Psychological Medicine, 2003, 33, 1061-1070. © 2003 Cambridge University Press DOI: 10.1017/8003329170300816X Printed in the United Kingdom Behavioural factors associated with symptom outcomes in a primary care-based depression prevention intervention trial E, LUDMAN;! W. KATON, T. BUSH, C. RUTTER, E. LIN, G. SIMON, M. VON KORFF anp E. WALKER From the Center for Health Studies, Group Health Cooperative and Department of Psychiatry and Behavioral Sciences, University of Washington Medical School, Seattle, WA, USA ABSTRACT Background, A randomized trial of a primary care-based intervention to prevent depression relapse resulted in improved adherence to long-term antidepressant medication and depression outcomes. ‘We evaluated the effects of this intervention on behavioural processes and identified process pre- dictors of improved depressive symptoms. Method. Patients at high risk for depression recurrence or relapse following successful acute phase treatment (N=386) were randomly assigned to receive a low intensity 12-month intervention or continued usual care. The intervention combined education about depression, shared decision- making regarding use of maintenance pharmacotherapy and cognitive-behavioural strategies to promote self-management. Baseline, 3, 6, 9 and 12-month interviews assessed patients’ self-care practices, self-efficacy for managing depression and depressive symptoms. Results. Intervention patients had significantly greater self-efficacy for managing depression (P-<0-01) and were more likely to keep track of depressive symptoms (P<0-0001), monitor early warning signs (P<0-0001), and plan for coping with high risk situations (P<0-0001) at all time points compared to usual care control patients. Self-efficacy for managing depression (P<0-0001), keeping track of depressive symptoms (P=0-05), monitoring for early warning signs (P=0-01), engaging in pleasant activities (P <0-0001) and engaging in social activities (P <0-0001) positively predicted improvements in depression symptom scores. Conclusions. A brief intervention designed to target cognitive-behavioural factors and promote adherence to pharmacotherapy in order to prevent depression relapse was highly successful in changing several behaviours related to controlling depression. Improvements in self-efficacy and several self-management behaviours that were targets of the intervention were significantly related to improvements in depression outcome. INTRODUCTION Given the chronic or recurring course of de- pression for a large proportion of individuals, identifying and treating patients who could benefit from maintenance pharmacotherapy and other preventive strategies has become an » Address for correspondence: Dr Evette Ludman, Center for Health Studies, Group Health Cooperative, 1730 Minor Avenue, Suite 1600, Seale, WA 98101, USA. important priority. Effective and easily dissemi- nable models of care designed to promote main- tenance of treatment gains and prevent future relapse and recurrences of depression are es- pecially needed in primary medical care settings. In primary care approximately 60% of patients with major depression have had two or more prior affective episodes (Katon, 1995; Katon et al. 1995) and over a third are likely to have a recurrence within I year of symptom resolution 10611062 (Lin et al. 1998). However, few primary-care settings are equipped to provide evidence-based psychotherapy, or the educational and ongoing psychosocial support to promote self-care among patients with chronic or recurrent de- pression. As with chronic medical illnesses, there are rarely sufficient resources to plan and sup- port patients in following personalized phar- macotherapy regimens and behavioural plans that maintain treatment progress. Such resources are important components of successful efforts to improve the care for populations with chronic illnesses (Wagner et al. 1996; Katon et al. 1997; Von Korff ef al. 1997). Several core components of cognitive-behav- ioural psychotherapy for depression lend them- selves well to adaptation for brief interventions for relapse prevention (e.g. pleasant events scheduling, self-monitoring of prodromal symp- toms, identification of high-risk situations). These strategies are designed to improve confi- dence and skills in long-term self-management of depression and acknowledge patients’ active ‘responsibility,’ or central role, for continued improvement, In a recent randomized effective- ness trial a low-cost intervention designed to combine education about risk of depression relapse, motivational enhancement and