Utilizing a Cognitive Engineering Approach to Conduct a Hierarchical Task Analysis to Understand Complex Older Adult Decision-Making During Over-the-Counter Medication Selection

Abstract

Background:

Adults aged 65+ (older adults) disproportionately consume 30% of over-the-counter (OTC) medications and are largely responsible for making OTC treatment decisions because providers lack awareness of their consumption. These treatment decisions are complex: older adults must navigate age-related body/cognitive changes, developed comorbidities, and complex medication regimens when selecting the right OTC. Yet little is known about how older adults make such decisions.

Objectives:

This study characterizes older adults’ cognitive decision-making process when seeking to self-medicate with OTCs from their community pharmacy, and demonstrates how hierarchical task analysis (HTA) can be used to evaluate a pharmacy intervention’s impact on their decision-making.

Methods:

A pre-/post-implementation approach, using a think-aloud interview process, was conducted with older adults within a community pharmacy setting as they completed a hypothetical scenario to treat either pain, sleep, or cough/cold/allergy symptoms. HTA developed a conceptualization of older adult decision-making regarding OTC selection and use before/after Senior Section implementation.

Results:

An HTA constructed from 12 purposefully-selected interviews (pre-n=9/post-n=3), consisting of 8 goals/15 sub-goals. While selecting an OTC, older adults considered quantity, cost, form, regimen, safety, strength, appropriateness of OTC safety, generic/name-brand, past experiences, and ingredients. The intervention reduced by half the number of factors considered.

Implications:

Older adult decision-making is more complex than just selecting OTC medication from a pharmacy shelf. HTA-informed decision profiles can provide pharmacists critical insights into safety issues that older adults may not be considering (e.g., factors related to safety, strength, or appropriateness of OTC for symptoms) so that pharmacists can support their decision-making.

INTRODUCTION

Over-the-counter (OTC) medications are nonprescription drug products that are considered safe and effective for use by the general public without healthcare professional oversight.^1,2 OTCs are widely available for purchase directly by consumers online and in any of the 750,000 mass-merchandise stores and 54,000 community pharmacies across the U.S.³ In 2019, OTC retail sales totaled 32.2 billion US dollars,⁴ with top sales in analgesics ($889M), sleeping aids ($57M), and upper respiratory medicines ($1,207M).⁵ Unfortunately, widespread access combined with unmonitored use can cause consumers to overestimate the perceived safety of these products, which may increase the likelihood of unsafe use, a mistake that is both prevalent and costly.⁶

Older adults (age 65+) account for 13% of the U.S. population, but consume 30% of OTC medications.⁷ What is unsettling about the prevalence of older adult OTC use is that certain OTC medications are associated with such geriatric outcomes as falls, worsened adherence, and increased frequency of adverse drug effects (ADEs).⁷ In fact, there is insurmountable evidence that older adults’ OTC use often results in harm. For example, older adult use of non-steroidal anti-inflammatory drugs to manage pain leads to 80,000 preventable ADEs each year,⁶ and acetaminophen-related toxicities account for up to 50% of all acute liver failures annually.⁸

Instead of informing their provider,⁹ older adults turn to the OTC aisles to manage sleeplessness, for which they take a variety of pain, allergy, and/or sleep products.¹⁰ Their behavior illustrates an unawareness about underlying medical conditions that can disrupt sleep, and a lack of knowledge about safe OTC options to treat such symptoms, compounding ADE risk.⁶ Cough, cold, and allergy medications (e.g., Dextromethorphan) can cause dizziness, lethargy, and nausea, all of which can increase fall risk. In addition, diphenhydramine can cause cognitive dysfunction, sleep disruptions, hepatic renal insufficiency and other anticholinergic side effects.¹¹ For these reasons, diphenhydramine and doxylamine are contraindicated for older adult use compared to other OTC treatment options, a fact supported by the Centers for Medicare and Medicaid Services.⁶

Clinical Resources for Older Adult OTC Use

It is critical for healthcare professionals to have valid, evidence-based, clinical practice resources that help to identify medications that may be unsafe for older adults. A widely used resource related to this topic area is the American Geriatric Society’s Beers Criteria for Potentially Inappropriate Medication Use in Older Adults (commonly referred to as the Beers List). The Beers List was initially developed using a modified Delphi process and has been reviewed and updated periodically, with the most recent version published in 2019.¹² The Beers List includes medications that pose elevated ADE risk for older adults (representing a drug-age interaction) and medications that older adults who have certain diseases should avoid (indicating a drug-disease interaction).

Despite the Beers List’s utility for preventing certain OTC-related harms in older adults, it remains an incomplete resource for identifying possible safety risks resulting from broader contexts of OTC use. For example, the increased nuance of clinical practice requires healthcare professionals to consider the ramifications of not only a patient’s age or diseases/conditions, but also the medications that they take to manage those conditions and the use regimen the patient is planning to follow. There also is a disconnect between the expectations created in clinical guidelines available to healthcare practitioners for safeguarding older adults’ OTC use, and practitioners’ ability to be aware of OTC misuse and appropriately intervening. Given these factors, it is largely the older adult’s responsibility to determine the safety of the OTC products that they seek to treat minor conditions. Yet, according to Albert and colleagues:

“we know surprisingly little about the ways older adults select OTC medications and decide when to start or stop use, how older people actually take these medications, or how involved clinicians and family members are in older adult OTC behavior.”^1(p909)

Indeed, older adults’ self-medication behaviors carry risks and have critical safety implications.

