Government & Politics

Befriending as an intervention for chronic depression among women in an inner city. 1: Randomised controlled trial

Befriending as an intervention for chronic depression among women in an inner city. 1: Randomised controlled trial
of 7
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
Related Documents
   10.1192/bjp.174.3.219Access the most recent version at doi:  1999 174: 219-224 The British Journal of Psychiatry   T Harris, GW Brown and R Robinson women in an inner city. 1: Randomised controlled trialBefriending as an intervention for chronic depression among   References Article cited in: permissionsReprints/ to To obtain reprints or permission to reproduce material from this paper, please write  to this article atYou can respond   serviceEmail alerting   click here top right corner of the article or Receive free email alerts when new articles cite this article - sign up in the box at the  fromDownloaded  The Royal College of PsychiatristsPublished by on July 13, 2011  go to: The British Journal of P sychiatry  To subscribe to  BRITISH JOURNAL OF PSYCHIATRY (1999). 114. 119-214 Befriending as an intervention for chronic depression among women in an inner city Randomised controlled trialt TlRRlL HARRIS, GEORGE W. BROWN and RUTH ROBINSON Background Earlier workon the protective role of social support in onset and course of depressive disorder suggested that its provision might be relevant or outcome. Aims To evaluate volunteer befriending as an intervention among women with chronic depression n inner London. Method A randomised controlled trial, with a waiting list control design, with outcome measured as Present State Examination (PSE)-based full or partial remission after one year. Results A statistically significant effect upon remission was found for befriending. Other treatments monitored naturalistically did not relate o remission, nor did initial duration ofchronic episode or comorbidity, but there was an association with initial PSE score.The findings are discussed n relation to standard drug-trial results and to another befriending ntervention with the elderly. Conclusions The results are not unpromising. Additional trials with less restricted intake conditions, and in more naturalistic general practice settings, might conf rm volunteer befriending as a useful adjunct to current treatments. Declaration of interest Research supported by the Medical Research Council. Befriending service funded by contributions from the Joseph Rowntree Foundation, he King s Fund, the Mental Health Foundation, NorthThames Regional Health Authority and the Baring Foundation. The decision to pursue an intervention trial of volunteer befriending focusing on chronic depression followed directly from research that had highlighted the likely pro- tective role of social support with regard to depressive onset (Brown et al, 1986). How- ever, evidence that absence of social sup- port contributed to the perpetuation of an established disorder has received less atten- tion, although a closely allied factor - he presence of a marked interpersonal diffi- culty with a core tie at the point of on- set - was found to predict chronicity (an episode at 'caseness' level for at least one year: Brown Moran, 1994). Moreover, some social support interventions had proved encouraging (e.g. bereavement groups, see Parkes, 1980; Homestart, see van der Eyken, 1982; Newpin, see Cox, 1993), although rarely escaping severe methodological criticism. The issues con- cerning support are so complex that fully convincing evidence is only likely to emerge in an experimental setting with a random- ised controlled trial (RCT) design. PURPOSE One reason for choosing to evaluate volun- teer befriending rather than some other form of supportive intervention was that it was feasible for us to set the service up ourselves and thus maintain some control over the allocation procedure of the experi- mental design, which is always difficult, if not impossible, when evaluating an on- going service. For obvious reasons, random allocation can be difficult to contemplate for those in a well-established service and we had little faith that such a procedure could be carried out effectively without setting it up ourselves. Another reason was potential cost-effectiveness. If volunteer befriending proved beneficial, it might turn out to be a cheaper way to deal with unmet 'See Paper 2, pp. 22S232 this issue need than the expansion of existing ser- vices. There were also aspects of befriend- ing that we considered might lead a depressed woman to find it in some ways more supportive than professional help. The knowledge that someone cares enough to visit without being paid and will be flex- ible about timing these visits, and the sense of equality afforded by the mutual self-dis- closure of friendship might, for example, contribute to her sense of self-worth and security (Cox, 1993). AIMS AND HYPOTHESES An RCT with an intention-to-treat design sought to test the hypothesis that those allo- cated volunteer befriending would more often show improvement or recovery than controls. Baseline severity of depression, comorbidity, duration of chronicity and other treatment experience would also be taken into account. Further aims, reported in the following companion paper, included exploration of the role of other psychoso- cia1 factors in remission, along with the characteristics of the volunteers and the quality of their relationships with the depressed women. DESIGN The intervention project A waiting-list control design was adopted with the aim of comparing a minimum of 40 chronically depressed befriended women with 40 similar controls. In terms of preva- lence, about half of all depressive episodes occurring during a one-year period to women in an urban setting in the UK have lasted at least 12 months (Brown Harris, 1978). This is despite the fact that some three-quarters of all onsets remit within one year of onset. Selecting such women for study had the twin advantage of dealing with a significant public health problem and avoiding some of the complications raised by a control group with a high rate of spontaneous remission. A word needs to be said about designs that avoid offering treatments of proven ef- ficacy, such as antidepressant drugs. Be- cause our previous work had highlighted the large number of depressed women not in contact with their general practitioner (Ginsberg Brown, 1982), selection from a wider population than surgery attenders enabled us to reach out to those who would be unlikely to be offered other treatments.  