Introduction
Depression is a major public health problem, and is one of the most common conditions presenting to primary care. In the United Kingdom, the point prevalence of depression is 2.8%, and for mixed anxiety and depression it is 9.2% (National Statistics, 2000). Depression is estimated to affect 17% of adults at some time in their lives (Scottish Executive Social Research, 2002).
In Scotland, the rate of antidepressant prescribing has increased rapidly over recent years, from 28.9 million defined daily doses in 1992/1993 to 153.2 million in 2008/2009 (Information Services Division Scottish Government, 2010), and similar increases have been reported in Europe (Middleton et al., Reference Middleton, Gunnell, Whitley, Dorling and Frankel2001; Van Marwijk et al., Reference Van Marwijk, Bijl, Herman and de Haan2001; Helgason et al., Reference Helgason, Tomasson and Zoega2004; Guaiana et al., Reference Guaiana, Andretta, Corbari, Mirandola, Sorio, D’Avanzo and Barbui2005; Reseland et al., Reference Reseland, Bray and Gunnell2006) and North America (Hemels et al., Reference Hemels, Koren and Einarson2002; Olfson et al., Reference Olfson, Marcus, Druss, Elinson, Tanielian and Pincus2002). The cost of depression and anxiety in the United Kingdom has been estimated at £17 billion, or 1.5% of GDP (Layard, Reference Layard2006).
The ‘correct’ level of antidepressant use at a population level is unknown. It has therefore not been possible to ascertain whether this increase in antidepressant drug use represents an appropriate response to a previously under-recognised and under-treated disorder, or indicates instead that low mood has become unhelpfully ‘medicalised’ for some people.
Nevertheless, NICE guidance has clarified the appropriate use of antidepressant medicines, psychological therapies and guided self-help in the management of depression (National Institute for Clinical Excellence, 2004).
A perceived shortfall in the provision of services for people with depression in Scotland led to the establishment of the ‘Doing Well by People with Depression’ programme in 2004. This initiative was funded by the Centre for Change and Innovation in the Scottish Executive Health Department. The ‘Doing Well’ programme described here was one of the seven pilot programmes funded by the Centre, and received a total of £480 000 over two years. Central funding for ‘Doing Well’ ceased in October 2006, and the service has since been maintained using mainstream funding from NHS Greater Glasgow and Clyde. The protocol for the intervention described here was based on the available evidence about collaborative care, much of which was subsequently incorporated into NICE guidance on depression (2004). The ‘Doing Well by People with Depression’ programme therefore had similar objectives to the improving access to psychological therapies programme in England (which began in 2006), though the two approaches developed independently.
The objectives of the ‘Doing Well’ programme described here were to reform existing services to:
• provide cost-effective, evidence-based care for people with new cases of low mood, adjustment disorder and depression at all levels of severity;
• provide some form of psychological therapy (including guided self-help) for all patients referred to the programme;
• encourage rational antidepressant prescribing;
• provide adequate capacity to cope with the high demand for depression care; and
• respect the choices of patients, the autonomy of clinicians and acknowledge the clinical uncertainties that can arise in depression care.
This paper describes:
• referral rates and the capacity of the programme to cope with demand for the service;
• the profile of antidepressants used for patients in the programme; and
• clinical outcomes based on the Personal Health Questionnaire.
Methods
‘Doing Well’ involved a small clinical team of mental health clinicians providing care for people with depression in their local general practitioners’ (GPs) practices. The programme was implemented in 14 general practices in Renfrewshire, a mixed urban-rural area with pockets of significant deprivation, located about 10 miles west of Glasgow.
A pilot was implemented in one general practice between July and October 2004 and then other practices were gradually recruited into the programme. By February 2006, a total catchment population of 79 600 people registered to these 14 practices was reached. The programme continues to operate in these practices at the time of writing (February 2010) and has been extended into neighbouring practices.
Clinical care
‘Doing Well’ accepted referrals of new presentations of low mood, depression or adjustment disorder for people aged between 18 and 64 inclusive. ‘New’ presentations were broadly defined as people who had not presented with affective symptoms in the previous six months, or who had begun treatment for a new episode in the previous two months. Patients were not accepted if they had a primary diagnosis of alcohol dependence, psychosis, bipolar affective disorder, dementia or terminal illness.