shared decision-making regarding the use of mainten- ance pharmacotherapy and low-intensity cog- nitive-behavioural interventions was tested among a population-based sample of high-risk primary care patients (Ludman ef al. 2000: Katon et al. 2001). Participants who received the intervention had significantly more favour- able attitudes towards antidepressant medi cation (Lin er al. 2003), greater adherence to antidepressant medication and fewer depressive symptoms over the 12-month follow-up period compared with usual care controls (Katon et al. 2001). The intervention was also associated with, modest increases in days free of depression and ‘modest increases in treatment costs (Simon et al. 2002). In this paper we examine behavioural factors that were targets of the intervention in order to understand some of the processes of care that underlie improved outcomes. This may be es- pecially important in effectiveness trials where intervention delivery is less tightly controlled. We assess whether this brief real-world inter- vention improved self-efficacy for managing E, Ludman and others depression and increased specific self-manage- ment behaviours such as self-monitoring, identi- fication of high-risk situations and pleasant events scheduling. We also examine whether dif- ferences in self-efficacy for managing depression and the performance of self-management behav- iours were associated with improved depression outcomes over time. METHOD ants and recruitment Participants were recruited from four large pri- mary care clinics of Group Health Cooperative (GHC), a group-model health maintenance or- ganization serving approximately 400000 mem- bers in Washington State. The four clinics serve a population of 88000 members and are staffed by 73 board certified family physicians. Patients enrolled in GHC are generally representative of the area population in terms of age, ethnicity, education and income. Computerized pharmacy and_ visit regis- tration records were used to identify all patients (18 to 80 years of age) receiving a new anti- depressant prescription from a primary care physician associated with a visit diagnosis of de- pressive disorder. A ‘new’ prescription was de- fined by an interval of at least 120 days since last use of an antidepressant based on prescription date and intended duration of use. Invitation letters were mailed to all potentially eligible patients who were then contacted by telephone 6 to 8 weeks after the initial prescription. Fol- lowing a documented oral consent procedure, participants completed a telephone screening assessment including the current and past de- pression modules of the Structured Clinical Interview for DSM-IV (or SCID) (First er al. 1997), We had a two-stage recruitment process. Patients with significant residual symptoms (.. four or more of the nine DSM-IV depression criteria), current dysthymia (ie. depressive symptoms present for >2 years), or history of recurrent depression (i.e. two or more prior depressive episodes) were invited to participate in an in-person baseline assessment to deter- mine eligibility for the randomized trial or for another trial for which we were recruiting sim- ultaneously. Eligibility criteria for the randomized trial included substantial improvement of the indexBehavioural factors in depression prevention trial depressive episode (defined as either less than four of the nine DSM-IV depression criteria or four DSM-IV depression criteria with an SCL depression score <1-0), and high risk of relapse (defined as a history of three or more lifetime depressive episodes or a history of dys- thymic disorder). Criteria for exclusion were score of >2 on the CAGE alcohol question- naire, plans to dis-enrol from GHC within 12 months, recent use of mood stabilizer or anti- psychotic medication, pregnancy or nursing, and current medication management by a psy- chiatrist. After a full explanation of study pro- cedures, written informed consent was obtained from all participants. Intervention The brief primary care-based relapse prevention program (Ludman er al. 2000) was a multi- faceted intervention that included an edu- cational book (Katon er al. 2002) and videotape about effective management of chronic or re- current depression, two in-person visits with a depression prevention specialist in the primary care clinic, three scheduled telephone monitor- ing contacts (2, 5 and 9 months after enrolment) and four personalized monitoring mailings (3, 6, 10 and 12 months after enrolment) for con- tinued monitoring of depressive symptoms and treatment adherence. Three depression