DH Au, VA Puget Sound Health Care System; JR Curtis, University of Washington; MB McDonell, VA Puget Sound Health Care System; SM Anderson, VA Puget Sound Health Care System; E Udris, VA Puget Sound Health Care System; SD Fihn, VA Puget Sound Health Care System

Objectives: Many large medical systems, including the VA, are developing user-friendly, internet-based systems for physicians and patients. Patients now have the potential to find online information and communicate more freely with their physicians and health care programs. Before web-based systems can be used effectively for clinical and research applications, it is necessary to ascertain if patients would have access. As part of a large ongoing study of quality improvement, we sought to determine whether patients in the Veterans Affairs medical system used or had plans to use the internet.

Methods: Between October 1999 and March 2000, we conducted a cross-sectional, mailed survey of 32,138 patients from the 7 VA General Internal Medicine Clinics (GIMC) participating in the Ambulatory Care Quality Improvement Project (ACQUIP). Participating sites were geographically diverse and included: Seattle, WA; San Francisco and West Los Angeles, CA; Little Rock, AR; Birmingham AL; Richmond, Virginia; and White River Junction, VT.

ACQUIP patients were mailed a health screening questionnaire containing questions about common comorbid conditions and sociodemographic characteristics. Patients who returned the screening questionnaire were subsequently sent an additional item that asked, "Do you currently use the internet or plan to begin using it in the next year?" Subjects had the options of completing the following responses, "Yes, at home," "Yes, at the library," "Yes, at the community center," "Yes, at another location," or "No use or planned use." Prevalence and predictors of current or planned internet use in the next 12 months were evaluated. Chi-squared tests were used to assess differences between those who did and did not report use or anticipated use of the internet. Logistic regression models were constructed to estimate effect sizes and control for potential confounding associations.

Results: 32,138 subjects returned the health screening questionnaire and were sent the additional internet item. 16,353 subjects returned the Internet item. Of these, 71% (11,682) reported that they did not use nor have plans to use the internet in the next 12 months. Those who used the internet did so primarily from home with a minority going to community centers (0.8%), libraries (1.3%) or other locations (2.3%). Using logistic regression to assess independent predictors of plans for or current internet use demonstrated that income, geographic location, and education were the strongest independent predictors. Comparing those subjects with highest versus the lowest income, patients in the highest income stratum had a 15 fold increased odds of plans for or internet use. West coast patients had twice the odds of plans for or internet use compared to those receiving care elsewhere.

Conclusions: Most patients who receive care from VA GIMCs do not use the internet. Income, education age, and location are strong predictors of internet use.

Impact: Implementing internet based resources for patients may serve to increase the information divide between patients of different socioeconomic status. Possible solutions include establishing user-friendly computer centers within VA medical centers for patient use. Using the internet to perform population-based research could possibly lead to biased findings and should be used with caution.