Integrated ACE model for consumer health information needs: A content analysis of questions in Yahoo! Answers

Authors


Abstract

Consumer Health Information (CHI) seeking on the Internet is becoming more popular, yet very complex as Social Q&A services emerge. Health information needs occur at various stages of health and illness. In response to the need to ensure that information provision is effective from prevention and treatment, this study attempted to uncover how health information seekers articulate their information needs. In this study, 440 health-related questions in Yahoo! Answers were first pre-identified into eight progressive stages of health information seeking. Each question was treated as an information seeking activity, and content of the question was considered as health information needs. Through analyzing questions in action, cognition, and emotion dimensions inspired by Kuhlthau's Information Seeking Process model, each question was further coded to reflect the nature of clarity, urgency, and length of information needs articulation. The findings uncover the fact that CHI seeking is multi-tasking and non-liner, and the way information seekers articulate the health information needs changes over the stages of health status. This study may inform consumer health information providers in designing services that may address the time-sensitive nature of CHI seeking.

BACKGROUND

Consumer Health Information (CHI) seeking on the Internet is becoming more popular, yet very complex as Social Q&A services emerge. A PEW study reported that 80% of Internet users have looked online for information about health topics (Fox, 2011). Among those who use social networking sites, 23% have followed their friends' personal health experiences or updates on the site. It highlights the importance of consumer health information needs issues. However, general consumers, often without medical training, tend to have difficulties in formulating their requests in few terms when finding health-related information online. According to Zhang (2010), the difficulties are at three levels. First, at the mental model level, everyday information seekers tend to use simpler or specific vocabulary to describe the disease or condition (Zeng, Kogan, Ash, Greenes, & Boxwala, 2002). Second, at the semantic level, the vocabulary often does not match the medical terminologies, for example, Unified Medical Language System (UMLS), or vocabularies used in health information websites (Keselman, et al., 2008). Sometimes everyday information seekers can't even find proper terms to describe their actual intentions (Zeng & Tse, 2005). Third, at the lexical level, misspellings and use of abbreviations are common in health information searching and often cause search failures (McCray & Tse, 2003; Smith, Stavri, & Chapman, 2002).

Past research confirm that health information retrieval through search engine often ends up with difficulties in locating the desired information, and the information may be too complicated to determine the information quality. When people can't assess the quality of information alone, they turn to second-hand knowledge (Wilson, 1983). There is no exception in consumer health information communication. Unlike clinic visits in which anxiety plays a role, Social Q&A services allow people to ask questions in their own language and receive answers from others. Even if the answers provided by others may require further confirmation and verification, the service is still popular on the Web. In this environment, CHI seekers ask instead of search for answers.

Previous studies on CHI needs are based on two approaches (Zhang & Fu, 2011). The first approach is from a clinical perspective that uses structured surveys to identify types of health-related information; the second approach is to analyze transaction logs to identify what people search for. Both approaches can provide insight into the types of health information needed, but they are not able to reveal CHI needs in-depth. In Social Q&A environment, general consumers tend to ask questions to gain first-hand information, or to submit the requests that may not be dealt effectively by search engines, or to get experiences provided by others with similar problems (Kim, 2010). Krikelas (1983, p. 13) suggests that “the characteristic of the problem may be a more critical indicator of potential behavior than various personal or work characteristics”. In other words, analyzing CHI questions may allow us better understand the characteristics of health information needs in Social Q&A environment.

Kuhlthau's Information Search Process (ISP) model (Kuhlthau, 2004) reveals a constructivist and holistic view of information seeking from the user's perspective in six stages: (1) Initiation; (2) Selection; (3) Exploration; (4) Formulation; (5) Collection; and (6) Presentation. The model incorporates three realms of experience: affective, cognitive and physical dimensions. In order to construct a complete view of consumer health information needs, this study not only explores the action taken by the consumers, but also explores their cognitive activities and emotional experiences. According to Wilson (1981), the interaction between user and information system must be guided by emotional needs and cognitive needs. Consumers asking questions in Social Q&A sites is also one example of dynamic information seeking behavior.

