Content analysis was conducted to study two hundred health-related transactions from My Info Quest, a collaborative text reference service participated by fourteen libraries and library systems nationwide. My Info Quest was launched on July 20, 2009, and as of July 2013, it offers service (Central time) from 8 am–10 pm Monday through Thursday, 8 am–6 pm Friday, 9 am–6 pm Saturday, and 2–4 pm Sunday. It uses Mosio Text a Librarian as the service software, which allows librarians to process text reference questions on a computer. Participating librarians take turns logging into the service software and responding to questions during the service period. All library users text their questions to the same number. They usually add a short code in the first message to identify their library; therefore when future messages arrive, the librarian on duty will know the home library of the user.

The unit of analysis in this study was a health-related reference transaction involving an independent question representing a distinct information need. A follow-up question asked to provide clarification was not considered an independent question. For example, a user asked an initial question about the new health care bill, “i was curious as to what are the details and terms of the new health care bill,” and then sent a follow-up question “does the new health care bill require all Americans to have health insurance and if not there will be a penalty,” to further specify his or her information need. In this transaction, the follow-up question does not represent a new information need; it functions as an attempt to better specify the original need. Therefore it is not considered an independent question.

A convenience sample of two hundred reference transactions was selected. The first one hundred health-related transactions were identified from both the 2011 and 2012 dataset of My Info Quest and included in the sample. Probability sampling was considered but decided against because of the following constraint: although questions and answers can be exported from My Info Quest’s account in Mosio Text a Librarian, each transaction is automatically defined as all message exchanges (related to one phone number) within fifteen minutes. This transaction definition is different from how the unit of analysis was defined in the study and the inconsistency could be problematic. For example, if a user texts a follow-up question fifteen minutes after the initial one, it will be considered as a new transaction by Mosio Text a Librarian. However, in this study, this follow-up question should be included along with the initial question in one single transaction. This difference posed an insurmountable challenge to identify an accessible study population where a random sample of health-related transactions could be selected.

The alternative was to adopt a nonprobability sampling design and select a convenience sample instead. The complete 2011 and 2012 datasets were carefully combed through from the first day of each year to identify health-related transactions. When the number reached one hundred in each yearly dataset, the examination stopped, yielding a sample of two hundred transactions. The transactions were analyzed to discover (1) what are the types of health-related questions asked in text reference service, (2) what is the length of health reference transactions, (3) how are health-related questions answered, and (4) what information sources are used to answer health-related questions.

The health-related questions in the two hundred transactions fell under nine categories, as shown in table 1. The frequency distribution of the question categories showed great divergence. There is no dominant question category, and the most frequently asked questions, those “seeking general factual information about human body, medical/health conditions, diseases, or medical concepts/jargons,” only represented one-fourth of the total number of questions.

The majority of the transactions (79 percent) contained only two messages, as indicated in figure 1—usually a question and an answer, indicating that health reference questions sent via texting are usually short, straightforward, and can be briefly answered.

In each transaction, the librarian’s answer was carefully examined, and they could be grouped into the following categories:

1. Library resources were used to answer the question (2.0 percent). For example,
   • User: “books on homemade tooth past or mouthwash”
   • Librarian: “SPL has a book called, ‘Natural beauty from the garden’ It has a mouthwash and toothpaste recipe. On the shelf at Franklin and Sylvan Oaks branches.”
2. Direct answer was provided, but without citation to sources (36.0 percent). For example,
   • User: “What causes snores?”
   • Librarian: “Snoring is caused by air vibrating against parts of the respiratory system (usually the uvula or soft palate). The vibration is due to a blocked passageway that can be caused by: throat weakness, mispositioned jaw, sleeping on one’s back, etc.”
3. Direct answer was provided, with citation to sources (34.0 percent). For example,
   • User: “How many calories in a taco time crisp bean burrito?”
   • Librarian: “427 calories, acc. to shapeit.com.”
4. Direct answer was provided without citation to sources but with suggestion to consult a medical/health professional (5.0 percent). For example,
   • User: “Does gum surrounding a tooth grow back after its lost?”
   • Librarian: “Unfortunately, they will not grow back naturally. You should consult a dental expert to be sure however depending on specifics.”
5. Direct answer was provided, not only with citation to sources, but also with suggestion to consult a medical/health professional (8.5 percent). For example,
   • User: “I’m taking 20mg of lexipro once a day. Will it hurt me if I have 3 glasses of beer?”
   • Librarian: “According to the drug interaction information through their website http://tinyurl.com/3s63tve, alcohol consumption is not recommended. We cannot make any medical recommendations; you should consult a pharmacist or physician. Thanks.”
6. No direct answer was provided (2.0 percent). For example,
   • User: “Is sobe life water with energy listed as a safe drink?”
   • Librarian: “I’m unable to find any reliable information discussing the safety of SoBe drinks.”
7. No direct answer was provided, instead sources were suggested for exploration or consulting a medical/health professional was recommended (10.5 percent). For example,
   • User: “What are symptoms of a kidney infection?”
   • Librarian: “There r many symptoms you can find then at web md http://www.webmd.com/a-to-z-guides/urinary-tract-infections-in-teens-and-adults-topic-overview.”
8. Incomplete transactions (2.0 percent), where librarians asked a follow-up question but users did not respond.

Overall, 53 percent of librarians’ answers contained citation to information sources. However, the way the sources were presented varied greatly. In some answers, the specific URL of the webpage about the disease/drug/condition in question was provided, either in full or shortened; following this URL, users would be able to directly view the relevant information that could help them with their questions. In some answers, only the name or the URL of a general medical information website was provided, such as “According to the Medline Plus database” or “According to www.everydayhealth.com.” In some others, there was merely indication that sources were consulted without any further specification, such as “According to several websites on piercings.”

Information sources used in answering users’ questions could be grouped into three categories, as shown in table 2.

Health-related text reference questions examined in this study cover a wide variety of topics. They are consistent with the health topics witnessed in other reference service venues.²⁷ It is worth noting that questions about healthy living, such as exercise, food, and nutrition were the second most frequently asked questions, showing an increasing awareness and pursuit of healthy lifestyle. Given library users’ growing interest in this topic, librarians may want to develop ideas to actively promote library resources and programs about healthy living. For example, automatic text or email alerts can be set up to notify users about relevant new books and events at the library. Libraries can also partner with health organizations to design relevant programs that seek to promote healthy living. This will ultimately help enhance the patient activation level in the nation. Patient activation refers to a person’s ability to manage their health and health care. Engaging or activating consumers has become a priority for employers, health plans, and policy makers. Libraries have a critical role to play in this movement as their resources, programs, and services can help more people gain knowledge, skills, and confidence to manage their health.²⁸

Using text reference service, library users can ask questions anonymously. At My Info Quest, users’ phone numbers are masked in all reference transactions, and a user ID number is displayed instead. Such anonymity makes it easier for users to approach librarians with sensitive questions. In this study, 6.5 percent of the questions were about sexual health, suggesting that text reference service could be a secure, convenient, and credible source for people, especially teenagers, to ask questions that they would be reluctant to ask in other reference service venues.

Close to 80 percent of the reference transactions in this study contained only two messages, one from the user and one from the librarian. This echoed the findings from Luo and Weak’s study about text reference questions in general.²⁹ They analyzed more than three thousand text reference transactions and about three-fourths of them consisted of two messages. This was because most of the questions were ready reference questions, which were usually short and straightforward and could be answered with a specific piece of information. In this study, half of the questions were seeking factual information, either about medical concepts or medical conditions/diseases in general or about the cause, effect, or symptoms of a particular medical condition/disease. Such questions could be answered briefly and directly, therefore would require merely two messages in a reference transaction.

This finding shows that texting is not a venue appropriate for a complex and lengthy reference interview. Library users, when texting a librarian, choose to use this venue for short and specific information needs. There are two contributing factors—first, each text message only allows 160 characters and this limits the amount information each message can convey; second, texting costs money if the user does not have an unlimited texting plan. When responding to text reference questions, librarians tend to be concise as well because of the character limit and cost concerns. As indicated in figure 2, text reference service usually operates on subscription-based gateway software (e.g. Mosio Text a Librarian and QuestionPoint), and the cost of such software is dependent upon the number of outgoing messages a library subscribes to. If a library can only afford a small number of messages, librarians will need to compose their responses succinctly.

