Patient and carer contact recording

Recording of all contacts by teams with both patients and carers was undertaken during March 2002. This used a contact recording system already developed by Ford et al (Reference Ford, Repper and Cooke1993) and further adapted for use in the UK700 study (Reference Burns, Fiander and KentBurns et al, 2000), after some simplification and re-piloting. Individual workers, using booklets for each of the 4 weeks, recorded data on duration, site and primary purpose of each contact (telephone as well as face to face). The research assistants remained in regular contact with the teams during the month. Reliability checks on the contact recording were carried out by comparing the contact recording data with case note contact data collected in two censuses.

The patient and carer contact recording data were used in descriptive analysis of the provision of team contact with patients and also as a variable (‘proportion of patient contact in vivo’) in the team typology analysis.

Who are these assertive outreach teams?

Table 1 lists results from the Team Organisation Questionnaire, their means and ranges and ‘patient and carer contacts’. Thirteen teams (54.2%) identified their locality population by aligning by borough, with a further eight (33%) by geographical area. Seven of the 24 assertive outreach teams were run by not-for-profit voluntary agencies.

Patient and carer contact recording

Cluster analysis results

Table 2 lists the variables used in the cluster analysis, their distribution and (where appropriate) their means and ranges, for both the total sample and for the three clusters identified from the dendrogram illustrated in Fig. 1.

Fig. 1

Dendrogram of teams using average linkage method. All methods use the L1 dissimilarity measure and ranked data.

Cluster C, the most distinct cluster identified by the analysis, consists entirely of voluntary agency teams that do not hold formal clinical responsibility (e.g. the Care Programme Approach, CPA). Inevitably, they lack integrated health and social care provision and dedicated beds. These teams have no input from a psychiatrist, fewer disciplines represented (mean=2.7) and tend to be smaller (mean=6.2 FTE staff, range 5–9). However, they provide the highest percentage of ‘in vivo’ contacts to patients (45%). They have very little service provision outside of weekday office hours, with only one team offering some weekend service.

Clusters A and B comprise teams with CPA responsibility, although cluster B includes one voluntary agency team. All but one of the teams provide integrated health and social care. Clusters A and B differ in several regards. Cluster B teams have no psychiatrist input and no dedicated in-patient beds, whereas cluster A teams had a mean of 1.0 FTE psychiatrist per 100 patients and 36% had dedicated in-patient beds. Cluster A teams also tended to be more multi-disciplinary (mean of four per team v. three in cluster B) and to have more part-time staff (full-time: part-time ratio of 2.0 v. 5.8). They also tended to have smaller individual case-loads (mean of 8.7 v. 10.4) and to operate more outside of office hours (57% of teams v. 25%).

The frequency of patient contacts also varied by cluster. The overall figure for all teams is approximately 1.3 (s.d.=0.5) contacts per week. However, this varied with 1.36 contacts per week for cluster A teams, 1.45 for cluster B teams and 0.97 for cluster C teams. Thus, the voluntary sector teams of cluster 3 showed a tendency to have fewer contacts per week but with a higher proportion of contacts taking place in the patient's home or community setting.

Sensitivity analyses

Use of the L2 metric clearly identified clusters A, B and C (Fig. 1). Scaling by the standard deviation and the range also clearly identified cluster C but failed to identify cluster B. This is probably because the distinctive characteristics of cluster B are described by skewed continuous variables that have less influence with these approaches. Omitting the ‘percentage of patient contacts in vivo’ variable did not detract from the clear identification of clusters A, B and C. Taken together, these analyses support the definition of three main clusters.

To see whether any individual teams were highly influential, we omitted one team at a time and examined the impact on the classification of the remaining teams. Of these 24 analyses, 20 identified cluster C as most distinct, followed by clusters A and B. One analysis identified cluster A as most distinct, followed by clusters B and C. The remaining analyses separated cluster A from cluster C but did not identify cluster B.