Older Adults’ Decision-Making for OTC Selection and Use

Older adults’ safe OTC medication use can be particularly difficult to achieve because of age-related physiologic complexity, comorbidities developed over time, and a more robust medication list and regimen they are expected to follow. Not only do these factors contribute to more complex decision-making for older adults, but cognitive impairment can also become an issue. Given that most older adults are unfamiliar with the need to determine if OTC medications interact with their other medications or how to appropriately dose an OTC medication, a lack of provider awareness about their patients’ OTC use may further permit the duplication of therapies and dangerous overdosing.^6,14 Such factors demonstrate a critical reason for examining and intervening in OTC medication selection and use behaviors.

An important aspect of selection behavior is understanding how older adults assess and make decisions about selecting an OTC medication. Decision-making is influenced by individual patient-level factors, features of the medication product, and the context in which decisions are made.^1,15 In the retail pharmacy, research shows that older adults tend to seek OTC information to make decisions by reading drug labels and package inserts, asking a pharmacist or their doctor for help, looking online, and gathering information from family/friends.¹⁶ However, as established earlier, older adults can misinterpret even formal information sources (e.g., drug label information) because of impaired cognitive functions (e.g., memory, concentration, comprehension).^17,18 Instead, older adults often rely on prior knowledge and experiences with the product, even though their positive experiences with medications as younger adults do not guarantee therapeutic safety with those same medications as they age.¹⁹ As a result, patient-centered, system-level interventions are needed to help older adults make safe medication choices.

As a result of a 2014 national summit on older adult OTC safety, Albert and colleagues described a need for research in five key areas of older adult OTC safety, including the perpetual and cognitive basis of OTC medication decision-making and behaviors affecting selection, dosing, and adherence decisions.¹ These experts recognized this complex topic could not be adequately addressed without interdisciplinary collaboration between human factors engineering, the social sciences, and healthcare providers.

Research has begun to investigate patient-level factors that inform decision-making that occurs in the pharmacy aisle. Study designs have utilized various qualitative data collection methods and taking one of three approaches in intervention development: (1) technological, (2) educational, or (3) system-level.

Technological interventions.

One avenue of research has focused on technological interventions to promote OTC medication safety, particularly for older adults. One study demonstrated that an interactive game improved OTC medication safety by educating older adults, even for people with low health literacy.²⁰ The game involved information regarding drug use, interpreting the drug facts label, resources to utilize when seeking advice, and medication list documentation. Another study developed a health IT intervention for older adults – an OTC consumer decision-making tool – through adapted participatory design and guided by user-centered research.²¹ This user-centered approach identified a variety of decision support mechanisms related to the sharing of knowledge and the usability of the technology. This study is unique because it was designed to investigate OTC medication decision-making at the point when decision-making occurs.²² Technological intervention studies, although incrementally beneficial in the environments for which they were particularly developed, are generally limited because they cannot systematically assess an older adult’s mental model while shopping for an OTC and are not generalizable to the “real world” experience.

Educational interventions.

Another avenue of research has focused on healthcare provider interventions to alter older adult decision-making in a way that allows for safer medication choices. A systematic review revealed a common conclusion among 49 articles that investigated OTC pain management and the pharmacist’s role: pharmacists are ideally placed to guide self-medication practices and recommending medical advice when needed and that all community pharmacists should be expected to handle counseling for OTC medications in addition to educating patients on their prescription medications.²³

System-level interventions.

To date, OTC decision-making research with older adults has focused on label design and information processing, which contrasts with the different perceptual, cognitive, and motor abilities of younger adults.¹ Importantly, de la Fuente and Bix focused on being able to identify features on drug packaging that should be redesigned to improve the perception and encoding of important safety information (e.g., warnings and indications).²⁴ As a consequence of this project, it was determined that task demand completion (e.g., identifying a product and following its instructions) relies on the interaction between the user, the task, and the package in the context of the community pharmacy environment. That is, a person must perceive a message as a “warning,” not only to comprehend the information but also to be able to encode the safety signal and take appropriate precautionary actions (e.g., not taking the medication).

Few studies have examined naturalistic decision-making in community pharmacy settings, so that system-level interventions can be developed to support older adults as they make tough medication choices. For example, one pilot study conducted by Stone and colleagues to identify decision-making factors associated with selection and use of OTC medications asked participants to complete a simulated OTC decision task. In this study, the laboratory environment was designed to remind the older adult of shopping for OTCs in the aisles of a community pharmacy. Results indicated that OTC selection was influenced by an older adult’s knowledge and beliefs about OTC safety, assessment of the severity of the ailment, medical constraints (e.g., known comorbidities and number of medications already taking), and cost and accessibility. In response to this dearth of information, this study was designed to evaluate an innovative pharmacy intervention for its impact on older adult decision making while selecting OTC medications.

