For each client served, behavioral healthcare providers enter thousands of data elements into their information systems. Provider organizations need these data to perform billing functions and meet governmental reporting obligations. With the advent of electronic health records (EHRs), long-term clients might have tens of thousands of data entries, as well as substantial amounts of supporting text. These massive amounts of data capture a large portion of provider organizations’ activities, and organizational leaders should be using these data to improve their organizations—financially, administratively, and clinically.
These data often are organized into tabular format reports needed to conduct day-to-day business, meet audit trail requirements, and give providers the hard-copy comfort associated with extracting the data they have entered. These reports might have subtotals, attractive formats, and even a graph or two, but they remain primarily a regurgitation of the details that reside in providers’ IT systems.
These list-type reports remain essential, but they do not help providers analyze data. They do not provide data in a big-picture context for staff. They do not easily show where problems reside or opportunities exist.
Yet new products that can arm the behavioral health administrator, program director, or clinician with highly interactive tools are entering the market. Administrators and staff with an interest in evidence-based practices or performance-based reimbursement can use these new tools to easily access summarized, organized, and analyzed data, identifying trends and relationships to foster better decision making.
Providers want reporting tools that answer questions, or at least raise questions and point them toward topics that need further exploration. For example:
Under what conditions is group therapy as effective or more effective than individual psychotherapy? Is group therapy always less expensive to provide than individual therapy?
Under what conditions does an extended service period result in better outcomes? What is the impact on waiting lists if length of stay is increased?
Which payers have authorization policies that result in poorer outcomes? What impact do various authorization policies have on inpatient episode frequency?
One approach to this reporting need is to give data access to decision makers through an interactive dashboard interface, in which graphs over time and color-coded status gauges can be used to monitor performance and identify trends. The dashboard concept has been around for a while, but new tools provide the ability to interact with data displays. For instance, a mouse click might allow the display to switch results from Medicaid to commercial payers. A single dashboard can provide access to hundreds of graphs showing length-of-stay by program, location, diagnosis, etc., over time.
Furthermore, dashboards now can be outfitted with “what-if” features. A graph may show an organization's actual dropout statistics over the past year, but a what-if feature can allow users to explore the financial, staffing, and outcome impact if all programs were as effective as the high-retention ones (figure 1).
Easier Statistical Analysis
Behavioral health agencies face important questions about utilizing resources to have the greatest possible positive impact on clients, such as: What treatments are most effective with clients with a given set of characteristics? What practice patterns produce the best outcomes based on the organization's history? Providers’ information systems contain the details to answer these questions, but the necessary analysis has not been an easy task. In the past, complex statistical packages required users savvy in the world of chi-squares and correlation coefficients to explore questions of this nature.
New tools, however, shield the user from detailed statistical modeling and permit free exploration of how client demographic and clinical characteristics combined with particular treatment modalities, durations, provider characteristics, etc., lead to measurable outcomes. The software chooses the appropriate statistical test for the type of data being analyzed and provides graphical output summarizing the results.
For example, figure 2 shows that a particular agency serves many children and adolescents (the blue line) and does moderately well with regard to improvements in the Axis V Level of Functioning. Staff now might ask:
What mix of services is most effective for adolescents with a given diagnosis?
Does the primary clinician's degree (e.g., MSW, PhD, etc.) make a difference in outcomes?
Does the clinician's level of experience affect outcomes? Does it matter more for male versus female clients? How about for clients age 13 and over versus 12 and younger?
New statistical analysis and predictive modeling tools make answering these questions much easier.
There has been a breakthrough in data analysis tools, and information now can be delivered in usable formats that contribute to decision making. These new tools make use of the volume of data and the length of data history now accumulating in behavioral health information systems. These tools can help bring data-driven, evidence-based, 21st-century decision support to behavioral healthcare.
The authors are with The Echo Group: George Epstein is Chairman of the Board, Joseph Viger is Director of Operations, and Paul Kirsch is Marketing Manager.