Correlation with independent scales

The Dartmouth Assertive Community Treatment Scale

The mean DACTS score for all teams was 3.4 (s.d.=0.4), with a range of 2.3–4.1. Three teams (12.5%) scored a mean of 4 or more (usually taken as ‘high fidelity’), seventeen teams (71%) scored means of 3–3.9 and four teams (16.5%) scored means of 2–2.9 (‘low fidelity’).

The 28 individual items in the DACTS were ranked according to their mean score. Nine items had a mean score of 4–5 (high fidelity), eleven items had a mean score of 3–3.9 and eight items had scores of less than 2.9 (low fidelity). Of these, four had a mean score of 2–2.9 and four had a mean score of 1–1.9. Table 3 lists those variables where London teams showed high and low fidelity.

Table 3

Dartmouth Assertive Community Treatment Scale variables demonstrating ‘high model fidelity’ (> 4) and ‘low model fidelity’ (< 3) across London assertive outreach teams

Score of ≥ 4 (‘high fidelity’ items)	Mean	Score of < 3 (‘low fidelity’ items)	Mean
H1 Small individual case-load	4.7	H7 Senior permanent psychiatrist on staff	2.9
H6 High proportion of team posts filled	4.1	H9 Substance misuse specialist on staff	2.2
O1 Explicit intake criteria for patients	4.4	H10 Vocational specialist on staff	1.8
O2 Intake rate of patients low	4.9	O4 Responsibility for 24-h crises	1.9
O3 Full responsibility for treatment services	4.2	S5 Frequency of contact	2.7
O7 Time-unlimited services	4.4	S7 Individualised substance misuse treatment	2.7
S1 High proportion of service in vivo	4.1	S8 Dual disorder treatment groups	1.1
S2 ‘No drop-out’ policy	4.4	S10 Role of users on team	1.7
S3 Assertive engagement mechanisms used	4.5

Figure 2 shows the mean DACTS score for all teams and for clusters A, B and C. It also presents the DACTS mean scores by its three dimensions: human resources, organisational boundaries and nature of services.

Fig. 2

Mean Dartmouth Assertive Community Treatment Scale (DACTS) scores for all teams, for each cluster and for each dimension of the scale.

Teams in cluster A tended to show higher fidelity in all three dimensions, although this was less marked in the ‘nature of services’ dimension. In the ‘human resources’ dimension, cluster C's fidelity score was reduced by their employing support workers without mental health qualifications. However, they also scored lower on other items in this dimension that were not staff-dependent: degree to which a team approach is used, how often a team meets to review all patients and how clinically active the team leader is. Within organisation boundaries, two of the seven items differentiated cluster C teams from A and B. Not surprisingly, given their voluntary agency status, these were their lack of involvement in the hospital admission and discharge of their patients.

The ‘nature of services’ dimension did not clearly differentiate between the three clusters.

The International Classification of Mental Health Care

Each of ten modules of care are scored high (3), medium (2) or low (1) on the level of specialisation provided by the team. It is also possible to rate the service as not applicable to the module of care (0). Figure 3 shows the mean ICHMC scores for each module of care, both for all teams and by cluster.

Fig. 3

Mean International Classification of Mental Health Care scores for each module of care, by all teams and by team cluster: module 1, establishing and maintaining relationships; module 2, assessment; module 3, care coordination; module 4, general health care; module 5, taking over activities of daily living; module 6, psychopharmacological and other somatic interventions; module 7, psychological interventions; module 8, (re)educating basic, interpersonal and social skills; module 9, interventions related to daily activities; module 10, interventions aimed at family, relatives and others.

All teams scored 3 for module 1, ‘establishing and maintaining relationships’. Teams scored highly on modules 2 (‘assessment’) and 3 (‘care coordination’), with means of 2.4 and 2.5, respectively. Overall mean scores for other modules were lower and mostly in the range 1.5–2. Cluster C teams consistently scored lower than cluster A and B teams in all modules other than module 1.