The Senior Section™

This study’s intervention represents a systematic effort to decrease older adults’ selection/use of high-risk OTC medications. Participatory design²¹ and human factors engineering²² frameworks guided the redesign of community pharmacies’ typical structural layout of their OTC aisles (called the Senior Section), which promotes safe OTC use by facilitating pharmacy staff communication.²⁴ The Senior Section is shown in Figure 1 and is comprised of the following features:

Figure 1:

The Senior Section Intervention Implemented in The Shopko Project.

• a dedicated section of well-lit shelving stocked with a curated list of OTC medications that are safer for older adults,¹²

• proximity and sight line to the prescription department to promote patient/pharmacy staff interactions about OTC medications,

• tools (e.g., lighted magnifying glass),

• strategically-placed signage to aid in selection of lower-risk OTCs for the treatment categories of allergy, cough/cold, sleep, or pain,²⁴ and

• shelving height to facilitate use by older adults with visual or physical impairments.

The Senior Section is the first and only physical redesign intervention to demonstrate effectiveness in reducing OTC medication misuse in older adults without increasing pharmacist workload.^12,15 More detail about the design and development of the Senior Section can be found elsewhere.^6,25,39

Although the Senior Section was designed principally to reduce OTC misuse, the data collection methods used (see below) were considered appropriate to assess patient decision making when evaluating and selecting an OTC medication. This research aim aligns with recent discussions by experts in older adult OTC safety highlighting the need to research the perceptual and cognitive basis of OTC medication decision-making and behaviors affecting selection.¹ It is largely the older adult’s responsibility to safely select OTC medications to treat minor conditions, but little is known about this selection process.¹ One robust method for displaying the older adult mental model about the OTC medication selection process is through cognitive task analysis.

Using Cognitive Task Analysis as a Conceptual Framework to Understand Decision Making

Cognitive task analysis (CTA) refers to a family of methods used in qualitative research for a variety of topics (e.g., aviation and healthcare) to understand expert cognitive decision-making that must occur when completing a task in complex work environment (see Figure 2).²⁶ This process considers both the tacit (declarative) and unconscious (procedural) knowledge that informs these decisions. CTA is grounded in constructivism, an epistemological framework that resists the idea that knowledge can be completely formalized and classified because knowledge is constructed, rather than acquired, and is context- and individual-specific.²¹ Constructivist methods instead focus on how the learner interacts and processes information and uses that information to guide the development of an intervention or generate change.

Figure 2:

CTA implementation follows the following 5 steps.⁶³

CTA is usually applied to interview and observational data associated with action research, which focuses on iteratively-developed transformative change by taking action (e.g., an intervention) and doing research (e.g., assessing that intervention). The goal of action research is knowledge elicitation or analysis, to measure the impact of the intervention or to illustrate the role of the intervention in the decision-making process.³⁵ Such an approach is beneficial for capturing unobserved knowledge, cognitive processes, and goal structures that underlie human behavior.³⁵ This is particularly useful when applied to data collected from expert users. Expert users can be described here as older adults who have had continuous, deliberate practice in solving problems in a domain, such as deciding which OTCs to select from their community pharmacy, such that their knowledge is automated and largely nonconscious. This approach makes it possible to draw out the automated decision-making processes and action steps older adults use while problem solving related to OTCs. This enriched understanding of their decision-making may inform intervention designs. For example, elements of the environment that facilitate safe OTC decision-making can be replicated in the intervention design; and apparent gaps whereby even older adults fail to safely select OTCs can be addressed during intervention development. In the pharmacy setting, decision-making comprises how consumers (e.g., older adults) make decisions, how prior knowledge influences decision processes, and how they categorize OTC products. Given the dozens of OTC medications on the market to treat particular symptoms, all with different attributes of values (risks, warnings, dosages, or ingredients) that present tradeoffs,²⁸ it is critical to understand how older adults navigate these intricacies when choosing to use an OTC medication. By applying CTA to consumers in the pharmacy setting, a systems-based approach is used to ascertain the usability, usefulness, and understandability of the community pharmacy OTC aisles (work environment) and other system factors (technologies and tools, pharmacists) as the human operator (older adult) works to complete tasks.^19,26 CTA also has the advantage of deriving valid conclusions from interviews and/or observational data collected from a small sample size of 3–15 expert participants.^30–32 As a result, CTA’s rich data analysis approach generates output that is a particularly useful representation of the decision-making process, and does so in a way that can guide research and inform intervention development in complex healthcare environments.^30,31,33

Examples of CTA in Healthcare

CTA has been widely applied in healthcare and significant work has been done to guide researchers and healthcare providers towards realizing the importance of understanding cognitive task demands. CTA-based studies in healthcare began as early as 1995, and have addressed such diverse topics as necrotizing enterocolitis, diabetes self-management, influence of electronic health records on patient-doctor communications, end-of-life decisions, post-anesthesia patient care handoff, patient prioritization, functional endoscopic sinus surgery, and patient self-care decision-making. The methodology is not only beneficial as a tool for eliciting the underlying cognitive decision-making of healthcare providers, but its output provides a particularly useful representation of the decision-making process in a way that can guide future research and inform the development of interventions in complex healthcare environments. In fact, a meta-analysis of CTA studies³⁴ identified empirical assessments that suggest CTA contributes between 12% and 43% more information for documenting performance-relevant processes than approaches that are not CTA-based. Few studies have developed a CTA for patients,^37,49 while many have been conducted to improve training. A focus of other studies has been to assess the impact of an intervention or resource in the healthcare environment.^26,40

Hierarchical Task Analysis.