preven- tion specialists (DPS), a psychologist, a psychi- atric nurse and a social worker, were trained to provide the intervention. Each received a (60-page training manual and participated in two half-day training sessions with a psychiatrist (W.K.), a primary care physician (E.H.B.L.), and psychologist (E.J.L.). ‘The goals of the intervention were to increase self-efficacy and behavioural skills for self- management of depression as well as to increase uptake of maintenance medication and improve long-term adherence to antidepressants to re- duce recurrence of depression. The intervention included both educational (Lin er al. 1995) and motivational enhancement components for de- cisions about maintenance medication (Miller & Rollnick, 1991; Rollnick er al. 1992; Botelho & Skinner, 1995), Cognitive behavioural com- ponents of the intervention included strategies deemed appropriate for a stand-alone brief intervention, i.e. a relapse prevention interven- tion that was not a continuation of ongoing 1063 acute phase treatment with a mental health specialist. These components, based on social- cognitive theory of depression and depression treatment (Brown & Lewinsohn, 1984; Lewin- sohn ef al. 1984; Ludgate, 1995), included en- couraging daily stress reducing strategies such as scheduling pleasant and social activities, reg- ular tracking of depressive symptoms using a depression inventory or symptom checklist, and identification of idiosyncratic prodromal symp- toms and use of self-monitoring strategies to identify recurrence early. Patients were also taught to identify and proactively plan for situ- ations that would likely lead to a worsening or recurrence of depression. During the two in-person visits each patient and DPS collab- oratively developed a written self-care plan in- corporating each of these elements. Self-care plans were shared with primary care providers and became part of patients’ medical records. Follow-up telephone calls and personalized mailings helped depression prevention specia ists monitor recurrence of depressive symptoms and treatment adherence and reminded and motivated patients to continue following their self care plan. Details of the intervention programme are described in detail elsewhere (Ludman et al. 2000; Katon et al. 2001). Depression prevention specialists communi- cated regularly with treating primary care phys- ns regarding treatment discontinuation, signs of depressive relapse, or other situations requiring clinical attention (e.g. intolerable medication side-effects). Treating primary care clinicians remained responsible for all pharma- cotherapy decisions. If needed, patients were referred for in-person consultation with an on- site liaison psychiatrist or for ongoing treatment in GHC’s specialty behavioural health clinic. Each DPS met with a supervising psychiatrist (in-person or by telephone) for 15 to 30 min each week to review medical management of cases Patients assigned to the usual care group could receive any services normally available inside or outside of GHC ~ including referral to specialized mental health care. No additional services were provided, but no services usually available were limited or withheld. Measures The in-person baseline interview and blinded telephone interviews at 3, 6, 9 and 12 month1064 included assessment of depression severity, self- efficacy, self-management behaviours, and medi- cation use (adherence). Overall fewer than 3% of the sample reported ‘don’t know’ to any of the items of interest. ‘Don’t know’ responses were coded as missing. Depression severity Depressive symptoms were assessed using a 20-item depression scale extracted from the SCL-90, scored on a 0 to 4 scale (Derogatis er al. 1974). The SCL has been found to have high reliability and validity in multiple studies with medical patients and to be sensitive to change in depressed primary care patients (Derogatis er al. 1974). A score of 1-72 on the SCL-20 has been shown to have the highest positive predictive value for major depression (Mulrow et al. 1995). Self-efficacy Self-efficacy for managing and preventing de- pression was assessed using a 6-item scale (Bush et al. 2001). The scale included two global ques- tions about managing and preventing depression and four behaviour-specific items such as: “How confident are you in your ability to recognize early when you are starting to get depressed". Self-efficacy scale scores could range from a mini- mum of 0 to a maximum of 10. Psychometric analyses indicate that the scale is internally con- sistent (a=0-79) and consists of a single