictive of episode duration are not necessarily also predictive of relapse and recurrence. The differences in results may be understood if one recognizes the importance of the difference in the condition of the patients (depressed vs. remitted). Presumably, the way interaction partners perceive and react to each other’s behavior varies depending on the patient’s condition. Signs of involvement displayed by subjects in the depressed state, for example, may easily be perceived as signs of support seeking and dependency, contributing to these subjects being experienced as demanding or needy (see also Bornstein, 1992). When the depression is

275

over, and the social environment does not place the other in a context of helplessness anymore, these signs of involvement may be perceived as appropriate. Showing higher levels of involvement would be positively received and reacted to in the latter situation, while they would not be in the first. The difference in condition may also account for the fact that in this study no significant interviewer or interaction effects were found. Possibly, in the remitted state, interaction partners behave more independently than in the depressed state. Indeed, there is evidence that the behaviors of patients and interaction partners are more strongly influenced by each other when patients are more depressed (Bouhuys and Van den Hoofdakker, 1991). The behaviors of the interaction partners in this study may thus have been too weakly associated to find an interviewer or interaction effect on the patient’s prognosis. This may also be partly a consequence of the fact that the interviewers in the present study probably were not very significant to the patients compared to people they interacted with in daily life (at least now they had remitted). There is evidence that behavior displayed in patient–stranger interactions is less related to depression outcome than behavior displayed when patients interact with significant others (Hale et al., 1997b; see also Schmaling and Becker, 1991). The latter fact points to a possible disadvantage of the present design, making the generalizability of the results to everyday life limited. Another factor limiting generalizability may have been severity of illness in our sample. Our sample consists of inpatients who suffered from severe depression, in many cases accompanied by psychotic features. Clearly, our group cannot be equated with, for example, an outpatient group. Other disadvantages of this study are that group size may still not have been large enough, and that medication was not controlled for. Notwithstanding these limitations, this study shows that non-verbal involvement behavior is relevant as a predictor of depressive symptoms, even when measured in remitted subjects interacting with relatively insignificant others. This result underpins the importance of behavioral observation in the search for

276

E.H. Bos et al. / Psychiatry Research 113 (2002) 269–277

interpersonal risk factors for depression. The observational method provides a more direct and concrete measure of social functioning than when patients are asked to reflect on this issue and to fill out a questionnaire. Such behavioral assessments may open new possibilities to unravel processes that underlie inadequate social functioning, and in doing so, bear relevance to the prevention of relapse in depression. Acknowledgments This study was supported by NWO (grant 94033-043). The authors are grateful to Jaap Jansen, Bill Hale, Gerda Bloem, Sjoerd Dijkema and Ingrid van der Spoel for their assistance in the ethological registrations, and to Hans Ormel for his comments on an earlier draft of this report. References Aiken, L.S., West, S.G., 1991. Multiple Regression: Testing and Interpreting Interactions. Sage, Newbury Park, CA. American Psychiatric Association, 1994. Diagnostic and Statistical Manual of Mental Disorders (DSM IV). Author, Washington, DC. Barnett, P.A., Gotlib, I.H., 1988. Psychosocial functioning and depression: distinguishing among antecedents, concomitants, and consequences. Psychological Bulletin 104, 97–126. Bebbington, P., 1987. Misery and beyond: the pursuit of disease theories of depression. International Journal of Social Psychiatry 33, 13–20. Bell, R.A., 1987. Social involvement. In: McCroskey, J.C., Daly, J.A. (Eds.), Personality and Interpersonal Communication. Sage, Newbury Park, CA, pp. 195–242. Billings, A.G., Moos, R.H., 1985a. Psychosocial processes of remission in unipolar depression: comparing depressed patients with matched community controls. Journal of Consulting and Clinical Psychology 53, 314–325. Billings, A.G., Moos, R.H., 1985b. Life stressors and social resources affect posttreatment outcomes among depressed patients. Journal of Abnormal Psychology 94, 140–153. Bornstein, R.E., 1992. The dependent personality: developmental, social, and clinical perspectives. Psychological Bulletin 112, 3–23. Bouhuys, A.L., Albersnagel, F.A., 1992. Do interactional capacities based on observed behaviour interfere with improvement in severely depressed patients? Journal of Affective Disorders 25, 107–116. Bouhuys, A.L., Jansen, J.H.C., Van den Hoofdakker, R.H., 1991. Analysis of observed behavior displayed by depressed patients during a clinical interview: relationships between