Within the family of CTA methods that can be employed to understand decision-making, hierarchical task analysis (HTA) commonly is used to develop task knowledge structures of expert systems and capture operation sequences (e.g., goals, timelines, interactions).³⁵ Based on a theory of performance,³⁶ HTA (like CTA) is used to analyze qualitative data (e.g., observations, open-ended survey questions, interviews with subject matter experts, walkthroughs). Using this method, high-level goals are decomposed into a hierarchy of sub-goals.³⁷ At each level of sub-goals, a plan, which explains how a goal should be accomplished, directs the sequence and possible variance of tasks as statements of the conditions necessary to achieve the goal.^38,39 HTA provides procedural knowledge that is more human-centered because it focuses on what the human operator will need to do when, how, and with what priority information. HTA also can unveil systematic errors that burden older adults when making decisions about selecting and using OTC medications.

Purpose & Study Objectives

For this study, CTA was used to understand older adult decision-making while they searched for and selected an OTC medication in a retail pharmacy setting, to manage or treat the symptoms of a minor illness not severe enough to see a doctor. The study objectives are to (1) characterize the cognitive decision-making process that older adults undergo when seeking to self-medicate with OTC medications in their familiar community pharmacy, and (2) demonstrate how this method of cognitive (hierarchical) task analysis can be used as a tool to evaluate the impact of a pharmacy intervention on their decision-making.

METHODS

Study Design

A pre-/post-implementation approach, using a think-aloud interview process, was conducted with older adults within a community pharmacy setting as they completed a hypothetical scenario to treat either pain, sleep, or cough/cold/allergy symptoms. An HTA-based coding scheme was applied to the interviews to conceptualize older adult decision-making regarding OTC selection and use before and after Senior Section implementation. These methods were designed to conceptualize an understanding about the impact of a pharmacy intervention on OTC selection decision-making. Prior to data collection, human subjects’ approval (2017–1354) was obtained from our university’s Institutional Review Boards (IRB) with appropriate data privacy and confidentiality protocols in place.

Setting

Three pharmacies in a Midwestern state were selected from within a single pharmacy chain. Each pharmacy was deliberately selected with the same physical layout and to represent communities with demographically similar patient populations, and were agreed upon by the study researchers and the pharmacy chain’s corporate management.⁴² Only three pharmacies were selected to pilot Senior Section implementation, with the plan to broaden its implementation in additional community pharmacy settings over the next few years.

Recruitment

In 2018–2019, mailings were sent to a random sample of 100 older adult patients from each pharmacy location (repeated up to three times). The mailing invited the older adult to participate in the study, included an information sheet and study flyer, and directed them to call a researcher if interested. Upon receiving a call, the researcher described the study and had the older adult attest to meeting inclusion criteria: aged 65+ years and purchased/considered purchasing an OTC medication 6 months prior to treat either pain, insomnia/sleep problems, cough/cold, or allergies. After agreeing to participate, the researcher scheduled an appointment to meet the older adult at the store and conduct the assessment. Participants received $20 for completing the study. This recruitment method resulted in a sample size accommodating a power analyses calculated for a separate research question.

Data Collection

Older adults participated in the study at the store location that was familiar to them, consistent with requirements of HTA.³⁶ A researcher greeted and consented participants when they arrived at the store, and then collected their demographics and self-reported medications and health conditions.⁶ Participants were equipped with an audio recording device and asked to choose a hypothetical symptom profile with which they were most familiar: pain, sleep, or cough/cold/allergy (see Table 1). These scenarios were developed to represent the typical scenarios that different types of older adult shoppers face when shopping for an OTC medication. Older adults were allowed to select whichever scenario was most relevant to them, because research shows that patients provide the most useful information about their decision making when they can refer to a concrete, memorable episode (e.g., treating a cold in the past few months).⁴⁹

Table 1:

Study scenarios for three categories of symptoms

Category	Scenario
Pain	You are having soreness and muscle aches. You have not taken any medication to help with these aches yet. And it’s not bad enough to call your doctor. So you’re here at the pharmacy to look for a medication that can help you feel better. Show me how you would go about choosing a medication to help you feel better.
Sleep	You are having some difficulty falling asleep or staying asleep. You have not taken any medication to help with this sleep problem yet. And it’s not bad enough to call your doctor. So you’re here at the pharmacy to look for a medication that can help you sleep better. Show me how you would go about choosing a medication to help you sleep better.
Cough / cold / allergy	You are having symptoms related to a cold or allergies, like a runny nose, stuffy nose, cough, or congestion. You have not taken any medication for your symptoms yet. And it’s not bad enough to call your doctor. So you’re here at the pharmacy to look for a medication that can help you feel better. Show me how you would go about choosing a medication to help you feel better.