With the ISP model as the theoretical lens that suggests information seeking process is dynamic and proactive, this study was to content analyze 440 health-related questions collected from Yahoo! Answers. Health information needs occur at various stages of health and illness (Buckland, 1994). This study recognizes that CHI needs is time-sensitive and intends to analyze the content of questions of eight different stages in action, cognition, and emotion dimensions. It further analyzes questions' clarity, urgency and length, in order to improve our understanding about the way health information seekers express their information needs and explore the differences of health information needs in different stages. Specific research questions are: (1) How are CHI needs expressed in eight stages within the affective, cognitive, and physical ISP framework? (2) How do these expressions differ in various information seeking stages in terms of clarity, urgency and length? The findings will inform library and information service providers in designing services that may address the time-sensitive nature of CHI seeking and to effectively educate CHI consumers in health information literacy.

LITERATURE REVIEW

Consumer Health Information

Consumer Health Information (CHI) refers to information about healthy living, medical conditions, and health and other services that is provided directly to consumers (Elliott & Polkinhorn, 1994). The Consumer and Patient Health Information Section of Medical Library Association (1996) defines CHI as “information on health and medical topics provided in response to requests from the general public, including patients and their families.” The organization further indicates that in addition to information on the symptoms, diagnosis, and treatment of disease, it also encompasses information on health promotion, preventive medicine, the determinants of health, and accessing the health care system. Deering & Harris (1996) pointed out that CHI is defined as “any information that enables individuals to understand their health and make health-related decisions for themselves or their families” (Patrick & Koss, 1995). Sangl & Wolf (1996) reported a variety of health information types for health care decisions-making, including: (1) information of health promotion, and preventive health behaviors; (2) information of treatments or services for specific diseases or chronic illnesses; (3) information related to health care providers and facilities; (4) health insurance related information.

Internet access to health information is recently being widely explored and utilized by health care service providers. The popularity of Internet and mobile access enables people with the ability of sharing what they are doing or thinking (Fox, 2011). The chance that people meet those who have experienced, or are in the similar situation is highly probable (Oh, 2011). Besides, health experts may not be readily available when people need to discuss their concerns. Social Q&A services allow people to ask questions in their own language and receive answers from others. It is not only for people to seek and share information, but also for people to exchange experiences and feelings and receive social support (Gooden & Winefield, 2007).

Kuhlthau's Information Search Process

Kuhlthau's Information Search Process (ISP) model is a six stage model of the users' holistic experience in the process of information seeking. ISP incorporates the attributes of feelings, thoughts and actions for the individual information searcher in each stage (Pettigrew, Fidel, & Bruce, 2001). From initiation, selection, exploration, formulation, collection to presentation, each stage integrates user's affective, cognitive and physical experience (Kuhlthau, 2004).

“Information seeking” is an active process of constructing. It involves fitting information into what one already knows, and extending this knowledge to create new point of views. In the ISP model, physical aspects illustrate the process from which people seek relevant information (exploring) to seek pertinent information (documenting). Cognitive aspects are described as a constructive process, the process in which an information seeker acknowledges a problem, identifies the information needs, and solves the problem (Kuhlthau, 1991). Affective aspects of ISP describe users' feelings from uncertainty, confusion/frustration/doubt, disappointment, to optimism, clarity, sense of direction/confidence, to satisfaction (Kuhlthau, 2004). Similarly, MacMullin & Taylor (1984) proposed that a model incorporates physical (actual actions taken), affective (feelings experienced), and cognitive (thoughts concerning process and content) realms of activity to represent the user's sense-making process in information seeking.

Past research has helped to bridge consumer health information and ISP, in terms of exploring the criteria consumers use to evaluate health information, and identifying quality indicator for online health information (Cline & Haynes, 2001; Eysenbach, Powell, Kuss, & Sa, 2002; Yi, Stvilia & Mon, 2012). In addition, perceptions and experiences of online health information were explored to enhance health information system design (Gray, Klein, Noyce, Sesselberg, & Cantrill, 2005; Liu, Narasimhan, Vasudevan, & Agichtein, 2009; Zhang & Fu, 2011) According to Zhang (2010), exploring the context of consumer health information seeking can increase the understanding of the link between information needs and information seeking patterns. Little has been known about consumer health information needs of physical, emotional, and cognitional aspects in Social Q&A.

