RESUMO
BACKGROUND: Little is known about the exact process of how patients search for medical information on the internet and what they retrieve. There is especially a paucity of literature on browsing for information on minor ailments, a term used for harmless diseases that are very common in the general population and thus have a significant impact on health care. OBJECTIVE: This vignette-based experimental study aimed to explore what kind of Web-based search strategies are applied and how search strategies, demographic characteristics, and the quality of the visited websites relate to finding the right diagnosis. Additional goals were to describe how searching on the Web influences one's perception of the severity of the potential diagnosis and whether or not the participants would discuss the information they found on the internet with their doctors. METHODS: Out of 1372 survey participants, 355 were randomly sampled, and 155 of them were recruited and assigned to one of four clinical scenarios. Each search term they used was classified as one of three search strategies: (1) hypothesis testing, (2) narrowing within the general hypothesis area, and (3) symptom exploration. The quality of the websites used was determined by using the DISCERN instrument. To compare the diagnostic accuracy of the participants before and after the internet search, a McNemar test was used. Chi-square tests were used to describe which factors are related to the chosen search strategy. A multivariate binary logistic regression model was constructed to predict which factors are related to finding a sound diagnosis after searching the internet for health information. RESULTS: Most participants (65.8%, 102/155) used the symptom exploration strategy. However, this depends on the assigned scenario (P<.001) and the self-estimated severity score of the symptoms before the internet search (P=.001). A significant relation was found between choosing an accurate diagnosis and age (odds ratio [OR] 0.94, 95% CI 0.90 to 0.98) and the clinical scenario, as well as the use of high-quality websites (OR 7.49, 95% CI 1.85 to 30.26). Browsing the internet did not lead to a statistically significant change in participants' beliefs about the severity of the condition (McNemar test, P=.85). Most participants (65%) shared their retrieved information with their physician and most of them (75%) received a positive response. CONCLUSIONS: Our findings suggest that most patients use a symptom-based approach; however, if patients expect the potential diagnosis to be severe, they tend to use a hypothesis verification strategy more often and are therefore prone to certain forms of bias. In addition, self-diagnosing accuracy is related to younger age, the symptom scenario, and the use of high-quality websites. We should find ways to guide patients toward search strategies and websites that may more likely lead to accurate decision making.