Interviews began at the front of the store (during pre-implementation), and then the older adult participant guided the researcher through the store to the pharmacy section where they were asked to think aloud while they shopped for an OTC medication to treat their chosen symptom profile. Conversely, at post-implementation, the researcher started each interview in front of the Senior Section, denoting the only change in data collection resulting from the Senior Section. The researcher observed the work and asked probing questions of the older adults while they interacted with the existing pharmacy system environment, which is a common task data collection technique for a think-aloud interview process.⁴⁴ The think-aloud session was completed once the participant selected an OTC medication in the aisle and indicated that this was the product that they would use to treat the chosen symptom profile.

For the duration of the interview, the researcher carried a clipboard with an attached video recording device to capture video and audio data of the older adult interacting with OTC medications and the pharmacy section (as back-up for the audio recording unit attached to the participant). Video recordings were used to collect data about the:

1. physical tasks completed during the decision-making process, which may align with the cognitive tasks elicited during the think-aloud interview,

2. certain patient characteristics that may be of interest, such as gait and mobility (e.g., limp, use of a walker) and vision (e.g., glasses), and

3. interaction between the participant and the products in their environment (e.g., with medication boxes, searching for correct one, pointing to things on the shelf, how quickly medication was selected).

The think-aloud process was identical for pre- and post-implementation, including the same probing questions guided by the SEIPS 2.0 model to target each component of the work system and how they influenced OTC selection and use, to elucidate factors important to older adults (see Supplementary Material 1).^6,21 Think-aloud interviews took approximately 20 minutes to complete, although interview duration was highly variable across participants (range: 5–50 minutes).

Transcripts of the simulated OTC medication selection were professionally transcribed. Video data were stored in a secure Box folder. All other collected data were stored in RedCap™ hosted at our university.

Data Analysis of Interview Transcripts

CTA follows basic qualitative coding data analysis methods. Data analysis was conducted in two stages. In Stage 1, the video data for each older adult were inductively analyzed to become familiar with the tasks performed to select an OTC medication. Non-verbal behaviors, interactions with OTCs in the aisle (e.g., touching products, reading product labels, searching for specific products), and mobility were noted. This inductive review substantiated that these data were appropriate for a CTA, so the data were subjected to a second review to develop and test initial CTA pilot codes. During codebook development, it was apparent that HTA was a suitable analytical method. As a result, HTA was used in Stage 2 of analysis.

In Stage 2, the codebook was further refined and finalized using the transcripts to identify independent HTA goals, with a second coder (KX) assisting with this process. The stop-rule that determines the level of task decomposition for our CTA was determined by the extent of detail provided in the think-aloud interviews. The final codebook, a listed representation of the HTA, is shown in Table 2. In total, 8 goals and 15 sub-goals were identified as codes for this process. Goals 1 and 2 were predefined due to the structure of the study, while all other goals were deduced from study data, primarily identified by the video data analysis and then confirmed using a sample of interview transcripts for validity and completeness of the HTA. Video data were also analyzed in Stage 2 if the coder considered the interview transcript to be ambiguous, to provide more context and further corroborate the HTA codes. For example, some physical tasks that were not verbalized in the transcript were later confirmed by reviewing the video recordings. Data analysis was continued until data saturation was reached.⁴⁴

Table 2:

Final Codebook with goals, sub-goals, plans (if applicable), term descriptions, and examples.

