Few studies of collaborative programs have included formal cost benefit analyses. Among those that do, differences in costing methodologies make comparisons difficult.
Three studies report cost savings for shared care programs, compared with care provided in traditional ways. Goldberg and others (1996) studied 30 patients suffering from new episodes of depression or anxiety disorders who received care in a hospital-based psychiatric service and compared their treatment costs with the costs for 30 patients matched for diagnosis and severity of illness in a primary carebased mental health team. There were no differences in clinical outcomes, but the primary carebased group had significantly lower health services costs because patients were less likely to be admitted to hospital and were less likely to be seen by a psychiatrist. When the authors expanded their analysis to include all patients with these diagnoses treated in both locations over a 1-year period, they found that 190 patients were treated by the primary carebased team for less than the cost of 78 patients treated by the hospital-based service.
Katon and others (1997) compared the costs of treating patients with major depression who were randomly assigned to a collaborative, multifaceted treatment intervention with the costs incurred by a control group who received usual care from their family physician. They found that the average cost of depression treatment per patient for 1 year in the collaborative intervention group was $1337, compared with $850 in the usual care group. When the authors calculated the costs per successful outcome (as defined by a 50% reduction in SCL-90 scores at 4 months), they found that the costs for the intervention group were $1783, and the costs in the usual care group were $1940. The authors concluded that the care was more cost-effective in the collaborative intervention group.
Kates (1988) compared the costs associated with providing consultations in family physician offices with the costs incurred for new consultations in a local community mental health clinic and found that the office consultations were $100 less expensive, with the saving being primarily due to fewer follow-up visits.
Two studies found no cost savings associated with collaborative care models. Meadows (1998) compared costs for treating 110 patients with chronic mental illness in a primary care consultation-liaison service after they had been transferred from a traditional outpatient clinic and found that the transfer was cost neutral. Shah (1995) used data from government sources in the UK and constructed a theoretical model to estimate the costs for hospital-based psychiatric services and those delivered in the primary care setting. He found that the operating costs of both service modalities were similar.
The data provided by these studies are very limited, but they suggest that costs for delivering psychiatric services in the primary care setting may be similar to costs in traditional outpatient services, or slightly reduced. The sources of cost savings appear to be as follows: greater treatment effectiveness for some diagnoses, reduced admissions to hospital, other mental health workers taking the place of the psychiatrist when appropriate, and avoidance of unnecessary follow-up visits.
In general, program objectives and program evaluation should reflect the larger goals of shared mental health care. These goals are to improve the accessibility and quality of mental health care available to primary care patients and to make more efficient and effective use of health care resources through the following actions:
Additional goals may include the following:
It is important that shared mental health care initiatives focus on achieving these and related goals and that they measure their success in doing so.
It is important that measurement practices be consistent enough to permit comparisons between programs and to permit a coherent body of knowledge about what works and what does not to develop over time. We suggest that, in describing and evaluating shared mental health care programs, the following list of potential descriptors and outcome measures be considered and that applicable items be systematically dealt with in a manner replicable by others.
1. Program and contextual variables. What is the setting, and what are the sources of funding? What are the programs inclusion and exclusion criteria? What are its policies regarding secondary referral? What are the program resourcesthe personnel (number and type) and dollar values required to make the program work? How does the program fit into the local mental health care system?
2. Patient variables. What are the age, sex, and demographics of patients served in collaborative care programs. What are their medical and social comorbidities?
3. Reason(s) for referral. More detailed and standardized information is needed about why family physicians refer patients and what they are requesting.
4. Presenting problems. More detailed descriptions of patients presenting problems and more objective measures of functional status prior to treatment are needed.
5. Diagnoses. Diagnostic schedules appropriate to primary care should be used, and more complete reporting of all diagnoses and identification of most responsible diagnosis should be encouraged.
6. Treatment. More detail is needed about the nature and duration of treatment provided, including the care provided to patients with complex or difficult disorders who are not referred on to secondary care. Evaluation of treatment should also include some attempt to measure the appropriateness of referrals to secondary care and the appropriateness of decisions about hospitalization.
7. Patient outcomes. What are the short- and long-term outcomes of treatment provided in shared mental health care programs? Data collection for program evaluation of shared care initiatives should focus more on level of functioning and reduction in disability, as well as on symptom status, measures of overall health status, and (for chronic or recurring illnesses) length of time in the community and length of time between episodes.
8. Family physician outcomes. What changes take place in family physician recognition of mental disorders, treatment behaviours, and referral practices?
9. Costs. Standardized methods of reporting costs per patient or per episode of care should be established.
Standard implementation of a set of variables such as these would permit more meaningful interpretation of research and evaluation findings from individual programs and provide greater capacity for comparison among programs and among different models of shared mental health care. Data gathered over longer periods of time will become increasingly important as collaborative programs seek to demonstrate changes in patient outcomes, family physician knowledge and skill, and impact on the structure and function of the mental health care system as a whole.