STUDY DESIGN AND METHOD

A total of 440 health-related questions solicited from Yahoo! Answers were first pre-identified and pre-categorized into eight progressive stages of CHI seeking inspired by Buckland (1994) and Zhang (2010): from when healthy; when asker thinks he/she might be ill; right before getting a medical test or checkup; when diagnosed or self-diagnosed as ill; before a treatment, surgery, or taking certain medications; when receiving or taking treatments, medications or exercise routines; after surgery or after treatment; to living with illness. Each question was seen as an information seeking activity, and content of the query was considered health information needs. Through analyzing questions in action, cognition, and emotion dimensions inspired by Kuhlthau's Information Seeking Process model, each question was further coded to reflect the nature of clarity, urgency, and length of information needs articulation. The proposed integrated ACE model of CHI needs in Social Q&A shown in Figure 1 illustrates that analyzing time-sensitive CHI needs in Social Q&A with perspectives in action, cognition and emotion may lead us to understand the user's sense-making process in seeking information that is central to lifelong wellness.

Data Collection

Yahoo! Answers is a major Social Q&A site. There are 15 categories in Health, including general health care, epidemic diseases, men's health, women's health, cancer prevention, mental health, emotional troubles, diet & fitness, pregnancy & parenting, sports injuries, alternative medicine and healing, drug and medication information, health awareness, Chinese medicine, and others, in Yahoo! Answers Taiwan. Among all 15 categories, questions for 4 categories were eliminated for their similarity in terms of content and nature of questioning and answering. These categories include emotional troubles, diet and fitness, pregnancy & parenting, and sports injuries. Data collection was performed in December, 2011. Forty questions were randomly selected for each category and a total of 440 resolved questions were collected.

Figure 1.

An integrated ACE model of CHI Needs in Social Q&A

Data Analysis

After 440 questions were downloaded from Yahoo! Answers to an Excel spreadsheet, each question was numbered, classified, reasoned as information needs, and analyzed with procedures of Grounded Theory, including open coding, axial coding, and selective coding (Glaser & Strauss, 1967). In this study, the ISP model is utilized to identify the characteristics of the questions in action, cognition, and emotion dimensions.

First, each question was classified into one health stage based on the context described in the question. Then open coding method was employed to code the questions. Codes were derived directly inductively by closely reviewing the action, cognition and emotion characteristics appeared in each question. From text-based analysis, information needs descriptions were categorized into lower-level themes and explicated for deeper meaning. Then the lower-level themes were consolidated into higher-level concepts. Through the process of iterative analysis and constant comparison, the ongoing discussions with experts were held to review the principal codes and ensure exclusiveness between categories, and to increase credibility.

The second part of analysis aimed to identify the questions for characteristics of clarity, urgency, and length. Clarity is defined as the extent to which speciality/ambiguity askers demonstrate through articulating the information needs. Urgency is defined as if askers use words to express urgency in their questions, such as “hurry,” “ASAP,” or “please respond by.” In addition, question length is also taken into consideration to analyze if questions of certain health category or health status would be in longer length. In some instances, frequency of each dimension's description was counted for additional times when a question addressed more than one need. Therefore, percentage shown in the Results was determined as the ratio of the total number of the observed description frequency of the same dimension in the same stage divided by the total description frequency of the same dimension in the eight stages. To ensure the reliability, in addition to the experts' advice on the principal codes, two independent coders were employed and the second coder coded 25% of the questions. After the second coder completed the coding of 110 questions, comparison was performed to identify discrepancy. In case of discrepancy, the two coders resolved the coding through discussion. If the first coder agreed with the second, the second coder's coding was used. The result of inter-coder reliability reached 82%.