The clinical team comprised one whole-time equivalent self-help support worker, 4.5 whole-time equivalent primary care liaison workers and 0.2 whole-time equivalent consultant psychiatrists. The staff complement is summarised in Table 1.
Table 1 Staff complement
WTE = whole time equivalents.
Self-help support workers are psychology graduates who have received local training in ‘practical psychological skills’. This is a skill-based course lasting 30 h which trains practitioners to use the self-help book Overcoming Depression: A Five Areas Model (Williams, Reference Williams2001). Their work was clinically supervised both by senior ‘Doing Well’ clinicians and by the local psychology department. Total individual clinical supervision amounted to 2 h/week.
Primary care liaison workers are experienced psychiatric nurses or occupational therapists with expertise in mental health. They also had training in the use of guided self-help techniques, but had additional training in cognitive behavioural therapy, interpersonal therapy and the pharmacological treatment of depression. A consultant psychiatrist was part of the team, acting to review more complex clinical cases and to supervise other clinicians. This amounted to 8 h of clinical input per week. Weekly clinical supervision was conducted in group format, involving all ‘Doing Well’ clinicians.
Each non-medical ‘Doing Well’ clinician worked to a timetabled 10-session weekly ‘template’ that comprised of
• one to two sessions of administration
• one session of group supervision
• five to six sessions of direct patient contact
• two sessions of professional development
‘Professional development’ sessions were allocated to a range of tasks, depending on a shared assessment of staff members’ training needs and preferences. Examples of professional development undertaken by different staff members included completing a two-year diploma course in cognitive-behavioural therapy, completing a course in interpersonal therapy, writing or presenting research papers, developing group work teaching materials and training in website development.
In addition, self-help support workers were allocated to spend two sessions per week in the local psychology department, where they took on a small caseload and received 1 h of weekly supervision. Quarterly meetings for all ‘Doing Well’ staff were held away from base in order to review performance data and review or formulate protocols and standards for clinical service.
All referrals were made electronically by GPs, and included an assessment of depression severity using the personal health questionnaire (PHQ) at the time of referral. The PHQ is a self-administered nine-item scale capable of producing a depression rating consistent with the Diagnostic and Statistical Manual of Mental Disorder 4th edition (DSM IV; Spitzer et al., Reference Spitzer, Kroenke and Williams1999). An electronic referral system linking primary and secondary care databases transferred this assessment PHQ score, relevant physical health care records and any comments from the GP directly to the ‘Doing Well’ team. Clinical and support staff in general practice received training in the use of the electronic referral system and assessment tools.
Patients referred to the service were asked to confirm in writing or by phone that they wished to attend, and only those who responded were offered an initial appointment. Patients were seen in their local health centre, and first assessments were scheduled to be of 50 min duration. All patients were introduced to self-help materials based on the ‘overcoming depression’ text (Williams, Reference Williams2001). ‘Guided self-help’ of this kind was conducted in accordance with the practical psychological skills training described above. Lifestyle advice with respect to exercise and alcohol use was given to all patients. An assessment of suicidal ideation was included at each review point. All patients would continue with the planned self-help intervention, as long as they showed an adequate clinical response.
Antidepressant treatment was recommended for patients scoring 15 or more on the PHQ, or those who had a PHQ greater than 10 and had not shown improvement in mood with previous treatment during this episode of care. Antidepressant treatment recommendations were made in keeping with a 3-drug local formulary (fluoxetine, citalopram or lofepramine). Particular attention was paid to patient information and education.
The ‘steps’ in care are represented diagrammatically in Figure 1.
Figure 1 Schematic representation of ‘steps’ in care related to response to treatment
Patients were typically reviewed fortnightly, and at each visit prescribing decisions were reviewed in the light of the patient’s PHQ score. If non-response had persisted beyond about eight weeks, consideration was given to the introduction or switching of medication, a ‘step up’ to more intensive psychological treatment or a referral to secondary care.
