The operation of managed behavioral health organizations is increasingly dependent on information technology. Developments in database software and new statistical techniques offer great opportunities to use empirical data to shape the policies and practices of these organizations. Managed behavioral health organizations can use information technology in their own decision making and to inform consumers, health care purchasers, regulatory agencies, and academic researchers about the effects of their practices. Accrediting agencies such as the National Committee for Quality Assurance and the Joint Commission on Accreditation of Healthcare Organizations request that managed behavioral health organizations release complex "performance measurements" that are designed to evaluate the organizations' performance.
To date, the use of information by most managed behavioral health organizations has been treated as "proprietary," and the information has not been available outside the organization. United Behavioral Health (UBH) has decided to take a different stance. It is making the results of research publicly available, including technical developments in areas such as provider profiling. The goal is to stimulate the kind of distributed, "open-source" development that is taking place in the software industry.
We contend that significant development of information technology will be vastly improved if managed behavioral health organizations adopt a nonproprietary development model similar to that of the open-source movement in the software industry (1). An open-source approach can benefit managed behavioral health organizations in several ways.
Computer source code is simply a set of instructions that describe how to build a computer program—analogous to a blueprint for a building. Source code is invaluable to a computer programmer who wants to understand how a program works, to extend the program, or to fix it. Recognizing that source code had tremendous value, academic computer scientists were among the first to share their code, making it available for scrutiny by their peers (2). Since then, open-source software development has become central to the existence of the Internet. This form of development is driven by its ability to harness legions of programmers working on problems in common.
A recent and popular example of the open-source approach to software development is the GNU/Linux project. This project endorses peer review but has an additional feature—a license to guarantee that anyone who uses the software is free to modify the source code for his or her own purposes. If a particular modification proves useful, its author is free to distribute that modification to others, who, in turn, are free to make further modifications. These and other features of the open-source approach would be beneficial for managed behavioral health organizations.
The arguments for open-source development in managed behavioral health, all of which are compatible with our competitive private health care system, are compelling. An overarching theme is that open-source development offers important economic advantages to individual managed behavioral health organizations and to the behavioral health care system as a whole.
To the extent that health care purchasers differentiate and select managed behavioral health organizations that make available their information technologies, code, and key data, these organizations will benefit from open-source development. Although not central to employers' purchasing decisions, UBH is increasingly asked to share its technical methods with current and prospective purchasers of behavioral health services, including methods for evaluating clinicians, performing case-mix adjustment, warehousing its data, and using new data warehousing technologies to examine the cost and quality of treatment.
Developing industry standards
Each industry develops standards that facilitate the overall growth of the industry. Such standards are almost nonexistent in managed behavioral health, even in areas in which competitor-specific development is wasteful and nonsensical. For example, given the high degree of clinician overlap among provider networks, does it make sense for each managed behavioral health organization to develop its own provider profiling methods, practice guidelines, and performance standards? Why should different managed behavioral health organizations use different treatment improvement measures?
Even in our competitive private system, managed behavioral health organizations that are in direct competition with one another have significant incentives to participate openly in the development of information technology standards. Failure to participate may mean that an organization remains silent or lags behind important new developments in the industry. In the software industry, Internet standards have emerged largely because the stakeholders have the choice of either participating in development or being left behind. For example, the World Wide Web consortium is an organization devoted to providing open technology standards for the Internet. Its members include technology leaders such as Microsoft, Sony, and Sun Microsystems. One of the primary motivators for these companies is the opportunity to ensure that the open standards are in line with the companies' respective business models.
Offsetting development costs
Technology development can be expensive—often too expensive to be shouldered by a single company. Distributing development costs across several companies can lower the cost for each company while producing a higher level of quality. For example, we know of no managed behavioral health organization that has the resources to address the methodologic challenges inherent in the single task of profiling clinicians, let alone to address policies and practices that affect many other aspects of the delivery of care. Meaningful progress in the development of appropriate methods for use by managed behavioral health organizations in the management of care will occur only if these organizations openly share their methods and data with each other and with health care purchasers, academic researchers, clinicians, consumers, and consumers' family members.