Final Codes
Goal	Sub-Goals	Plan	Task Description	Example Participant Quotes
0. Treat Symptoms with OTC		Plan 0: 1–2, then 3–4 in any order, then 5–6 in any order, then 7, then 8.	Highest-level goal that must be determined for the ensuing cognitive tasks to take place (not coded)	Overarching goal aligns with the scenario provided to them in the study (see Table 1).
1. Identify Category of Symptoms*			given the nature of this study, participants identify the category of symptoms by assessing which category of sleep, pain, cough, cold, or allergy, they most frequently experience / identify with the most	Interviewer: “So between sleep, pain, cough, colds, or allergy, what resonates more with you?” Participant: “Well, the allergies, asthma, that’s my problem” - Participant
2. Experience Symptoms*			given the nature of this study, “experience symptoms” is simulated by the interviewer, when they provide the scenario to the participant	The participant is given the symptoms they are experiencing in the scenario (see Table 1).
3. Locate OTC Symptom Category			represents the cognitive effort to determine where a symptom-specific OTC category might be located	“I’m looking for signs. Well, I would think it would be here with health and beauty. Well, I don’t see where their, even where their aspirin would be here, first aid. Okay. Well, we must be getting closer. Okay. Here’s vitamins, health, wellness, pain relief, cough, flu, allergy, and ear care. All right. Well, we must be in the right area here. Let’s see what they’ve got here.” - Participant
4. Identify First OTC			represents participant’s cognitive identification of the first OTC option that they may or may not select to manage their symptoms (can be verbal or physical)	“I’ve tried this [Unisom], and I take maybe one every four months or something, so I don’t use a lot.” - Participant
5. Assess		Plan 5: 5.1–5.10 in any order	participants assess the selected OTC using a variety of factors to determine whether an OTC should be selected and taken to treat their symptoms	-
	5.1 Assess Quantity		the number of dosage units in the package	“Well, I’m getting250 caplets for $17.99. I would probably look at it and get my calculator phone out and see which is the better deal. If I’m getting them here, I’m getting 100 caplets. They’re both 650 mg. I’m getting 250. But then I also look and say, okay, I’m only taking one a day, so maybe I’m better off going the 100 caplets that take me 100 days versus 250 days, because I don’t know on pain relievers if they lose their potency or not.” - Participant
	5.2 Assess Cost		the cost of the package
	5.3 Assess Form of Medication		liquid, tablet, caplet, soft-gel, dissolvable, etc.	“I want the dissolvable… I hate taking pills, and it works faster than the non-dissolvable, and that’s what I wanted was to use it to get to sleep” - Participant “I usually go for the generic brand that is comparable to it, and then I like the, just the capsules… I have
				difficulty, a lot of difficulty swallowing, and so the smaller, the easier to go down, the better.” - Participant
	5.4 Assess Safety		the perception of medication risk	“The only other thing about ibuprofen, it’s, it does have some effects on bleeding and blood pressure, sodium retention, and increases your risk for heart disease if you take it regularly, so I don’t take it a lot.” - Participant
	5.5 Assess Strength		the amount of active ingredient per dosage unit	“And I’m only going to take one, so I can do extra strength with one” - Participant
	5.6 Assess Regimen		dosage, frequency, pattern of use, etc.	“Well, if 250 days, and I’m only taking one a day, there’s some people that probably take maybe more than that a day. I’m only taking one a day, and as far as the regular arthritis, I’m not even usually taking one a day of the regular arthritis. The nighttime one, I am. The regular arthritis, I’m, like I said, maybe a couple a week.” - Participant
	5.7 Assess Past Experiences		the participant’s historical experience taking the selected medication	“Mentholatum, that reminds me of being a little kid and having rub on my chest, which I never was sure it worked. But it smelled good, so. I guess, no, probably I wouldn’t take any of these.” - Participant
				“But I know what to take from my past experiences” - Participant
	5.8 Assess Appropriateness of OTC for Symptoms		relating the purpose of the selected OTC to the symptoms experienced	“I take Sudafed for sinus and headache” - Participant “I tend to focus on this, you know, cold and flu (Nyquil) or there’s chest congestion. There’s a severe versus what looks like the mild for the Robitussin.” - Participant
	5.9 Assess Generic vs. Name Brand		a comparison of generic and name brand products	“Shopko has a pain relief which has the same acetaminophen that Tylenol has. So if this were, for example, on sale or if the Tylenol PM wasn’t available, I’d know it’s the same ingredient. It’s basically the same thing under different packaging” - Participant
	5.10 Assess Ingredients 5.10.1 Assess Active Ingredients 5.10.2 Assess Inactive Ingredients		assess ingredients to best treat symptoms experienced Active = active ingredients Inactive = dyes, other formulation ingredients, emoluments	“Okay. So I have no idea what cetirizine is. It’s an antihistamine. This one looks like it’d be pretty good for stopping, again, it’s got corn starch in it, which is okay.” - Participant
6. Consider Alternative		Plan 6: 6.1, then 6.2–6.3 in any order	considers an alternative OTC option	After a first product is identified and assessed, any additional product identified and assessed is considered to be an alternative.
	6.1 Identify Alternative OTC		cognitively identifies an alternative OTC option that may be used to treat symptoms
	6.2 Evaluate OTC	Plan 6.2: 5.1–5.10 in any order	assess the selected OTC using a variety of factors to determine whether an OTC should be selected and taken to treat their symptoms
	6.3 Evaluate relative to other picked OTC(s)	Plan 6.3: 5.1–5.10 in any order	assess the selected OTC in relation to another selected OTC using a variety of factors to determine whether an OTC should be selected and taken to treat their symptoms	“I think the Sudafed is more of a decongestion, as whereas, the Claritin is more of the antihistamine action, the antiallergy action. And I’ve already taken both with knowledge of my doctor and pharmacist. But I have fond I’m not taking the Sudafed particularly.” - Participant
7. Select Treatment			participant chooses a product	-
	7.1 Select Primary Treatment		participant chooses an OTC product(s)	“I would walk over here, and I would take Claritin. And I would probably take, in my case, a small one (Picks up a small box of Claritin)” - Participant
	7.2 Select Secondary Treatment  7.2.1 Decide that concurrent (OTC or non-OTC) product is needed	Plan 7.2: 3, then 4, then 5–6 in any order	participant chooses another product in relation to the primary treatment selected. Participant may determine secondary treatment is necessary.	“I would probably pick this one right here [Loratadine]… And if my nose is real stuffy, I will get these breathe strips” - Participant
8. Treat			Marks the end of the think-aloud interview.	“I would go with… the store brand, and I’d go with the 3 milligrams for $7.00.” - Participant

^*These tasks/goals are predefined due to the structure of the study. All other tasks/goals were deduced from study interviews.

Using the finalized codebook, a sample of transcripts were coded that were purposefully selected to reflect think-aloud processes interrupted only by interviewer probes for more detailed information (the reason for which is described in the Discussion). Two reviewers (AM & AG) independently analyzed and coded the transcripts to identify explicit statements related to independent HTA goals. The reviewers met in-person to discuss each transcript and to identify and resolve coding discrepancies.