HARRIS ET AL Our research and that of others (e.g. Brugha et al, 1992) had also found that many are unwilling to accept medication, or at least to take it consistently. Our inter- est in befriending stemmed precisely from the possibility that it might come to be seen as an alternative by such women. Any de- pressed women who had recently started regular treatment by psychiatrists or psychotherapists were excluded, because it would be impossible to distinguish its effects from those of the experimental inter- vention. However, other 'treatments' - for example, if already taking antidepressants or having contact with a health care profes- sional for over one year, or any 'alternative' treatments started after first interview - were not grounds for exclusion but were taken into account at the analysis stage. Other samples used for comparative purposes Recent criticism of the generalisability of trial results on the grounds that those not included in the trial may differ crucially from those included (Licht et al, 1997) sug- gests that it is important to view such trials in a comparative context. This is possible in the present instance because the issue of chronicity has been studied in two other groups of depressed women living in the same area, although clearly they do not have the same status as a randomly allo- cated control group. One representative sample of 395 Islington mothers (Brown et al, 1986) provides 35 women with de- pression that had lasted at least one year who were followed up after 12 months in the early 1980s and whose rate of remission therefore provides an independent assess- ment of what to expect in our control series. The other, a consecutive group of female psychiatric patients with depression from three north London hospitals, inter- viewed along the same lines in 1985- 1988, initially provided a group of 65 with depression lasting over 12 months (Brown et al, 1994). However, this report only deals with the 18 who, when first inter- viewed, had already been depressed for at least 12 months and whose outcome had therefore been monitored prospectively. PROCEDURE: RECRUITMENT AND EXCLUSION CRITERIA Befriending service side Funds were obtained for Choice, the Isling- ton branch of the Family Welfare Associa- tion, to extend a befriending service provided for ex-mental hospital patients. Female volunteers were recruited through advertisements in the local press, churches and health centres. All those responding were given several interviews before being accepted for training, which required three full Saturdays. Selection criteria for the volunteers were not specified in advance and in practice were developed by Choice. Current depressive symptoms in a potential volunteer were not an exclusion criterion. Befriending was defined as meeting and talking with the depressed woman for a minimum of one hour each week, and act- ing as a 'friend' to her, listening and 'being there' for her. Although the emphasis was on facilitating confiding, the three-day training also encouraged volunteers to ac- company their befriendees outside on trips, . to broaden their range of activities, to offer practical support with ongoing difficulties and to create the 'fresh-start' experiences often found to precede remission. (These points are discussed in the following com- panion paper.) Criteria for matching volun- teers with depressed women again evolved as the social workers proceeded: although similarity of background experience was considered, there was no attempt to restrict matches in terms of ethnic or cultural back- ground. Research side Research staff were responsible for recruit- ing the sample of chronically depressed women. Identification was by postal screen- ing of all women registered with (rather than attending) certain general practices in Islington. A version of the General Health Questionnaire (GHQ-30; Goldberg et al, 1970) was used, amended to distinguish chronic from shorter episodes. Next, tele- phone screening, using an abbreviated form of the shortened Present State Examination (PSE-10; Wing et al, 1990) - he clinical interview of the main research - eliminated a large number of probable 'false positives': especially common were women with min- or depression or major depression of less than one year. Among the exclusion criteria were severe comorbid conditions of sub stance misuse or eating disorders, but not comorbid anxiety without these other con- ditions. Also, during the initial telephone interview, anyone expressing lack of inter- est in the provisional offer of befriending was excluded. At third contact at home, a full clinical interview made a final psychi- atric assessment of whether the woman was chronically depressed at a caseness level and confirmed her interest in befriend- ing, although it was specified that this might not be available immediately. Once such a woman was identified, research staff immediately allocated her randomly to be- friending or control group through a sealed envelope system. There were equal numbers of control and intervention envelopes, so it was a constrained form of randomisation. The interviewer asked another member of the team to select an envelope for her interviewee in the presence of a third work- er. Names of those selected for befriending were immediately passed to Choice. Volunteers received the same initial home interview as