Improving quality with peer review
Peer review can dramatically improve the quality of any project. Adoption of an open peer-review process among managed behavioral health organizations would mean that every stakeholder in the delivery of behavioral health services could contribute to development by commenting, criticizing, making suggestions, or developing alternative approaches. Perhaps most important is the greater credibility that published performance measures would have. Academic standards of objectivity require that there be an opportunity to detect error and investigator misconduct. No such opportunity exists in the case of numbers released by managed behavioral health organizations to accrediting organizations and to others, despite obvious incentives and the potential for bias.
Improving communication among organizations
Communication is often considered a method for attaining an outcome, but it can also be considered an outcome of open-source development methods. As observed in the software industry, improvements in communication channels can result from collaborative efforts among managed behavioral health organizations. In addition to informal methods, such as Internet mailing lists, more structured approaches to communication can be developed. Many Internet standards have been developed through a request-for-comments process administered by the Internet Society (3). A request-for-comments document—for example, a technical report—results from the discussion process and can then be publicly referenced. Managed behavioral health organizations could use the same method, which would allow clinicians, consumers, and academic researchers to participate in the development of the system's information technology standards, including privacy and confidentiality standards.
Open-source development is not a panacea. Even projects that would benefit from open-source development involve risks. Early visionaries of the open-source software movement recognized that each project must have its own set of rules for development (4). For example, with some projects a single individual has complete control over the final choices; usually this individual is the person who initiated the project. With other projects, an elected committee or core team makes the final decisions. Some projects use a hybrid of these two approaches. We suggest that to control risk, managed behavioral health organizations should adopt a similar approach in which each open-source project develops a core team of individuals who establish a set of project rules or guidelines. These guidelines can be strict or loose, as long as they adhere to an open-source approach.
Licensing is another method that can be used to control risk. In the software industry, the participants in each project choose from among several software licenses. Some licenses are very specific about the current and future uses of the software—for example, the GNU General Public License holds that current and future extensions to the software must always include the source code. Other licenses are less restrictive—for example, the Berkeley Software Distribution License holds that current and future extensions to the software do not have to include the source code. The choice of license is open to the author of the code or the project team. Managed behavioral health organizations can adopt a similar approach. Each organization or project team could create its own license for a collaborative project, or the project itself could have a license.
Open-source development in managed behavioral health should include open peer review of the source code, algorithms, and data; freedom for anyone to contribute regardless of their organization or company; and sets of rules, guidelines, and licenses to reduce risk. Each of these appears to be a critical aspect of the open-source movement in the software industry, and we expect all these elements to be necessary in the behavioral health care world.
Two possible outcomes that are particularly important to behavioral health care are the development of a set of standards—for example, operational definitions, specifications, and data-exchange formats—and the development of a set of common tools—for example, source-code libraries and data analysis tools. These two elements are common to all managed behavioral health organizations. Thus all managed behavioral health organizations stand to benefit from open-source development.
Most open-source projects begin with only one or a few individuals who need to solve a problem. To this end, UBH has already begun to incorporate many of the elements of an open-source approach into its standard development activities. UBH's behavioral health sciences department publishes research for peer review, makes source code available through these publications, and shares data with academic partners (5,6). For example, UBH has described statistical issues it has encountered in the development of a provider profile and has made this discussion and other technical reports available on its Web site (7). Thus others inside and outside the managed behavioral health system may adopt, modify, and extend UBH-initiated technical developments in significant ways that benefit UBH and other managed behavioral health organizations.
Despite these substantial beginnings, an open-source approach in managed behavioral health requires the adoption of other key elements. First, more open lines of communication need to be established so that anyone, regardless of geographic region, company, or organization, can get involved in the improvement process. Second, issues related to controlling risk must be addressed. Third, and most important, a large number of organizations and individuals are needed to establish the critical mass that is necessary for open-source development to be successful. Finally, managed behavioral health organizations need to create work environments in which technologists' natural drive to collaborate is recognized and encouraged.
Dr. Cuffel is vice-president of research and evaluation and Dr. McCarter is senior research analyst at United Behavioral Health, 425 Market Street, 27th Floor, San Francisco, California 94105 (e-mail, firstname.lastname@example.org).