RESULTS

The video data of all 87 older adult interviews were analyzed in Stage 1 and, for Stage 2, a sub-sample of 12 interview transcripts (9 pre-, 3 post-implementation) were analyzed.⁴⁹ Demographic characteristics for the participants selected for the HTA were largely similar (Table 3). The only notable difference was education, because the only participant who received an education only up to eighth grade was included in the sub-analysis. Regardless, most participants remained college or technical graduates.

Table 3:

Demographics of Total and Sub-sample Participants

	Total (87 participants) n (%)	HTA Sample (12 participants) n (%)
Health Status  Fair  Good  Very Good  Excellent	10 (11.5%) 28 (32.2%) 37 (42.5%) 11 (12.6%)	1 (8.33%) 4 (33%) 6 (50%) 1 (8.33%)
Gender (male)	30 (34%)	3 (33%)
Education  Up to Eighth Grade  High School or GED  Some college or technical school  College or technical school graduate	1 (1.15%) (19.5%) (20.69%) 51 (58.6%)	1 (8.33%) (8.33%) (16.67%) 8 (66.67%)
Race (white)	81 (93.10%)	11 (91.67%)
	Mean	Mean
No. of Prescribers	2.36	2.34
No. of Pharmacies	1.33	1.36
No. of Medications	9.68	9.42
Age (years)	73.7	73.8

Final HTA Configuration

In this HTA, the overall goal of the task is Task 0: Treat OTC Symptoms with OTC. Eight goals (Table 2, left column) were identified, which comprised the tasks necessary to reach the overall goal. Goals 1–4 did not consist of sub-tasks, while goals 5, 6, 7 had 10, 3, and 2, respectively. Goal descriptions are in Table 2. During the think-aloud process, participants verbalized what they were considering when making selection decisions (see examples in Table 2, right column).

An aggregated representation of the 12 participants included in the HTA are shown in Figure 3. The sequence of goals (i.e., goals 1–8) was determined by averaging the order in which each goal appeared in all 12 transcripts. For example, goal 1 was the first goal to appear across all transcripts.

Figure 3:

The Hierarchical Task Analysis representation of older adult decision-making in the selection of an over-the-counter medication.

Pre-Post Analysis

The average number of assessment factors considered pre-intervention was 6.33 factors. At post-implementation, participants considered only 3.14 factors on average. Table 4 shows the proportion of participants who considered each factor. During the pre-implementation phase, the most frequent factors considered were cost and quantity (78%), followed by regimen, past experiences, and generic vs. brand name (all 67%). Alternatively, at post-implementation, past experiences was by far the single most frequent factor (at 71%), with the next-most-frequent factor being appropriateness of OTC for symptoms (57%). Although medication quantity and cost showed the highest occurrence at pre-implementation, they were among the least deliberated at post-implementation (along with form and inactive ingredients, all at 14%).

Table 4:

The Extent that Participants that Considered Each Assessment Factor

Factor	PRE n = 9 n (%)	POST n = 3 n (%)
Quantity	7 (78)	1 (33)
Cost	7 (78)	1 (33)
Form	5 (56)	0 (0)
Safety	4 (44)	1 (33)
Strength	4 (44)	1 (33)
Regimen	6 (67)	2 (67)
Past Experiences	6 (67)	1 (33)
Appropriateness of OTC for Symptoms	5 (56)	2 (67)
Generic vs Brand Name	6 (67)	0 (0)
Ingredients	5 (56)	0 (0)
Active Ingredients	2 (22)	0 (0)
Inactive Ingredients	0 (0)	0 (0)

DISCUSSION

In today’s U.S. healthcare system, most decision-making responsibility falls to older adults when selecting an OTC to treat their symptoms. Because physician approval is not required to purchase OTCs, reducing harms depends heavily on the older adult’s ability to understand and use the medication appropriately.^1,28,45 There is insurmountable evidence that older adults are not using OTC medications safely. This unsafe use is exacerbated by inadequate knowledge about age-related physiological changes that may make certain OTC medications dangerous to take, as well as declining cognition and lapses in memory.^7,14 Considering such patient responsibility, this study responds to the prominent call for understanding the decision-making processes surrounding OTC behaviors among older adults¹ to improve OTC medication safety.

This HTA study, the first to investigate the holistic decision-making process surrounding OTC medication selection in the older adult’s natural shopping environment, demonstrated that older adult decision-making is far more complex than just picking OTC medication off a pharmacy shelf. In the aggregated view of the older adult OTC medication model that the HTA produced, system and individual-level factors that inform decision processes were identified and compared to prior profiles about shopping behavior. First, this HTA diagram, particularly the assessment factors regarding quantity (5.1), cost (5.2), form (5.3), strength (5.5), and generic versus brand name (5.9), aligns with the older adult decision-making processes described by Paliwal (2017):

• treatment decision-making – deciding to take an OTC by themselves or deciding to seek a recommendation from a healthcare provider or family/friend) and