In Luo’s study about the essential competencies requisite for providing text reference service, “ability to compose answers to patrons’ questions concisely, quickly, and accurately” and “ability to interpret patrons’ information needs with limited context in text messages” were identified to be the no. 1 and no. 5 most important competencies, indicating that librarians are fully aware of the affordance of texting as a reference venue and the need to be precise, brief, and fast in their answers.³⁰ To help librarians achieve these competencies, and thus effectively and efficiently answer health-related questions, adequate training is necessary. In addition, a cheat sheet readily available to librarians during their shifts could be helpful too. Such a cheat sheet may contain quick links to library policy and procedures about answering health related questions, frequently used health reference resources on the web (the sources identified in this study may offer some ideas in this regard), a list of texting lingo to help librarians “translate” text messages, and tools like a URL shortener or character counter that could help librarians compose their messages concisely.

In terms of answering questions, this study showed that librarians’ behavior varies greatly. This inconsistency is possibly caused by two factors: (1) librarians participating in My Info Quest are from different libraries, which might have different policies regarding answering health-related questions; (2) My Info Quest is not built upon any existing consortium; instead, it is a self-organized collaborative service and managed by volunteering librarians. The voluntary nature of this collaborative makes it awkward for anyone to police participating librarians. Even if policies and procedures are well established and communicated, some librarians may still exhibit a lack of awareness or a lack of willingness to comply.³¹ Although the inconsistent question-answering behavior is understandable, it is certainly harmful to library users’ experience and perception of the service. For instance, if a library user uses the service multiple times and sometimes his questions are answered with citation with sources and sometimes not, it can be confusing and disappointing to the user when he wants to explore the source and do further research. Thus it is important that libraries establish clear guidelines about answering health-related questions and provide sufficient training to help librarians adhere to the guidelines and therefore ensure consistent service quality.

The prelude to formulating question-answering guidelines is to fully understand how library users are asking health-related questions. The examination of the text reference transactions in this study suggests that health-related questions are usually stated in two manners—seeking factual information (e.g., “What are symptoms of a kidney infection”) and seeking advice (e.g., “I’m taking 20mg of lexipro once a day. Will it hurt me if I have 3 glasses of beer?”). To answer questions in the former category, it is necessary to provide a direct answer and a citation to the source. In this study, close to half of the questions were answered without citing the source. This might be a result of librarians’ concern of the restrictions of texting as a reference service venue. Yet providing sourced answers is an important mark of reference service and should not be compromised because of the limitations of the service venue. Since most of the sources used in answering text reference questions are web sources, URL shorteners like bit.ly can be helpful in saving characters and shortening the message.

As for questions in the latter category, the literature is abundant with discussions of the challenge of answering such questions. Library users often blur the line between health information and health advice and they do not understand that librarians’ role is limited to helping them with health information and not offering health advice.³² To address this challenge, it is necessary to educate library users about what librarians can and cannot do when assisting them. When answering their questions, librarians should suggest that users consult a medical professional for advice. In the meantime, librarians can still tease out the information needs from the advice-seeking questions and answer them with specific information and source citations.

In conclusion, through a detailed analysis of health reference transactions, this study provides an in-depth understanding of how library users ask health-related questions via texting and how librarians answer these questions. Such an understanding is critical to identifying the common issues in delivering health reference service via texting and to developing strategies to enhance service quality and improve user experience. With health literacy being a national health and public policy priority, it is important for libraries to play a more active role in assisting the public in their quests for health information. As texting-based health programs grow rapidly, more and more library users will become familiar with receiving, processing, and managing health information via this communication venue. Therefore libraries need to be well prepared to offer health information assistance via this venue as well. This study hopes to inform the practice of health reference service via texting and help libraries make evidence-based decisions on establishing service policies and procedures, providing training for librarians, and ultimately implementing the service successfully.