RESULTS

Proposed typology and integrated ACE model of CHI needs in Social Q&A

Through constant comparison method, a typology of health-related information needs description (Table 1) was established based on the results of Grounded Theory analysis, and it was used as the coding framework for the quantitative content analysis. On action dimension, they are pre-identified into describing symptoms, asking question directly, expressing emotion, providing incentives to encourage immediate response, and describing problem situation. On cognition dimension, questions were pre-identified into seeking explanation, seeking treatment information, seeking the cause of illness, asking for a referral clinic, doctor or venue, inquiring the safety and efficacy of treatment, asking for experience sharing, inquiring information about medical regulation, seeking information on surgery and treatment effect, and seeking other information on treatment. On emotion dimension, these questions were pre-identified into confusion/puzzlement, frustration/fear, anger and dissatisfaction, self-blame, clarity, helplessness, and appreciation and gratitude.

Table 1. A typology of health-related information needs in Social Q&A
original image

Statistics of Each Stage Question

This study found the Stage 1 (when healthy) questions most common, accounted for 31%. Second most-common are Stage 2 questions (when asker thinks he/she might be ill), accounted for 23%. This pattern shows that health information seekers are not necessarily people who are ill; healthy or health-aware people are more responsible in seeking to upgrade their understanding of health issues. In other words, this study supports findings from previous research that distinct health motives may underlie health information seeking behavior as both wellness and illness information are sought (Weaver, et al., 2010). The least common are Stage 3 (right before getting a medical test or checkup) questions, only accounted for 2%. The percentages of each stage are shown in Figure 2.

Figure 2.

Percentages of Each Stage

Properties of CHI Needs in Action, Cognition and Emotion

Results of properties of CHI needs in action, cognition and emotion are shown in Figure 3, 4, and 5. Regardless of stages, we found that on action dimension, asking questions directly (59%) is the most common behavior observed from the questions.

Figure 3.

Percentages of Action Properties

On cognition dimension, over half of the questions are attempting to seek explanation (52%) and are much higher than seeking treatment information (24%).

Figure 4.

Percentages of Cognition Properties

The most common characteristic on emotion dimension is confusion/puzzlement, accounted for 49%. It is quite obvious that the emotion associated with CHI seeking is mostly negative. Appreciation and gratitude (10%) seem positive; however, they are actually used by the askers as a strategy to solicit immediate response.

Figure 5.

Percentages of Emotion Properties

Action Characteristics of 8 Stages

In Table 2, Table 3, and Table 4, the frequency was counted by coding the total of how many times a dimension was observed in each stage. Two types of percentage rates were provided. In each cell, percentage shown in the second row was calculated by the frequency divided by the total frequency of a dimension. Percentage shown in the third row was calculated by the frequency divided by the total frequency of a stage.

Table 2. Frequency of Descriptions on Action dimension
original image

Askers in Stage 2 tend to describe symptoms more frequently than others (36%). In situation like this, they think they might be ill, in order to get others' attention and ensure the usefulness of information received, they need to describe symptoms in detail. They also have the high ratio for asking questions directly in action dimension. Here is an example:

Recently, I smell my hands constantly. I wash my hands when I think they smelled. My hands also must be wiped off until the hand soap smell is gone. Do I have obsessive-compulsive disorder? (Stage 2/Mental Health)

Askers would like to take action for describing problem situation in query statement. 76 counts of situation descriptions were found in 56 questions. Related situations include describing the history of present illness (HPI) (28%), describing examination or treatment process (7%), describing habits and routines (15%), describing the mood and feelings (12%), and describing concerns or confusing matters (39%). The most used situation expression is describing concerns or confusing matters, followed by describing the history of present illness.

Cognition Characteristics of 8 Stages

Frequency of description on cognition dimension were shown in Table 3. The results show that in all stages but Stage 5 and Stage 8, askers seek treatment information more often than seek explanation.

Table 3. Frequency of Descriptions on Cognition dimension
original image

When people are chronically ill or becoming so, their priority is to seek treatment information (64%) to overcome illness, instead of explanation about their condition. For example:

I cough all the time, especially in cold weather. Is there any better treatment for my cough caused by sensitive tracheal? (Stage 8/General Health Care)

Askers in Stage 5 and Stage 6 also need more information on treatment rather than explanation. They need other treatment information that they have not yet known.