the depressed women. Each was allocated one befriendee, with the proviso that if a particular relationship broke down then another match should be attempted. After one year volunteers, all control group members and befriendees were re- interviewed. The same set of measures was used to assess any changes in mental state and other treatments. At follow-up, non- recovered controls were offered befriending but this group is not considered in this re- port because the logic of an RCT demands random allocation. MEASURES USED IN THE EVALUATION Clinical state A shortened version of PSE-10 was used, extended by the Bedford College criteria to date onset and offset of episodes of de- pression and anxiety, as well as to assess the severity of symptoms (Finlay-Jones et al, 1980). The telephone screening used an even shorter form, focusing only on core symptoms of depression. The Bedford Col- lege threshold for 'caseness' has been found similar to probable major depression ac- cording to Research Diagnostic Criteria (RDC) (Dean et al, 1983). It is defined by PSE depressed mood and at least 4/10 core symptoms; 'borderline caseness' of depres- sion is defined by the presence of depressed mood and 1-3 core symptoms (see Finlay- Jones et a1, 1980). Similar criteria were ap- plied to conditions that are often comorbid with depression, such as anxiety, ea~g is- order, drug misuse and alcohol misuse. Baseline clinical predictors thus included duration of depressive episode, presence of comorbid conditions and total PSE score.  BEFRIENDING AS AN INTERVENTION FOR CHRONIC DEPRESSION. I Previous work (Brown et al, 1988) had distinguished full 'recovery' (moving from caseness to absence of depressive symp- toms) from 'improvement' (moving from caseness to medium borderline caseness or to RDC minor depression). The aetiological processes appeared very similar for both, and a joint index of 'either-recovery-or- improvement' was employed, but here the terms full and partial remission are used (after Frank et al, 1991). In practice, those with remission in this enquiry therefore contain a small proportion who retain resi- dual depressive symptoms (for a commen- tary on this issue, see Paykel et al, 1995). A minimum duration of two months was taken for remission. Three women (one control and two in the intervention group) found at follow-up to have been in remis- sion for less than two months were subse- quently re-contacted to establish whether the minimum duration criterion had been reached. Other treatment during follow-up period In a community-based intervention, alter- native treatment-seeking behaviour could not ethically be discouraged as part of the controlled design. However, details were collected about medication taken and dura- tion of attendance with any other treatment personnel. RESULTS Management of the intervention Of 4182 women screened by postal questionnaire, 606 were considered to be possibly suffering from chronic 'caseness' depression. Subsequent telephone screening established that 248 of these appeared to have only a minor or non-chronic form of depression, that seven had an additional psychiatric disorder (such as alcohol or drug misuse) of such prominence that they were unsuitable, and 36 were likely soon to move away from the area. This left 315 possibly suitable women. When asked about their interest in befriending, 204 were excluded: 161 indicated lack of inter- est, 30 were unsure and 13 were regularly seeing another therapist. Of the remaining 11 1 seven lost contact with the project be- fore they could be interviewed, leaving 104 who had accepted in principle the idea of participating in a befriending scheme. At the home visit a clinical interview estab- lished that all women met the clinical criteria for inclusion. This surprising lack of 'false positives' was probably a function of the high symptom threshold set before embarking on a home interview. The first 18 months saw 43 women allocated to befriending and 43 to the control group. The logic of the RCT design requires that only these women are reported on here (see Fig. 1). The advertisements for volunteers attracted 202 respondents, 78 of whom showed continued interest and so were seen by the social workers. Of these, 16 with- drew, six were not considered suitable and 56 underwent the training. In all, 48 matches were made, with three volunteers matched twice. Again, for reasons of the RCT design, this report focuses only on the first 43 matches. Nature of sample There were no important demographic dif- ferences between the two experimental groups. The ethnic composition was repre- sentative of the locality. The depressed women were drawn from a wide age range but the heaviest concentration was among those aged 2540 years and nearly two- thirds were in a partnership. Despite this, more than half had no children at home, and only about one-fifth were single mothers. About one-third worked full-time but about two-fifths did not work at all. Factors predicting remission Simple comparison of experimental and control groups (Table I Overall 'remission' as defined (including partial remission to 'borderline case' level; see Measures section) occurred in 65 (28143) of the befriended group and 39 (17143) of the controls (x2=4.66, d.f.=l, P<O.OS). (Partial remission from case or major depression to borderline or minor de- pression had occurred in four of the group TpMe I Percentage of chronically depressed mwn with remiuion within 13monthsof interview in the different sample groups Sample group Remission Intervention project Allocated befriending 65 28143) Befriending project controls 39 (17143) Two comparison series Islington population 40 (14135) North London patients 50 (9118) Sample selection procedure Stage n = 4182 Postal screening I Replied with apparent chronic depression n = 606 Stage 2 rue' dhronic Moving Other Depression Telephone n = 315 n = 36 disorder minor or screening for prominent non-chronic symptoms n=7 n=248 Stage 3 Showed Refused Unsure Seeing Asking about befriending y n = 30 therapist ther = 13 I Stage 4 Face-to-face interview control series not reported here n=ia Stage 5 Allocated befriending Control group Random allocation n = 43 n = 43 Fig. I Befriending s an intervention for chronk dinid depression.  