Role of the voluntary sector

Almost one-third of London's assertive outreach teams are run by voluntary agencies. One further team previously had been non-statutory but was recently incorporated into a National Health Service trust. These non-statutory teams were clearly identifiable as a distinct group on the cluster analysis as well as on the DACTS and ICMHC, where they held a lower fidelity to the assertive outreach model as measured by the DACTS and offered lower levels of specialisation on the ICMHC modules.

The cluster C teams differed in other ways also, often having a specific target group or holding to a particular ideology of care: two teams have a particular focus on African–Caribbean patients; one team focuses on young people (less than 25 years old); and another team targets asylum seekers and recent immigrants. Several spoke of working to a ‘social inclusion model’ and of holding a ‘true team approach’. In two teams all staff are employed as mental health workers or support workers rather than by their professional backgrounds, and in one team all staff received the same salary. Voluntary agency staff were also more likely to be reported as living in the area served. One team spoke of having a ‘pre-engagement’ role, with a view to later engagement of their clients in the local statutory services.

Voluntary agency teams may not play such a strong role in assertive outreach provision outside London but their role within London is clearly significant. There would appear to be a need for more consideration of the place of the non-statutory sector teams within the local health economies and service provision. Their role in implementing the statutory responsibility of the CPA – a set of principles for organising mental health care in the UK that is required in law – is already developing. One team already holds CPA responsibility and agreement is in place for a further team to take this on in the near future. Only one voluntary agency team reported poor relationships with local statutory services.

Integrated health and social care provision and hospitalisation

An integrated health and social care service was one of only two components demonstrated to be associated with reduced hospitalisation in a systematic review of home treatment services (Reference Burns, Knapp and CattyBurns et al, 2001). In our sample only half of the statutory teams fulfilled this integrated provision, although it was clear that this was changing rapidly: 79.2% of teams (19) had no designated in-patient beds and only 29% (7) retained any medical responsibility for their admitted patients. There is a fault-line in patients' care at the point of hospital admission. Receiving teams may not fully understand or accept assertive outreach approaches, and thereby fail to achieve the earlier discharge that has been identified as one of its strengths. This has implications for expectations of the effectiveness of assertive outreach teams in reducing the length of hospital stay. Continuity of responsibility of medical care across community and in-patient provision by assertive outreach services requires careful consideration.

Contact with patients

Several of the assertive outreach teams within the study were already employing some form of patient and carer contact recording, either as part of a wider patient information database or for the purposes of clinical audit. Although issues of quality and comparability made it impossible to use these in this study, this is an encouraging development. These do not, as yet, record the provision of specific, defined interventions, and it is highly likely that it is such interventions (rather than patterns of contacts) that affect outcomes (Reference Burns, Creed and FahyBurns et al, 1999).

The Dartmouth Assertive Community Treatment Scale

The mean score on the DACTS for all teams was 3.41, which would indicate moderate fidelity. Compared with the US teams investigated by Teague et al (Reference Teague, Bond and Drake1998) this is close to the ‘ACT-like’ teams (mean=3.47) but lower than the ACT replication teams (mean=4.01).

More striking, however, is the great variation in the London teams' DACTS mean scores. Most teams (71%) reflected a moderate level of fidelity, with three teams scoring above 4. However, with the highest at 4.14, this indicates that even these teams were not implementing some of the features of the model. Four of the teams scored a mean of below 3, suggesting that there were many items of the DACTS that were not being adhered to. Cluster A teams have the highest fidelity scores. When ranked by the DACTS mean score, ten of the top eleven teams were from cluster A. Conversely, four of the six cluster C teams had the lowest DACTS means (the other two cluster C teams being ranked sixteenth and seventeenth).