behavioral factors and clinical concepts of activation. Journal of Affective Disorders 21, 79–88. Bouhuys, A.L., Van den Hoofdakker, R.H., 1991. The interrelatedness of observed behavior of depressed patients and of a psychiatrist: an ethological study on mutual influence. Journal of Affective Disorders 23, 63–74. Bouhuys, A.L., Van den Hoofdakker, R.H., 1993. A longitudinal study of interaction patterns of a psychiatrist and severely depressed patients based on observed behavior: an ethological approach of interpersonal theories of depression. Journal of Affective Disorders 27, 87–99. Burgoon, J.K., 1985. Non-verbal signals. In: Knapp, M.L., Miller, G.R. (Eds.), Handbook of Interpersonal Communication. Sage, Beverly Hills, CA, pp. 344–390. Burgoon, J.K., Newton, D.A., Walther, J.B., Baesler, E., 1989. Non-verbal expectancy violations and conversational involvement. Journal of Non-verbal Behavior 13, 97–119. Cahn, D.D., Frey, L.R., 1992. Listeners’ perceived verbal and non-verbal behaviors associated with communicators’ perceived understanding and misunderstanding. Perceptual and Motor Skills 74, 1064–1095. Cappella, J.N., 1983. Conversational involvement: approaching and avoiding others. In: Wiemann, J.M., Harrison, R.P. (Eds.), Non-verbal Interaction. Erlbaum, Hillsdale, NJ, pp. 37–68. Cegala, D.J., Savage, G.T., Brunner, C.C., Conrad, A.B., 1982. An elaboration of the meaning of interaction involvement: toward the development of a theoretical concept. Communication Monographs 49, 229–248. Cohen, J., 1968. Nominal scale agreement with provision for scaled disagreement of partial credit. Psychological Bulletin 70, 213–220. Coker, D.A., Burgoon, J.K., 1987. The nature of conversational involvement and non-verbal encoding patterns. Human Communication Research 13, 463–494. Coyne, J.C., 1976. Toward an interactional description of depression. Psychiatry 39, 28–40. Coyne, J.C., Burchill, S.A.L., Stiles, W.B., 1990. An interactional perspective on depression. In: Snyder, C.R., Forsyth, D.O. (Eds.), Handbook of Social and Clinical Psychology. Pergamon, New York. Davis, D., Holtgraves, T., 1984. Perceptions of unresponsive others: attributions, attraction, understandability, and memory of their utterances. Journal of Experimental Social Psychology 20, 383–408. Dill, J.C., Anderson, C.A., 1999. Loneliness, shyness, and depression: the etiology and interrelationships of everyday problems in living. In: Joiner, T.E., Coyne, J.C. (Eds.), Interactional Nature of Depression: Advances In Interpersonal Approaches. American Psychological Association, Washington, DC, pp. 93–125. Geerts, E., Bouhuys, A.L., 1998. Multi-level prediction of short-term outcome of depression: the role of non-verbal interpersonal and cognitive processes and of personality traits. Psychiatry Research 79, 59–72. Geerts, E., Bouhuys, A.L., Meesters, Y., Jansen, J.H.C., 1995. Observed behavior of patients with seasonal affective dis-