I have white spots on skin. Doctor said it can be treated by sunlight therapy and steroids. But I heard that taking steroids may not be good for health, and I don't want to risk it. Is there any other treatment for white spots? (Stage 5/Drug and Medication Information)

Askers at Stage 5 have more information needs on a referral clinic or doctor (77%), the safety and efficacy of treatment (78%) and experience sharing (75%). Before a treatment, surgery, or taking certain medications, it is typical that patients need enough information to be confident in diagnosis and treatment. They also need experience sharing from others. Social Q&A services offer a potential venue for them to get such information, in addition to the authoritative information from healthcare providers.

Emotion Characteristics of 8 Stages

Frequency of description on emotion dimension were shown in Table 4.

Table 4. Frequency of Descriptions on Emotion dimension
original image

The results show that no anger was found in Stage 3, 4, 5, and 8, but more anger statements were found in Stage 1 (36%), followed by Stage 6 (27%) and 7 (27%). For askers at Stage 1, they just express their dissatisfying emotion on Yahoo! Answers. Here is an example:

[…] I am only 62 kg, but 6 kg overweight. I don't understand why women care so much about their weight, and most men only care about women's appearance. It seems that it is difficult to create good impression on women just because you are over 60 kg. [.] (Stage 1/Others)

For questions in Stage 6, anger was shown with unsuccessful treatment or diagnostic results. Here is an example:

[.] I want to blame × Hospital for my mental disorder. It can't be cured and I have been kept in X hospital for a long time. Those doctors have no skill and knowledge at all. They just want cash. (Stage 6/Alternative Medicine and Healing)

Confusion/puzzlement is the most common emotional factor at all stages except Stage 8. Askers at Stage 8 feel more helplessness emotion (38%) than askers of other stages. When living with illness, the emotion becomes troubled or helpless, and they would not feel so confused compared with other askers. One asker stated his/her helplessness like this:

I have been taking a medicine for quite a while, but my situation still can't be improved. I really don't know what's the problem with me. I feel very uncomfortable every day, especially in this season. (Stage 8/Health Awareness)

Askers in Stage 1 and Stage 5 develop more appreciation and gratitude. The clarity in expressing their emotion is higher than askers of other stages. Askers in Stage 2 feel frustrated or lost (27%) more frequently than askers of other stages. It suggests when an asker thinks that he/she might be ill, they will turn in fear and hope to seek support and advice from others.

Common Characteristics

Among 440 questions, up to 97% of askers directly ask questions in their question statement, and almost half of the askers describe symptoms in the statement (47%). We also find that only 6% of CHI questions in Yahoo! Answers show urgency in the statement, and these questions are better in clarity and in longer length. It is worth noting that only 6% of the questions are seeking information related to the cause of the illness, and most of them belong to Stage 2. In stages of might be ill, before a treatment or surgery, or taking certain medications, or even when healthy, the percentages of askers' confused/puzzled emotion are higher. The expressions of these negative emotions are explicit, but they show no differences in length.

In cognitive dimension, when askers seek treatment, the most common actions taken are asking questions directly and describing symptoms. Of 110 questions for seeking treatment information, action of asking questions directly sums up to 96% and 66% for action of describing symptoms. It reveals the fact that askers describe the symptoms in order to facilitate health information communication on social networking sites.

Of 322 questions in which askers' demonstrate confusion/puzzlement, 302 (94%) questions show clarity. They feel the need to clearly express their information needs; however, the length of most questions (60%) is short, which means less than 50 characters in Chinese.

Seeking recommendation for clinics or doctors is not very common. Although only 7% askers have such information needs, we found that such questions are mostly found in the Stage 5. Other two properties of cognitive dimension: inquiring about the safety and efficacy of treatment and asking for experience sharing, are also shown centralized in Stage 5.

Emotional clarity refers to the ability to understand, label, and identify one's specific emotions (Salovey, Mayer, Goldman, Turvey, & Palfai, 1995). 75 questions are observed with emotional clarity, and all of them also show with clarity in content.

