Development of a practice-based patient cohort for primary care research

Efforts to develop "laboratories" for primary care research have largely focused on the development of networks that recruit subjects on a project-specific basis. We sought to develop an alternative model--a representative cohort of adult primary care patients maintained for use in multiple projects.

In 2001, research assistants in waiting areas of a representative sample of 16 family medicine practices in North Carolina approached all adult patients during a 4-week period. Follow-up has been maintained for 3 years. In 2004 and 2005, the cohort was refreshed by adding eight new practices. Each consenting subject was administered a four-page self report questionnaire that included items on demographics, risk factors, health status, and quality of life.

Of 10,649 eligible patients approached in 2001, 6,811 (64%) completed the enrollment questionnaire, of whom 5,575 (81.9%) consented to be included in the cohort. African Americans, Latinos, and older persons were enrolled at rates paralleling the state's adult population. Poor general health, chronic illness, and risk factors for chronic disease were more prevalent in the cohort than in the general population. Over 3 years, cohort members were included in multiple studies, and 77% of the original cohort remained active. The per-subject enrollment cost varied between 27 US dollars and 45 US dollars; annual program maintenance costs were estimated at approximately 35,000 US dollars.

The research cohort has these advantages over traditional practice-based research networks: patient focus rather than physician focus, a structure that places few demands on practices, ability to target racial and ethnic minorities, and a better-defined patient population. As is true of all aspects of research infrastructure, the cost of development and maintenance is significant.