E.H. Bos et al. / Psychiatry Research 113 (2002) 269–277 order and an interviewer predicts response to light treatment. Psychiatry Research 57, 223–230. Geerts, E., Bouhuys, A.L., Van den Hoofdakker, R.H., 1996. Non-verbal attunement between depressed patients and an interviewer predicts subsequent improvement. Journal of Affective Disorders 40, 15–21. Geerts, E., Bouhuys, A.L., Bloem, G., 1997. Non-verbal support giving induces non-verbal support seeking in depressed patients. Journal of Clinical Psychology 53, 35–39. Geerts, E., Kouwert, E., Bouhuys, A.L., Meesters, Y., Jansen, J.H.C., 2000. Non-verbal interpersonal attunement and extravert personality predict outcome of light treatment in seasonal affective disorder. Journal of Affective Disorders 59, 193–204. Gotlib, I.H., Hammen, C.L., 1992. Psychological Aspects of Depression: Towards a Cognitive-Interpersonal Integration. John Wiley and Sons, New York. Hale, W.H., Jansen, J.H.C., Bouhuys, A.L., 1997a. Depression relapse and ethological measures. Psychiatry Research 70, 57–64. Hale, W.H., Jansen, J.H.C., Bouhuys, A.L., Jenner, J.A., Van den Hoofdakker, R.H., 1997b. Non-verbal behavioral interactions of depressed patients with partners and strangers: the role of behavioral social support and involvement in depression persistence. Journal of Affective Disorders 44, 111–122. Hamilton, M., 1967. Development of a rating scale for primary depressive illness. British Journal of Social and Clinical Psychology 6, 278–296. Harlow, R.E., Cantor, N., 1994. Social pursuit of academics: side effects and spillover of strategic reassurance seeking. Journal of Personality and Social Psychology 66, 386–397. Holahan, C.J., Moos, R.H., Bonin, L.A., 1999. Social context and depression: an integrative stress and coping framework. In: Joiner, T.E., Coyne, J.C. (Eds.), Interactional Nature of Depression: Advances in Interpersonal Approaches. American Psychological Association, Washington, DC, pp. 39–63. Joiner, T.E., 2000. Depression’s vicious scree: self-propagating and erosive processes in depression chronicity. Clinical Psychology: Science and Practice 7, 203–218. Lara, M.E., Klein, D.N., 1999. Psychosocial processes underlying the maintenance and persistence of depression: impli-

277

cations for understanding chronic depression. Clinical Psychology Review 19, 553–570. Lewinsohn, P.M., 1974. Clinical and theoretical aspects of depression. In: Calhoun, K.S., Adams, H.E., Mitchell, K.M. (Eds.), Innovative Treatment Methods in Psychopathology. Wiley, New York, pp. 63–120. Patterson, M.L., 1983. Non-verbal Behavior: A Functional Perspective. Springer-Verlag, New York. Paykel, E.S., 1991. Depression in women. British Journal of Psychiatry 158, 22–29. Paykel, E.S., 1994. Life events, social support and depression. Acta Psychiatrica Scandinavica 89, 50–58. Perris, C., 1987. Towards an integrating theory of depression focusing on the concept of vulnerability. Integrative Psychiatry 5, 27–32. Robins, C.J., Block, P., 1989. Cognitive theories of depression viewed from a diathesis-stress perspective: evaluations of the models of Beck and of Abramson, Seligman, and Teasdale. Cognitive Therapy and Research 13, 297–313. Rounsaville, B.J., Prusoff, B.A., Weissman, M.M., 1980. The course of marital disputes in depressed women: a 48-month follow-up study. Comprehensive Psychiatry 21, 111–118. Schmaling, K., Becker, J., 1991. Empirical studies of the interpersonal relations of adult depressives. In: Becker, J., Kleinman, A. (Eds.), Psychosocial Aspects of Depression. Lawrence Erlbaum, Hillsdale, NJ, pp. 169–185. Segrin, C., Abramson, L.Y., 1994. Negative reactions to depressive behaviors: a communication theories analysis. Journal of Abnormal Psychology 103, 655–668. Segrin, C., 1998. Interpersonal communication problems associated with depression and loneliness. In: Anderson, P.A., Guerrero, L.K. (Eds.), Handbook of Communication and Emotion: Research, Theory, Applications, and Contexts. Academic Press, San Diego, pp. 215–242. Troisi, A.T., Pasini, A., Bersani, G., Grispini, A., Ciani, N., 1989. Ethological predictors of amitriptyline response in depressed outpatients. Journal of Affective Disorders 17, 129–136. Weissman, M.M., Klerman, G.L., 1977. Sex differences and the epidemiology of depression. Archives of General Psychiatry 34, 98–111. Zeiss, A.M., Lewinsohn, P.M., 1988. Enduring deficits after remissions of depression: a test of the scar hypothesis. Behaviour Research and Therapy 26, 151–158.