The results reveal different patterns of the CHI stages on types of CHI needs. Stage 1 is the most common stage in four categories (epidemic diseases, drug and medication information, health awareness, and Chinese medicine). It shows that even in healthy stage, consumers still have a lot of health information needs. Askers belong to the above 4 categories have more confusion/puzzlement emotion, accounted for 73% and have more seeking explanation in cognition and more asking questions in action, accounted for 69% and 98% separately. In mental health category, 63% of the questions are classified into Stage 2; they also demonstrate more confusion/puzzlement. Seeking explanation and seeking treatment, each accounted for 48% and 35%, are the most common cognitive properties. In addition to 95% askers take action of asking questions directly, there are up to 78% questions describing symptoms to seek judgment about whether they really have mental health disorders from peers.

CONCLUSION AND DISCUSSION

In Social Q&A environment, answerers may share their knowledge, experience, opinions and information with askers, and are highly motivated by certain reasons, including enjoyment, efficacy, learning, personal gain, Altruism, community interest, social engagement, empathy, reputation, and reciprocity (Oh, 2011). On the other hand, information seekers' questioning behavior is motivated by recognizing a knowledge gap, disturbing feelings, or lacking of social resources (Zhang, 2010). As a result, exploring query statements posted by information seekers may lead us to a deeper understanding of health information communication on Social Q&A sites.

In summary of the findings, action-taking is the most commonly expressed characteristic for health-related questions in Yahoo! Answers, followed by emotional expression. This finding shows the proactive nature of health information seeking. When taking the information seeking process in the timeline, the current study found that over half of the query statements can be identified as “before diagnosed as ill.” This pattern shows that health information seekers are not necessarily people who are ill; healthy or health-aware people are more responsible in seeking to upgrade their understanding of health issues. The diversity of the profile of health information seekers and consumers suggests a demand of designing a health information retrieval system that can fit their personal circumstances and health situation (Alpay, Verhoef, Xie, Te'eni, & Zwetsloot-Schonk, 2009).

The results explain the reason why Social Q&A services for CHI grow so popular. The technical features such as 24/7 operations and anonymous questioning and answering allow users to participate in an open forum, yet to still maintain their privacy. From the content aspect, unlike a clinic session where only few health/medical problems of a specialty may be solved, the one-stop-shopping concept for Social Q&A-based health information services supports the information needs across the diverse health and medical problems in any health/ill status at any point of time. From social aspect, the findings of the study reveal the complexity of emotions elicited from the query statement. It is a common practice that health information seekers not only seek information to improve decision making, but also look for emotional support and guidance during difficult periods in their lives.

However, the complexity of seeking health information on Social Q&A sites remains in several aspects. Large amount of questions in Yahoo! Answers uncover tremendous and diverse health information needs. However, the current information retrieval model of Yahoo! Answers relies on search by title keyword. After scanning through 440 query statements, it was found that titles of the questions were not self-explanatory enough to facilitate effective keyword search. Very often a mismatch between title and content was found. That is to say, the actual questions or concerns that an information seeker have in mind, don't coherently translate into titles with limited word length. Also, the content analysis of the questions uncovers a series of undergoing activities during the search formulation stage. For example, providing incentives to encourage immediate response or seeking treatment information. It suggests the multi-tasking nature of health information seeking and retrieval.

With the Action-Cognition-Emotion framework, this study was able to further the existing understanding of consumer health information. While past query-oriented CHI seeking studies primarily focus on behavioral aspect, this study extends the focus to emotion and cognition. The concepts of clarity, urgency, and length also aid in accounting for the integrated relationship between how information needs are addressed and the explicit linguistic characteristics. However, limitations exist. Not all Action-Cognition-Emotion factors were self-evident. Some tacit health information needs can't be found in questions. In addition, only a 25% percent sample was drawn to employ intercoder reliability. Besides, one critical factor missing from this framework is the concept of motivation. Oh (2011; 2012) studied motivation of answering health related questions on Social Q&A sites by conducting online survey. While Oh's study focuses on answerers' motivation and answering strategies, this study suggests that health information seekers' motivation is equally important in understanding the holistic process of health information communication. For example, question posters not necessarily the individuals of health information needs, in many cases, posters ask questions on others' behalves. It is interesting in the next step to differentiate various types of health information seeking motivations, and further validate how motivation will impact health information seeking in terms of action, cognition, and emotion.

Ancillary