Weekly team meetings were held to review clinical cases and workload. ‘Doing Well’ information systems allowed for the ready identification of patients who were not responding to treatment or needed a change in their care. The workload for individual clinicians could also be monitored and amended as necessary.
All patients included in this analysis gave written consent for their anonymised data to be used. Ethical approval for the study was granted by the Local Research Ethics Committee (Ref AC04/073). PHQ scores were tracked continuously throughout treatment. Statistical analyses were carried out using SPSS for Windows version 15.
Results
The outcomes for all patients referred as part of the pilot (July 04–October 04) and the first 24 months of the full programme implementation (October 04–October 06) are described in Figure 2. Patients were grouped into one of five categories for analysis (Table 2).
Figure 2 Flow chart of subjects referred to programme during July 04–October 06
Table 2 ‘Discharge category’ definitions used in analysis
A total of 1584 patients were referred during the study period. The flow of patients through the study is represented in Figure 2. Eighty-one patients referred (5%) did not meet inclusion criteria and are not included in further analysis. This group included 28 people aged over 65 years. Two further patients were excluded from analysis at their request. Of the 1501 patients eligible for inclusion in the study, 294 (19%) did not attend any appointment offered. Thirty-eight people (3% of those eligible) declined the offer of an appointment. Therefore, 332 (22%) people were appropriately referred but dropped out of the programme before assessment.
A further 320 (21%) people ‘disengaged’ from treatment after being seen at least once. This group includes one person who committed suicide and 11 who moved away as well as those who chose not to maintain contact with the service. Those who disengaged from the programme included patients who did not respond to treatment as well as people who had been showing signs of clinical improvement.
Of the 1501 patients included in the study, 654 (44%) completed treatment. One hundred and ninety-five (13%) patients were referred to other services. Of those referred elsewhere, 68 people (35%) were referred on for longer-term psychological therapies (including counselling), 44 (22%) received support outside NHS mental health services (eg, social work, employment support and physical health care) and 15 (8%) required multidisciplinary care from the local community mental health team (Figure 3). After assessment by the ‘Doing Well’ team, 32 patients (16%) were recognised to have mental or physical health problems other than depression, and were referred back to primary care for appropriate management.
Figure 3 Services used for the 195 patients (13% of eligible subjects) transferred requiring non-‘Doing Well’ care
After full recruitment of GP practices, monthly referrals remained broadly stable, averaging 1.3 referrals/month per 1000 population.
The mean age of the 1501 subjects was 38 years; 68% were female. PHQ scores were recorded for 1499 of the 1501 subjects who met inclusion criteria. Overall, 449 (30%) had a PHQ < 15 at the time of referral indicating mild to moderate depression; 1050 (70%) patients had a PHQ of greater than or equal to 15 indicating moderate to severe depression.
The mean waiting time between referral and first appointment was 17 days, and the median wait 15 days. Eighty-seven percent of patients were seen within four weeks. Longer waiting times often resulted from patient requests, or reflected staffing problems because of absence or annual leave.
The mean number of contacts for people completing treatment was 5.0 (median 5), and for those disengaging from treatment was 4.1 (median 4). Total mean contact time was 151 min for patients completing treatment (median 135 min), and mean contact time for people who disengaged was 95 min (median 80).
Table 3 shows the mean drop in PHQ from the score recorded by GP at referral to the last available score recorded by the ‘Doing Well’ team.
Table 3 Change in personal health questionnaire by discharge group
For the treatment complete groups this was the PHQ recorded on discharge. For the disengaged and transfer groups, the PHQ was the ‘last observation carried forward’ from the final contact.
A drop of five points or more is considered a clinically significant change in depression status. The mean drop of 10.6 PHQ points in the treatment complete group is therefore both clinically meaningful and statistically significant (P < 0.001). Although statistically significant, the fall in PHQ noted in the disengaged (1.4 PHQ points; P < 0.001) and transfer groups (1.3 PHQ points; P < 0.01) does not represent a clinically meaningful change.
The lack of response to treatment in the disengaged group was evident early in treatment (Figure 4). Response in the treatment complete group was maximal at four to five contacts.