HARRIS El AL allocated to befriending and in two con- trols.) All reached the minimum duration criterion of two months' remission. A simple proportional reduction in error (PRE) conveys the relative size of an effect: [Experimental proportion remitting (0.65) -Control proportion remitting (0.39)] =0.43 [I OO -Control proportion remitting (0.39)] This takes the proportion not expected to improve (in terms of the control series fig- ures) as a base from which to compare the improvement in the experimental remission rate over that of the control group. Put an- other way, the intervention appears to have increased the remission rate by 43 among those not remitting spontaneously. Comparison by taking account of completers: whether befriending received or merely ollocoted Table 2) Remission occurred in 56 (10118) of the women who met their volunteer not at all or only once but then refused further contact. Although remission tended to be higher among those receiving the full 12 months' befriending as opposed to 2-6 months 76 (13117) and 63 (518), re- spectively - his did not reach statistical significance. (There were no instances of befriending lasting between 7 and 11 months.) Thus, 72 (18125) of completers experienced remission. Generalisability: remission rates in two local noturalisticolly studied samples Table 1 gives remission rates in the two other samples mentioned. These figures of 40 (14135) in the community sample and 50 (9118) among the patients did Table 2 Percentage of chroniully depressed women with remission within 13 months of interview for various degrees of befriending Degree of befriending Remission Received for I2 months 76 (13/17) Received for 2-6 months 63 (518) Saw social worker but not 70 (7110) volunteer Refused after seeing volunteer 38 (318) once Table 3 Percentage of chroniulty depressed women with remission according to total PSE sc m at first interview and befriending allocation Total PSE score Allocated befriending Controls Total not differ substantially from the rate found in the befriending control group. Clinical features predicting remission Length of episode prior to intake into the project was not associated with outcome, nor was comorbidity with anxiety (or the minor misuse of alcohoVsoft drugs or mild eating disorders not excluded by the selec- tion criteria), either at first interview or at any point during the episode. However, higher total PSE score at first interview did predict non-remission, although sub- scores of the core depressive symptoms did not do so. This effect, however, was confined to a small group of eight scor- ing more than 36 (see Table 3). Other treatment during follow-up period Alternative treatment was associated neither with remission nor with inter- vention status: 33 of those with remission and 61 of those without had consulted at least one professional because of their depression, and figures were 45 (19142, one not known) in the control group com- pared with 40 (17142, one not known) in those allocated befriending. A similar picture emerged for use of psychotropic drugs: proportions taking these were 9 for those with and 17 for those without remission, with 12 for controls and 14 for those allocated befriending. A composite measure taking account of both attendance at other services and antidepres- sant medication showed no association with remission in either group. If anything, those without other treatments were doing better. This was not explained by the sever- ity of symptoms: although scoring positive on this composite treatment measure tended to go with the very high PSE cores of 37 or more at first interview, numbers were small (13 (4131) v. 7 (4154)). Moreover, numbers scoring 26 or more were identical among those positive and negative on this composite tmatment index. There is, therefore, no need to control for other treatments in multivariate analyses. DISCUSSION Main findings In considering the aims and hypotheses out- lined, first the data did show some effect of the intervention upon remission over one year: 65 of those allocated befriending and 72 of those completing (i.e. in regular contact with a befriender for at least two months) experienced remission compared with 39 in the control series. There was some hint that those who went as far as to meet a volunteer just once and then dropped out did no better than con- trols (only 38 remitted), but numbers were small and this difference fell short of statistical significance. Those allocated be- friending who, in the end, chose not to take it up with the social workers also fared no worse than those who did, and the follow- ing companion paper details psychosocial factors that may account for this. Comparison with other trials A recent RCT with a chronically depressed elderly group in a population enquiry in Kentish Town has also suggested an effect of regular visits from a community nurse (Blanchard et al, 1995), and organisations such as Newpin and Homestart have con- tinued to expand their activities. The differ- ence in 65 v. 39 remimng here is reasonably close to the 55 v. 20 differ- ence in the elderly series, the lower rate of 20 in their control series reflecting their shorter study period and consequent re- duced opportunity for spontaneous remis- sion. Using the PRE statistic described earlier to estimate effect size, the interven- tions appear to have brought about 44
Similar documents
View more...
Related Search
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks

We need your sign to support Project to invent "SMART AND CONTROLLABLE REFLECTIVE BALLOONS" to cover the Sun and Save Our Earth.

More details...

Sign Now!

We are very appreciated for your Prompt Action!