You are being asked to justify, or plan for, the number of children and youth that you are serving in your system. Too few, and you risk underserving your population. Too many, and you risk overpathologizing your population. How do you come up with an estimate of how many children and youth you should be serving?

How do I use this?
Annual prevalence estimates give you the number of children and youth who should be served each year by your service system. Comparing the number of persons served to the number estimated to be in need serves as a rough check on whether the number of service slots available in the system meet the mental health service needs of your population. However, this will not tell you whether the persons you're serving are actually the persons in need of mental health services. It's entirely possible that the people in services are not the persons most in need of services; we need clinical data on the children being served to insure that they indeed are those persons most in need of services.

Steps in the process:
In the best case scenario, you would have epidemiological data that would allow you to answer this question directly. However, the United States government has never conducted such a study. There are no nationwide epidemiological data on the prevalence of children and youth's mental health needs. This puts us in a difficult position. Because there are no such data, we have to turn to consensus estimates based on local, regional, and even international studies from geographies with conditions similar to ours. The technical term for a prevalence estimate generated this way is a 'synthetic prevalence estimate.' It does not involve local data collection. Rather, it relies on a series of steps for creating the most reasonable estimate of the local prevalence. The two documents below walk you through the steps used to create synthetic prevalence estimates for two different locales. The first walks through the process used to estimate the service population in Massachusetts. It was created in the remedy phase of the Rosie D. lawsuit in Massachusetts, for parties involved in that lawsuit. The second was created for the city / county of San Francisco as part of their annual state review by an external quality review organization (EQRO). Each was designed to be completely transparent about the data and assumptions employed in creating the estimate.

What do I still not know?:
Synthetic prevalence estimates differ in terms of the detail that they provide. For instance, a synthetic prevalence estimate may simply give you a single number of how many children and youth should be served in a given locale. Or it may be constructed in such a way as to give even more detailed estimates, such as variation by poverty status, age, or ethnicity. The critical issue is how rigorously these estimates have been constructed: more detailed estimates require additional assumptions to be made.