factor explaining 50% of the variance (Bush et al. 2001). Depression self-management behaviours Five separate items were used to measure whether participants had performed depression self-management behaviours in the past month. The five behaviours were: participating in pleasant activities; participating in social activi- ties, keeping track of depressive symptoms; look- ing out for early warning signs of depression; and, anticipating and planning for situations that were likely to cause depression or make depression worse. For three behaviours, (keep- ing track of depressive symptoms, looking out for carly warning signs of depression, and an- ticipating and planning for situations that were likely to cause depression or make depression worse), participants were asked (yes) no) whether they performed cach of these behaviours in E. Ludman and others the past month. These were scored as present or absent. Patients who reported that they en- gaged in pleasant and social activities were asked to rate frequency on a 5-point scale (not at all=0; less than once a week=1; once a week =2; several timesa week =3; and daily =4). Results were analysed as a binary variable (sev- eral times a week or daily versus once a week or less). Antidepressant medication adherence Medication adherence was based on self-report, Participants were asked ‘How many days in the past month did you take an antidepressant?” A participant was considered adherent if they reported taking antidepressants > 25 days of the previous month. If a participant was prescribed fluoxetine they were considered adherent if they reported taking medicine >15 days of the pre- vious month, allowing for every other day dos- ing. Self-report of antidepressant use correlated highly with automated pharmacy refill data (86-7% agreement, x =0°66). This level of con- cordance between self-report and automated pharmacy data is very similar to results found in a prior study (Saunders e¢ al, 1998). Medical co-morbidity Co-morbid medical illness was assessed with the Chronic Disease Score (CDS). This standard measure for assessment of chronic medical mor- bidity uses algorithms based on computerized pharmacy records over a 6-month period (Von Korfi ef al. 1992; Clark et al. 1995) . This CDS is a cost-based measure and has shown good ability to predict mortality, hospitalization and use of general medical services. Statistical methods We tested for intervention and control group differences in baseline demographic and clinical process measures using chi-square tests (for categorical measures) and ¢ tests (for continuous and semi-continuous measures) To estimate the effect of intervention on SCL depression scores via change in process variables, we estimated two sequential models. Models were fit using generalized estimating equations (GEEs) (Diggle er al. 1994). The first set of models estimated the effect of intervention on behavioural process measures. BecauseBehavioural factors in depression prevention trial intervention can only affect process measures after baseline, GEE models for intervention effects included repeated measures taken at fol- low-up time points (3, 6, 9 and 12 months). Models for intervention effects adjusted for the baseline value of the dependent behavioural variable, age, gender and chronic disease status. The second set of models estimated SCL depression scores (at 3, 6, 9 and 12 months) as a function of behavioural process measures. Because behavioural process predictors change over time, we decomposed these predictors into within individual and between individual (cross- sectional) effects (Neuhaus, 2001). Between in- dividual effects are comparisons between people performing and not performing the behaviours; within individual effects are comparisons within individuals who change their behaviour over time. Let xij represent the process variable for the ith inidvidual at the jth timepoint. Between ividual effects are estimated using each in- dividual’s mean level of the predictor over time, ,. Between individual effects estimate expected differences in SCL scores for individuals with different values for the process predictors. Within individual effects are estimated using the difference between each individual’s mean pro- cess predictor and the values observed at each timepoint, ¥)—x,. Within individual effects estimate the expected difference in SCL scores for individuals whose process variable change. Models for the effect of process variables on SCL scores adjusted for the baseline SCL de- pression score, treatment group (intervention or usual care), age, gender