Figure 4 Average PHQ scores by contact for disengaged and treatment complete groups
There was a significant effect of gender on PHQ outcomes, with women showing an average PHQ improvement of 1.36 points greater than men (P = 0.006). Age groups over 35 years were significantly different from the 16 to 25-year age group (P < 0.001). There was no significant effect of deprivation on outcomes measured by the PHQ (P = 0.28).
A final PHQ score of less than five was considered to be ‘remission’, (Kroenke et al., Reference Kroenke, Spitzer and Williams2001) and a PHQ score of less than 10 was considered to represent partial recovery. Since such terminology is only appropriate for clinical cases, the following analysis was conducted for the 1082 people referred to Doing Well with a PHQ score of 10 or over in the ‘treatment complete’ (n = 602), ‘transfer of care’ (n = 180) and ‘disengaged’ (n = 300) groups. Figure 5 shows that the proportion of patients referred with a PHQ score of 10 or more, who left treatment with a PHQ of less than five, was 42% in the treatment complete group, and 1% and 3% in the disengaged and transfer of care groups, respectively. Seventy-nine percent of the people in the treatment complete group had a PHQ score of less than 10 at the end of treatment, compared with 7% in the ‘disengaged’ and 10% in the ‘transfer of care’ groups. These changes are statistically significant (χ 2 = 248.471, DF = 2, P < 0.001).
Figure 5 Proportion of patients with a final PHQ of less than five (‘remission’) or less than 10 (‘partial recovery’) categories (initial PHQ ⩾ 10; n = 1082)
Of the 1169 people who attended at least one appointment, 610 (52%) were prescribed an antidepressant at some point during their treatment. The rate of antidepressant use increased in proportion to the severity of depression as measured by the PHQ, as shown in Figure 6.
Figure 6 Percentage of patients receiving an antidepressant prescription by PHQ score at referral (total subjects 1169; PHQ 0–4 n = 10; PHQ 5–9 n = 76; PHQ 10–14 n = 259; PHQ 15–19 n = 408; PHQ 20–27 n = 385)
Discussion
A diagnosis of depression covers a wide range of situations: from low mood and anxiety related to personal crises, through to major depressive illness with psychotic symptoms. This heterogeneity has led to depression being criticised for being ‘an over-inclusive term with a lackof conceptual clarity between symptom, syndrome, episode andillness’ (Casey et al., Reference Casey, Dowrick and Wilkinson2001). The National Institute for Clinical Excellence (NICE) stated:
“the most significant limitation is with the concept of depression itself… it is too broad and heterogeneous a category, and has limited validity as a basis for effective treatment plans” (National Institute for Clinical Excellence, 2004).
The difficulties caused by this lack of diagnostic clarity are compounded by two further issues. First, depression is a disorder that tends to remit quickly: the median time to recovery for depression is 12 weeks or less (Ustun and Kessler, Reference Ustun and Kessler2002). Whether psychological or pharmacological interventions are being evaluated, it is likely that 50% of any research group will recover within eight weeks (Andrews, Reference Andrews1996). Second, the placebo effect in antidepressant trials is unusually large, accounting for about 60% of the improvement shown by active drug groups in clinical trials (Andrews, Reference Andrews2001).
Depression is therefore difficult to define, and the effectiveness of treatments can be hard to assess. The methodological problems faced by researchers have clinical consequences: GPs have to recognise, diagnose and manage a protean condition with a complex aetiology, which may require medical, social and psychological intervention or none at all.
A substantial research literature describes how ‘service delivery systems’ can best be organised to deliver care. This evidence supports system redesign involving ‘collaborative care’ for people with depression. Although this term includes diverse approaches to treatment, Bower et al. (Reference Bower, Gilbody, Richards, Fletcher and Sutton2006) helpfully categorise them into three groups: (1) the introduction of case managers to support treatment; (2) efforts to improve liaison between primary and secondary care; and (3) the introduction of methods to monitor and share information about the progress of individual patients.
Meta-analysis of 34 studies has identified four effective components of these complex interventions: patient recruitment by systematic identification, the use of case managers with a mental health background, improved liaison between primary and secondary care and provision of regular supervision for case managers.
