Effect of a multifaceted approach to detecting and managing depression in primary care

Background and setting

The prevalence of depression is 3-9%.1 It is more common than diabetes and asthma and, by 2020, will rank second only to ischaemic heart disease as a worldwide disease burden.2 About 90% of episodes of depression are managed in primary care. However, case identification and treatment are suboptimal. 3 4Interventions that focus on any one component of the identification or management process have demonstrable research efficacy but limited clinical effectiveness.5Models that target several aspects of depression management simultaneously may be more beneficial.6 This paper describes attempts to introduce such a multifaceted model.

Newcastle and North Tyneside District Health Authority met the costs of the project and provided a list of five general practices that might participate. From these, we recruited one large practice serving a population with low levels of deprivation and a less well resourced practice serving a more deprived area (see table 1 for details).

Practice A- The first practice had a list size of about 10 000 and served a predominantly affluent white British population. The practice team comprised six partners and 13 other staff, including a counsellor and a psychologist.

Practice B- The second practice had about 5500 registrations. It had two surgeries located in deprived urban inner city areas with high levels of unemployment and large ethnic minority populations. There were three partners, with a series of locums employed to cover one partner's study leave. The surgeries had access to a counselling service for individuals from ethnic minority groups. Each surgery referred cases to different community mental health teams.

The problem

The problem was how to implement effective methods of identifying and managing adult depression in general practice without distorting normal working patterns. We applied a “chronic disease management” approach, aiming to establish an integrated care pathway.7 A potential barrier to implementation was that primary care teams were under pressure to make other changes and had limited time and resources available. We seconded an H grade mental health nurse to facilitate the project, and a specialist registrar provided sessional input.

Key measures for improvement

We used a prospective before and after design to evaluate changes in depression management. We defined five measures of improvement in advance:

• All identified cases of depression to be placed on a register; an appropriate indicator would be if about 6% of patients aged 18-65 years were registered

• Antidepressants to be prescribed at agreed doses for the acute, continuation, and maintenance phases of depression

• Drugs to be prescribed from an agreed practice formulary reflecting clinical and cost effectiveness data (in practice A the drug of choice would be lofepramine8) and the “Northern Regional Quality Marker” (which states that selective serotonin reuptake inhibitors should comprise <25% of prescribed antidepressants)

• A call and recall system to be instituted so that patients newly prescribed antidepressants would be reviewed at two and six weeks

• Referrals to secondary care to be for reasons agreed at the “interface meetings” with mental health services (see below) and would comprise no more than 10% of cases seen.

We assessed staff views of the intervention using a modified version of the client satisfaction questionnaire.9 This was rated on a 1-4 scale (1=strongly disagree, 4=strongly agree).

Strategies for change

The project facilitator worked alongside a steering group at each practice to devise and deliver

• Staff training to enable reliable detection of cases of depression

• Specific computer codes that flagged cases and their clinical status on a register

• In-house treatment guidelines to aid case management

• Referral and shared care procedures agreed jointly with mental health services.

Effects of change

Practice A— At baseline, 285 cases of depression were known to practice A, compared with the predicted 399 (95% confidence interval 331 to 414). After intervention, 362 cases (6.6%) were registered (see table 2). Although only 66% (51/77) of new cases of depression were placed on the register, all other aspects of case management improved after the intervention. There was no record of prescribing not in accord with the guideline, and lofepramine comprised 56% of the antidepressants prescribed. The mean client satisfaction questionnaire score for staff (n=9) was 3.3 (SD 0.8).

Practice B— The predicted number of cases of depression for practice B was 208 (95% confidence interval 137 to 292). At baseline, 52 cases (1.6%) were identified, and after the intervention 71 cases were registered (2.2%). An improvement in sensitivity after training was countered by a decrease in specificity (see table 2). No systematic data were available from the practice after the active phase. The mean client satisfaction questionnaire score for staff (n=5) was 2.9 (SD 0.7).

Lessons learnt and next steps

The results of our intervention might have been easy to predict given the contrasting resources and capacity available in the two practices. Certainly, a key handicap at practice B was the more primitive information technology system and the greater reliance on handwritten information. Familiarity with the philosophy of chronic disease management probably affected engagement with the project. Staff at practice A had previous knowledge and experience of integrated care pathways for hypertension and diabetes; those at practice B did not. The completion of the planned intervention was more difficult at practice B because it was harder to maintain the involvement of the same core staff at each stage. The general practitioners in particular found it difficult to attend training sessions unless locums were provided, and training exercises sometimes had to be repeated. This partly explains the failure to complete the process: it proved impossible for everyone to meet to agree the in-house treatment guideline.

Beyond these tangible barriers, other elements were also important. For example, ownership of the project at practice A clearly extended beyond the steering group. Several individuals, including the practice manager, provided active leadership. Team members seemed committed to the process and showed willingness to review and change their practice. At practice B there was no consensus onhow depression should be managed, and difficulties in giving priority to the project meant that a shared understanding never developed. Once the input of the project team was withdrawn, staff at practice B could not sustain their planned work on the register or their guideline.

Practice B is far removed from the research settings where models of primary care management ofdepression are usually tested. This project suggests that further modifications of the depression management model are needed if it is to help staff and patients in more disadvantaged circumstances. It may not be feasible to introduce this model unless a practice has prior experience of chronic disease management and computerised records are used routinely. If these components are in place, two basic challenges remain—namely, active leadership from within the primary care team and sufficient incentives for a critical mass of team members to engage in the process. Even then, prioritising this work over other commitments is unlikely without a strong steer from primary care trusts and strategic health authorities.

In summary, attempts to implement “whole system” approaches to depression management in primary care produce inconsistent results.15^–17The key elements for success are probably organisational change combined with individual commitment to behavioural change. However, the basic challenge of creating and maintaining an effective system of management of depression in primary care without the artificial support of a research project remains.17