and medical co- morbidity, For all GEE models, time was included as a categorical measure to allow non-linear trends. These models also included a treatment group by time interaction to allow differences in time trends for the two groups. We assumed an in- dependence working correlation structure and adjusted standard errors for clustering at the individual level by using robust (or empirical) covariance estimates. We present the overall (average) effect of treatment on process out- comes at follow-up based on a linear combi- nation of estimated covariates. Models for SCL depression scores assume that the effects of process outcomes are constant over time, We did not conduct an a priori power analysis in order to determine the necessary sample size 1065 Table 1. Baseline characteristics of participants: frequencies and means Intervention (= 194) ‘Age, mean (£0.) 464119) 456(133) Female, % 154 19 Caucasian, % 923 880 Married, % 60 550 allege >I year, % 382 870 Employed fll or part-time, % 780 Bi SCL Depression”, mean (6) 083039084035) CSF, mean (sD) osi-4 (1228-0) 1008-2 (9845) Self-efficacy’, mean (5.0) 66(18) 687 ‘Taking antidepressant srs a3 medication, % * SCL is the Hopkin’s Symptom Checklist 20 depression items (range 0 to 4) + CDS, Chronic Disease Score is a measure of chronic medical morbidity using algorithms based on computerized pharmacy re Conds over a six month period. As is common with cost-based measures, the distribution of scores is skewed (non-normally di tributed) {Self-efficacy is the mean score of the Gitem depression seli:management scale developed for this study, scores range from ‘minimum of O10 a maximum of 10 to see a significant change in the behavioural process outcomes defined and measured in this study. Power was based on depression outcomes that were the primary outcomes of the ran- domized controlled trial. RESULTS Details of screening and recruitment are de- scribed in earlier publications (Katon er al. 1999, 2001). A total of 386 patients were en- rolled in the randomized trial (194 assigned to the relapse prevention programme and 192 to continued usual care). Among these, 377 (98%) remained enrolled throughout the follow-up period ; 315 (82%) completed all blinded follow- up assessments. As reported previously, the proportion of patients completing all follow-up assessments was 88% (170 of 194) in the inter- vention group compared to 76% (145 of 192) in the usual care group (z?=943, P=0-002), Those completing all follow-up assessments had significantly lower SCL depression scores at baseline (mean 0-82, s.p.=0-36, v. mean 0-9" 8.D.=0-38; 1=2:15, df =384, P 03). Parti pation in blinded follow-up assessments was not significantly associated with age, sex, or health services costs in the 6 months prior to enrolment.1066 E, Ludman and others Table 2. Intervention effect on behavioural factors Time point Intervention v. Control diflerencesf Variable Baseline 3 month Gmonth month I2month ORS 98% CL Py Selfecacy, mean Inter, 6627297227743 748 B=039 O14-0.64 ” Contol 85679 GST 699 7m Tracking Inte, TB S28 MS 34D 608 408, 2Es9 0 tee sympioms,% Control «573.1300 123015931590 Monitoring Inter, 40218227922 786.7690 327 2ahael tt ‘warning signs, % Control S156 S484 612055286197 Anticipating high Intere. 5103.74.26 DOE 6950, ror 4-272 tet risk-situations,% Conttol S313 S495 S651. GOL Pactcipating in Totery. 7371 739274957703. TL 080153 NS pleasant activities Control 640671406796. GFL at leat several Times a week, % Pactcipating in nterv, 7944562 44575282 S532 lol o7sa37 NS. social activites at Control 44274033. 50-08 997 S52 least several times a week, % Interv, Intervention; f, repression estimate. + GEE models for interveation effets included repeated measures taken at follow-up time points (3. 6, 9 and 12 months). Models for intervention ellects adjusted forthe baseline value of the dependent behavioural process variable, age, gender and! medical co-morbidity {Baseline value is unadjusted § Odds ratios, unless stated otherwise 4 P value indicates significance over al follow-up time points. POOL; *9** P50% of control patients stayed on maintenance phase antidepressant medication over the year-long study period Similarly, the finding that the intervention did not improve pleasant or social activity sched- ig over time may reflect the fact that close to 95% of participants endorsed regular (i.e. at least once per week) participation in pleasant and social activities at baseline. Nevertheless, participation in pleasant and social activities was highly predictive of improvement in depressive symptoms over