The DACTS is a multi-dimensional measure and mean scores have limited value. Nine individual DACTS criteria scored a mean of 4 or above, spread across the three dimensions of DACTS, indicating that most London assertive outreach teams are doing well at implementing these elements of ACT. Some of these were expected, given our team inclusion criteria, such as having ‘small individual case-loads’ and ‘time-unlimited services’. Others included providing a ‘high proportion of service in vivo’ and the team having ‘full responsibility for treatment services’ (such as housing and employment support). However, eight DACTS items scored a mean of less than 3, suggesting that London assertive outreach teams differ significantly from their US counterparts in score. Of these eight, three related to expertise and service provision in substance misuse care and a further two to the low level of senior psychiatrist and vocational specialist input to the teams. The ‘frequency of contact’ with patients, the ‘role of users on team’ and having ‘responsibility for 24-h crises’ also showed poor fidelity within DACTS.

Exporting health care structures from one national system to another is likely to reveal differences. Examples of this are: the presence of staff in the team who have at least one year of training or experience in substance misuse or vocational specialities; and the details of what ‘integrated health and social care’ means. An understanding of how the teams fit into their own national system of health care is essential for interpreting their ‘fidelity’ scores.

In addition, we found that the DACTS omits team characteristics that we considered important, both from the literature and from clinical experience for assessing care within a non-US context. Only 6 out of 14 variables that were considered in the cluster analysis are reflected in the DACTS. We believe that the DACTS is a more ‘culture-bound’ instrument than has been acknowledged previously and we would recommend that in a non-US context the other eight items used in this cluster analysis be included.

The DACTS overall mean score tells us little about the profile of the service characteristics of teams. It is quite possible for two teams to score equally on the overall DACTS while incorporating substantially different components of care. More work needs to be done to establish the relative importance of different components of care within the assertive outreach model.

Implications for future assertive outreach services in the UK

The role of assertive outreach teams in mental health services in the UK is developing rapidly, with policy support in the National Service Framework for Mental Health (Department of Health, 1999) and the Mental Health Policy Implementation Guide (Department of Health, 2001). Our study suggests that currently the term ‘assertive outreach team’ is not indicative in London of a single implementation model, nor necessarily of teams including the characteristics required by Department of Health guidance. Recruiting clinical psychologists was reported as particularly difficult and few teams had achieved it. Similarly, there were some teams with a high proportion of part-time staff, which was seen as unsatisfactory. It is also clear that some of the accepted wisdom of the US ACT model and the means of assessing its implementation through the DACTS do not easily translate to the UK system of care. Assertive community treatment is being adopted as policy in a number of countries outside the USA. We would propose that their likely heterogeneity presents a clinical challenge but also a research opportunity in distinguishing effective from redundant components of the prescribed model.

Members of the Pan–London Assertive Outreach Study Group

Tom Burns, Christine Wright, Peter James, Adele Greaves, Christine Benfell, Nan Greenwood (St George's Hospital Medical School); Paul Bebbington, Sonia Johnson, Joanne Billings (University College London and Camden and Islington Mental Health and Social Care Trust); Stefan Priebe, Walid Fakhoury, Joanna Watts (Unit for Social and Community Psychiatry, Barts and The London School of Medicine); Matt Muijen, Iain Ryrie, Rebecca Walwyn (Sainsbury Centre for Mental Health); Ian White (Medical Research Council Biostatistics Unit).

CLINICAL IMPLICATIONS

1. ▪ London assertive outreach teams vary considerably in their service characteristics and in their fidelity to the assertive community treatment model.

2. ▪ The teams form three groups on cluster analysis, supported by differences in scores on independent scales of model fidelity and interventions. Teams in the voluntary sector are clearly distinct in their practice.

3. ▪ Currently, assertive outreach teams in London do not necessarily match the characteristics proposed in UK guidance.

LIMITATIONS

1. ▪ We cannot assume that London teams are representative of assertive outreach teams elsewhere in the UK.

2. ▪ We cannot comment on the relative effectiveness of the teams in each of the three clusters.

3. ▪ Teams are rapidly evolving and practice was not consolidated in all services.