Unfortunately, the applicability of these studies to primary care in the United Kingdom is limited. Twenty-seven out of 34 studies were located in the United States (where their effectiveness proved to be greater than in trials conducted elsewhere). In 18 trials, patients were specifically prepared to take antidepressant drugs, and in 21 studies psychological therapies were not available as a treatment option (Bower et al., Reference Bower, Gilbody, Richards, Fletcher and Sutton2006).
Clinicians and managers therefore face the following challenges when providing care for people with depression:
• How to respond appropriately to the clinical uncertainties relating to treatment for individual patients.
• How to implement effective system redesign and ensure that this can be sustained over time.
• How to ensure that redesigned systems have the capacity to cope with the high incidence and prevalence of depression in the community.
‘Doing Well’ responded to these issues by implementing significant changes in four areas of service delivery.
First, ‘Doing Well’ represented a substantial service reorganisation. This was a complex intervention, implementing a number of changes simultaneously. It was also an ‘integrating’ intervention, making sense of a range of existing therapeutic options, including self-help, pharmacotherapy and secondary care specialist care. Secondary care expertise was deployed in primary care, and ‘service delivery systems’ were developed to minimise waits and waste.
Second, the redesigned system sought to make better use of information. The PHQ was recorded electronically at every contact, and was routinely used to guide care, monitor outcomes and distribute work within the team. It became a useful ‘shorthand’ for talking about depression severity and was widely used in discussions between GPs, mental health workers and patients. Systematic use of objective measures in this way was unusual in mental health systems in the United Kingdom at the time of this initiative. Implementation of the Quality Outcomes Framework in general practice from 2006 introduced the widespread use of objective measures of depression severity. The high rates of completion of the PHQ in this programme are consistent with recent studies suggesting that the measure is acceptable to both patients and GPs (Dowrick et al., Reference Dowrick, Leydon, McBride, Howe, Burgess, Clarke, Maisey and Kendrick2009). An assessment of the relative accuracy of the PHQ was beyond the scope of this study, though recent work has suggested that this measure may overestimate the prevalence of depression compared to the Hospital Anxiety and Depression Scale (Cameron et al., Reference Cameron, Crawford, Lawton and Reid2008; Kendrick et al., Reference Kendrick, Dowrick, McBride, Howe, Clarke, Maisey, Moore and Smith2009).
Third, ‘Doing Well’ provided a psychological approach to care (guided self-help) as the default intervention for all patients, supplemented by antidepressants or more intensive psychotherapies where required. The emphasis on prompt (though brief) psychological intervention inverts traditional practice, which tends to lead with drug treatment or to refer a small proportion of patients to specialists for relatively lengthy and sophisticated psychological treatment.
Finally, ‘Doing Well’ invested in staff training and well-being so as to support their engagement and relationships with patients. Staff encouraged patient choice and participation in the decisions about treatment. Although staff satisfaction was not assessed as part of this study, the clinical training, work schedule and support provided by ‘Doing Well’ was intended to encourage staff to feel autonomous, respected and confident in their work. It seems likely that this sense of confidence and optimism shared between staff and patients was an important aspect of the therapeutic relationship and will be investigated in a future study.
Strengths and limitations of the study
This study included a large cohort of unselected patients presenting with depression and low mood in everyday primary care practice. The routine use of the PHQ as part of an electronic clinical record generated substantial outcome and service use data for those participating in treatment.
The principal weakness of the study is the observational design and the lack of a control group. Given the propensity of mild-to-moderate depression to remit spontaneously over time, a randomised controlled trial comparing ‘Doing Well’ stepped care against ‘treatment as usual’ would be required to demonstrate greater efficacy and effectiveness.
Little is known about the outcomes of those who did not attend, or who dropped out of treatment. There are no follow-up data after discharge. The impact of ‘Doing Well’ on GP workload was not examined in this study. There was no economic evaluation. The impact of the programme on antidepressant prescribing is described in a separate paper.
Clinical outcomes
A small team of 5.7 whole-time equivalent clinicians was able to provide effective psychological and pharmacological care for about 100 new cases of depression each month. The stable referral rate over two years suggests that some equilibrium between demand and capacity had been reached. Almost all patients were seen within two weeks, which was significantly faster than standard care in this setting.