time. Self-efficacy for managing depression im- proved quickly and stayed high among inter- vention patients, but improved over time in both groups. Not surprisingly, we observed the greatest improvement in self-efficacy among intervention group patients between baseline and the 3-month follow-up interview. Patients received the most concentrated intervention contacts (educational materials, in-person visits and first telephone contact with the depression prevention specialist) during this time. Limitations of this research include the sample of primary care patients studied. Our sample included patients from one region of the country with high rates of employment, and high levels of education, Participants were members of a single, large organized system of health care with family practice physicians for primary care providers, The results reported here may not generalize to more diverse racial and ethnic groups, patients from lower socio-economic status and other types of primary care settings. There is some evidence that depression treat- ment interventions initially developed in homo- genous populations are also effective in more diverse (¢.g. uninsured) populations (¢.g. Smith et al. 2001), nevertheless, whether this pro- gramme would provide similar outcomes with subjects of differing racial ethnic and socio- economic characteristics receiving care in other settings warrants further study. Other limi- tations of this research include our reliance on retrospective self-report, the demand character- istics of the assessment situation and the limited sensitivity of our measures of self-management behaviours that met only a standard of simple face validity. Although these measures were brief, we believe that brief questionnaires are helpful for recruiting more representative (i.e. less motivated) patients and for maintainingBehavioural factors in depression prevention trial relatively high rates of participation in follow-up data collection. Future research would benefit from measures that provide a better distribution of responses. It is encouraging that a minimal intervention with high participation rates and modest care delivery costs can improve self-efficacy for man- aging depression and demonstrate improve- ments in self-management behaviours that are related to improved outcomes over time. Primary care systems need to begin to adapt services to better serve the needs of patients with recurrent and chronic medical and psychiatric illness. Intervention programmes such as the one described here meet many of the criteria believed to be useful in psychotherapy for refractory depression, such as individualizing assessment, setting reasonable goals (as in man- aging a chronic medical illness), implementing lifestyle changes (¢.g. regular aerobic exercise, improved sleep hygiene), enhancing medica adherence, increasing positive reinforcement (developing daily activity schedules), managing cognitive distortions, and using other beha- vioural strategies (Thase, 1997). A brief inter- vention acknowledging patients’ central role for continued improvement can support self- management by improving confidence and skills in long-term self-management of depression. This study was supported by NIMH grant no. MH. 4rn9. REFERENCES Boielho, R.J, & Skinner, H, (1995). Motivating change in health ‘behavior: implications for health promotion and disease preven: tion. Primary Care 22, 565-589, Brown, R. & Lewinsobn, P. (984). A psychoedueational approach 1o the treatment of depression: comparison of group, individual, ‘and minimal contact proceduces. Journal of Consulting and Clinical Psychology 82, 774-183. Bush, T,, Russo, J, Laman, E., Lin, E., Von Kort, M, Simon, G., Katon, W. & Waiker, E. (001). Perceived self-efficacy for de- pression sell management. A reliable and valié self-report measure With predictive validity. Poster presentation at the American Psychological Society Meeting, June 2001, Toronto, Canada Clark, D., VonKorfl, M., Sauaders, K., Baluch, W.M. & Simon, G.B, (1995). A Chronic Disease Score with empirically derived weights. Medical Care 38, 983-795, Derogatis, L.. Rickels, K., Ublenbuth, E. H. & Covi, L. (1974). The Hopkias Symptom Checklist: a measure of primary symptom dimensions. In Psychological Measurements in Pyychopharmacol- ‘ogy: Problems in Psychopharmacology (ed. P.Pichot), pp. 79-110. Rargerman: Basel, Switzerland. Diggle,P. J, Liang, K. Y. & Zoge, SL. (1994). Analysis of Longi- ‘udinal Data. Oxford University Press: New York 1069 First, M., Spitzer, R., Gibbon, M. & Williams, J. (1997), Strue- tured Clinical Inverview