Interactions between Health Searchers and Search Engines

George Philipp, Ryen W. White

Abstract

This study investigates the bidirectional relationship between a user's internal state of health anxiety and their interactions with web search engines. By analyzing large-scale search logs, the research identifies how subtle variations in query formulation serve as proxies for medical history and how search engine result pages (SERPs) can inadvertently exacerbate health concerns. The findings suggest a reinforcement loop where anxious users are more likely to engage with "concerning" content, potentially leading to cyberchondria.

Introduction

Health anxiety—the tendency to interpret common symptoms as signs of serious illness—has found a new catalyst in the digital age: the search engine. This phenomenon, often termed cyberchondria, describes a "vicious cycle" where online health searches escalate rather than alleviate user concern. This paper addresses the lack of empirical understanding regarding how a user's psychological state influences query phrasing and how search algorithms respond to these nuances.

Methodology

The researchers conducted a longitudinal study using anonymized Bing search logs from approximately 190,000 users over a six-month period. The study focused on "headache" queries, a common symptom with diverse potential causes. The search process was modeled as a five-stage pipeline:

  1. BeforeSearching: Historical medical search behavior.
  2. QueryFormulation: The specific wording of a "landmark" query (the first symptom-only search).
  3. SERP Generation: The results provided by the engine.
  4. Click Behavior: The user's selection of results.
  5. AfterSearching: Subsequent search behavior or health-seeking intent (e.g., searching for a doctor).

A logistic regression classifier was employed to assign "concern scores" to web pages based on their discussion of serious vs. benign conditions.

Query Formulation and User State

A key finding is that subtle differences in query wording are highly discriminative of a user's prior medical concern. For instance, a user searching for "headache everyday" typically exhibits different historical search patterns than one searching for "daily headache." The researchers mapped these variations across 19 high-level and 117 low-level features, including intensity, duration, and associated symptoms.

Figure 1: High-level and low-level query formulation features used to categorize user symptom descriptions.

The Reinforcement Loop and Search Results

The study identifies a significant reinforcement loop. Users with a history of intense medical searching are 19% more likely to click on "concerning" results, even when benign explanations are available. Furthermore, search engines exhibit variance in results for positions 2-8; users who received more concerning SERPs had searched for 12% more serious conditions in the preceding 24 hours.

Figure 6: Statistical analysis of the relationships between SERP concern, click behavior, and past medical search history across different time windows.

Interestingly, the research uncovered a "HUI Paradox": viewing concerning pages sometimes led to a short-term reduction in real-world health-seeking queries (e.g., "find a doctor"). This suggests that for some users, the act of searching serves as a temporary, albeit potentially maladaptive, substitute for professional medical consultation.

Conclusion and Future Directions

The study highlights the sensitivity of search engines to user anxiety. While top results remain stable, the surrounding "noise" in SERPs can lead worried users down a path of escalating concern. Future search engine design should aim to detect health anxiety and provide balanced, unbiased information to prevent the escalation of unrealistic medical beliefs. Limitations include the focus on a single symptom (headache) and a single search engine, suggesting that broader cross-symptom and cross-platform studies are necessary to generalize these findings.

Summary updated

Abstract

The Web is an important resource for understanding and diagnosing medical conditions. Based on exposure to online content, people may develop undue health concerns, believing that common and benign symptoms are explained by serious illnesses. In this paper, we investigate potential strategies to mine queries and searcher histories for clues that could help search engines choose the most appropriate information to present in response to exploratory medical queries. To do this, we performed a longitudinal study of health search behavior using the logs of a popular search engine. We found that query variations which might appear innocuous (e.g. "bad headache" vs "severe headache") may hold valuable information about the searcher which could be used by search engines to improve performance. Furthermore, we investigated how medically concerned users respond differently to search engine result pages (SERPs) and find that their disposition for clicking on concerning pages is pronounced, potentially leading to a self-reinforcement of concern. Finally, we studied to which degree variations in the SERP impact future search and real-world health-seeking behavior and obtained some surprising results (e.g., viewing concerning pages may lead to a short-term reduction of real-world health seeking).
BibTeX
@inproceedings{philipp2014interactions,
  author = {George Philipp and Ryen W. White},
  title = {{Interactions between Health Searchers and Search Engines}},
  booktitle = {International ACM SIGIR Conference on Research and Development in Information Retrieval},
  year = {2014},
  doi = {10.1145/2600428.2609602},
}
Figure