No data are available for comparison with clinical outcomes in routine care, but the mean drop in PHQ of 10.6 in the treatment complete group represented significant clinical improvement. Similarly, 79% of patients who were referred with a PHQ of 10 or more showed a partial recovery, and 42% achieved remission.
Table 4 compares the ‘Doing Well’ intervention described here with a similar programme in Doncaster, England, funded by the Improving Access to Psychological Therapies programme in the NHS in England (Clark et al., Reference Clark, Layard, Smithies, Richards, Suckling and Wright2009; Richards and Suckling, Reference Richards and Suckling2009). Both programmes focussed on depression treatment using a stepped care model based on guided self-help. The referral rate, staff complement, depression severity at referral and duration and number of contacts were all equivalent. Clinical outcomes were comparable, with a mean PHQ fall of 8.5 for patients with depression of less than three months’ duration, similar to that in the ‘Doing Well’ programme of 10.6 points.
Table 4 Comparison with IAPT site in Doncaster
IAPT = improving access to psychological therapies; PHQ = personal health questionnaire.
1Includes six vacancies as of September 2007.
2Treatment complete group.
3Treatment complete and transfer-of-care groups where illness duration is <3 m (n = 226).
The ‘Doing Well’ service seemed to be effective for both men and women, across all ages and for people living in deprived areas as well as those from more affluent backgrounds.
Twenty-two percent of patients did not attend treatment, which is consistent with a 20% non-attendance rate in psychiatric outpatient settings (Mitchell and Selmes, Reference Mitchell and Selmes2007), but is significantly lower than the 39% non-attendance rate for a primary care mental health team described by Gilbert et al. (Reference Gilbert, Barkham, Richards and Cameron2005).
There is evidence that outcomes for patients who ‘disengaged’ were significantly worse than for those who continued in treatment; this may also be true of those who failed to attend any appointments. Given the difficulties that people with depression may have in accessing treatment because of their condition, this finding requires further investigation.
The impact of ‘Doing Well’ on GP workload was not directly examined in this study, but we note that the programme was initially welcomed by all local GPs and sustained by participating practices for more than two years.
NICE identified 25 interventions that support effective depression care, almost all of which were implemented by ‘Doing Well’ (Table 5). There is therefore some justification for assuming that the implementation of this pragmatic approach is likely to have had a real effect. However, this initiative does not permit the identification of ‘active ingredients’ (Bower, Reference Bower2003) within such a multifaceted intervention.
Table 5 NICE guidance on management of depression (National Institute for Clinical Excellence 2004; ticks indicate ‘Doing Well’ compliance with these recommendations)
1‘Doing Well’ did not carry out population-based screening for depression, but would systematically assess individuals on the basis of their individual presentations.
2Although the intervention is designed to be brief, ‘Doing Well’ does advise on continuing treatment.
Although we can make reasonable predictions about what best practice for depression care should look like, it is difficult to identify potential mechanisms from this complex interaction of different factors, and we still do not know what critical mass of change within such interventions is needed to be effective.
Despite the limitations of this study, this description suggests that the ‘Doing Well’ model of collaborative care endorses the general applicability of this approach in British healthcare settings. These findings are consistent with the international literature on ‘collaborative care’ for depression, but are more generalisable to routine care in the United Kingdom.
There may be advantages compared to standard NHS care for depression, particularly in terms of waiting times, access to psychological therapies and rational prescribing for depression.
The comprehensive implementation of NICE guidelines for depression is feasible in primary care, and this model of care seems to be welcomed by GPs. However, despite these encouraging findings, the clinical efficacy of adherence to the whole NICE ‘package’ of guidance on depression has yet to be established.
Acknowledgements
The ‘Doing Well’ clinical programme and associated research were funded by the Centre for Change and Innovation in the Scottish Executive Health Department between 2004 and 2006, and by NHS Greater Glasgow and Clyde thereafter. Neither funding source had any involvement in the production of this paper. Ethics approval for the study was granted by the Local Research Ethics Committee for NHS Greater Glasgow & Clyde (Ref AC04/073).