for DSM-IV Axis I Disorders (SCID-1), Clinician Version. “American Psychiatric Press: Washington, Be. Katon, W. (1995), Will improving detection of depression in primary fare lead to improved depressive outcomes? General Hospital Payehiaary V7, 1-2 Katon, W., Von Korfl, M. Lin, E, Walker, E., Simon, G., Bush, T Robinson, P, & Russo, J. (1995), Collaborative management to achieve treatment guidelines: impact on depression in pre mary cate, Jownal of the American Medical Association 273, 1026-1031, Karon, W., Vou Korfl, M. Lin, F., Unutzer, J. Simon, G.. Walker, ,, Ludman, F. & Bush, T. (199%). Population-based care of depression: effective disease management strategies to decrease prevalence. General Hospital Psychiatry 19, 169-178. Katon, W., Von Korfl, M., Lin, E. Simon, G., Walker, E., Untitze, Jn Bush, T, Russo, J.& Ludman, E. (1999). Stepped collaborative ‘cre for primary care patients with persistent. symptoms of ‘depression: a randomized tral. Archives of General Psychiairy $6, 1109-1115, Katon, W.5, Rutter, C, Ludman, E,, Von Kori, M. Lin, B. ‘Simon, G. Walker, E, Unutzer, J. & Bush, T, 2001). A random- ized trial of relapse provention in primary care. Archives of General Psychiatry 8, 241-247 Katon, W., Ludman, E., Simon, G., Lin, E., Von Korf, M., Walker, E,, Sullivan, M., Bush, T. & Carter, L” (2002). The Depression Helpbook. Bull Publishing, Company: Bouléer, Colorado Lowinsoha, P., Antonuecio, D., Steinmetz, J. & Teri, L. (1984). The Coping with Depression Course: A Psychoeducational Intervention for Unipolar Depression. Castalia Press: Eugene, OR. Lin, E.H.B,, Von Korff, M., Katon, W., Bush, T., Simon, G. E., Walker, F. & Robinson, P. (1995). The role ofthe primary care physician in patients adherence to antidepressant therapy. Med fal Care 33, 67-74 Lin, E.H, B., Katon, W., Von Korfl, M., Russo, ., Simon, G.E., jas, T., Rutter, C, M., Walker, E.A. & Ludman, E. (1998). Re lapse of depression in primary care: rate and clinical predictors. Archies of Family Medicine 7, 443-49. Lin, E,11.B, Katon, W.J., Ludman, E.5., Simon, G.E., Bush, T.M., Rutter, C. & Vou Korff, M. (2003). Enhancing adherence to long-term antidepressant treatment for high risk primary eare patients. General Hospital Peychiairy (in the pees). Ladgate, J.W. (995). Maximizing Psychotherapeutic Gains and ‘Preventing Relapse in Emotionally Distressed Clients. Professional ‘Resource Press: Sarasota, FL. Ludman, E., Von Korff, M., Katon, W., Lio, E, Simon, G., Walker, Bush, T. & Wahab, S. (2000). The design, implementation and acceptance of a primary care-based intervention to prevent depression relapse International Journal of Psychianry in Medicine $30, 229-245, Miller, W.R. & Rollnick, S. (1991). Mosivational Interviewing Preparing People to Change Addictive Behavior. Guilford Press: New York. Malrow, C., Williams, J. J, Gerety, M., Ramirez, G., Montil, ©. & Kerber, C. (1095). Casefinding instruments for de- pression in primary care settings. Arwals of fnternal Medicine 122, 513-921 ‘Neuhaus, J.M. (2001), Assessing change with longitudinal fand clustered binary data. Annual Review of Public Health 22, 115-128, Rollnic, S., Heather, N. & Bell, A. (1992), Negotiating behaviour ‘change in medical setings: the development of brief motivational intersiewing. Journal of Mental Health 1, 25-37 Saunders, K., Simon, G., Bush, T. & Grothaus, L (1998). Assessing ‘the feasibility of ‘using computerized pharmacy refill data to monitor antidepressant treatment on & population basis: a com- parison of automated and selfreport data. Joumal of Clinieal Epidemiology $1, 883-890, Simon, G-E,, Von Korfl, M, Ludman, E.., Katon, W., Rutter, C, ‘Unuteer, I, Lia, EH.B, Bush, T. & Walker, E, (2002).1070 Cost-ffectiveness of a program to prevent depression relapse in primary care, Medical Care 40, 941-950. ‘Smith, J.L., Rost, K.M,, Nutting, P-A. & Elliot, C.B. (2001), Resolving disparities in antdeprssant treatment and quality of life outcomes between uninsured and insured primary care patients with depression. Medical Care 29, 910-922. ‘Thass, M.E, (1997). Psychotherapy of refractory depressions Depression and Anxiety, 190-201 E, Ludman and others Von Korfl, M., Wagner, E & Saunders, K. (1992) A chronic disease Score from’ automated pharmacy data. Jownal of Clinical Epidemiology 8,197 203, Von Kori, M., Gruman, 3, Schaefer, J, Curry, S.J. & Wagner, E, H. (1997) Collaborative management of chronic illness. Annals of Internal Medicine 127, 1097-1102. ‘Wagner. E.H., Austin, B, T.& Yon Kor, M.(1996). Organizing care Tor patients with chronic illness, Milbank Quarterly 74, S11-544