• purchase decision-making – obtaining information from the OTC drug label or from a pharmacist, comparing prices and deals among various options (and between generic or brand name), and making a final selection decision, typically choosing the product that has maximum and fast relief, at a lower cost, and in an easy to swallow dosage form).⁴⁶

Second, these findings compliment other research indicating that the drug label and the community pharmacist are underutilized information sources.⁴⁷

Utility of HTA

For this study, an HTA was developed as a tool to evaluate the Senior Section’s impact pre- and post-implementation. As shown in Table 4, pre-post analysis of individual HTA sub-goals demonstrated that the Senior Section altered the weight of each sub-goal in their decision-making mental model. That is, factors that were most frequently considered during post-implementation were more aligned with medication safety concerns (e.g., appropriateness of OTC for symptoms), while factors considered important at pre-implementation related more to superficial aspects of the OTC product (e.g., cost, quantity). This data analysis approach, which is novel for the pharmacy setting, allowed us to identify nuanced differences in older adult decision-making as they shop for OTC medications in the Senior Section.

HTA, as well as other CTA methods, have been used in other disciplines as a best practice method to understand expert decision-making, but this is the first study to apply CTA methodology to the community pharmacy space.⁶ As such, this study contributes to a growing body of literature surrounding patient ergonomics.^48,49 HTA was particularly appropriate for this study because of the level of detail that can be interpreted, and its usefulness in the early stages of intervention development and iterative design.⁴⁴ This HTA can inform research to improve older adult OTC medication safety when shopping in a community pharmacy setting -- conceptualizing systems-focused interventions that alter the salience of critical decision-making sub-goals (e.g., safety-related assessment factors). Each sub-goal presents an opportunity (e.g., a safe decision) and a threat (e.g., an unsafe decision or error). There is the potential to improve safety by simply addressing a single sub-goal using a systems approach to the intervention development and implementation strategy. Safety-related sub-goals should be targeted, including the importance of medication ingredients, strength, safety, appropriateness, and regimen.

Reviewing the video data confirmed that a pharmacist can be involved at many points along an older adult’s decision-making process but may only be initiated when the older adult engages with the pharmacist. When pharmacists have the opportunity to engage with older adults about safe OTC selection, these HTA findings may help guide more patient-centric shared decision-making. These results can inform pharmacists’ considerations about what is important to the older adult’s mental model for selecting an OTC product. Such information can effectively supplement pharmacists’ reliance on the Beers Criteria¹² to help older adults avoid harmful medication use. Given that determinations about older adults’ safe OTC selection/use are often more complex than reflected by recommendations from a clinical guideline, HTA-informed decision profiles can provide pharmacists critical insights into safety issues that older adults may not be considering. For example, the pharmacist could make recommendations about such factors as safety, reduced strength, or appropriateness of OTC for symptoms, rather than having the older adult rely on such things as cost, form or quantity or not considering characteristics of the product that could result in potential harm.

Strengths and Limitations of Study

Several advantages support using HTA for this research objective. First, HTA elicits more detail than other task analysis methods, while maintaining clearly-defined boundaries for the start and stopping points of the decision process. This HTA approach was designed to clearly start at the time the older adult first experienced symptoms and stop when they decide which medication they intend to select to treat their symptoms. It should be noted that the symptom scenarios were always the same and included the same goal – to “treat the issue” – which provided structure for comparing the intervention effect between pre- and post-implementation participants throughout the decision-making process. However, it was a study design artifact. Second, HTA is relatively flexible in representing important goals that do not necessarily correspond to physical tasks at specific times, as it represents ongoing potential goals and sub-goals that could be triggered at any time (e.g., asking a pharmacist for help).⁵⁰ This HTA was an aggregate representation of the older adult shoppers in this study.⁴⁴ By doing this, we developed a more complete representation of decision-making processes that may take place while any older adult shops for an OTC medication. Third, HTA outputs make excellent tools for designing an intervention, or evaluating an intervention to improve its design.⁵¹ This diagram can serve as a tool for future work in this space, as the Senior Section continues to be implemented in pharmacies. Finally, a “good” CTA (and, by extension, HTA) is one that is valid, reliable, generalizable, appropriate in scope, complete, and useful.⁵² Table 5 demonstrates how the HTA developed for this study complied with these 6 criteria.

Table 5:

Six Factors in Cognitive Task Analysis Quality⁴⁵

Trait	Description	Evaluation of our HTA
Valid	• Construct validity: appropriate choice of methodology, reliance on domain practitioners as a source of information (either as the focus of analysis, or as members of the analysis team), observation of work practices and inspection of tools and artifacts in the work domain itself • Face validity: whether the relationship between the goals of the analysis and the analysis methods is apparent.	• Construct validity: demonstrated by using HTA as the appropriate methodological approach for the goals of this research, recruiting participants that were older adults themselves, and collecting data in their natural work environment, which was studied closely in pilot work while developing the Senior Section intervention.13,22 • Good face validity: the goals of this analysis (evaluating the Senior Section) and the analysis method (analyzing older adults as experts who need to use the Senior Section) align
Reliable	• Redundancy - did the analyst interview enough people, see enough tasks performed, watch activities under enough conditions that he or she started hearing and seeing the same things repeated, without novel information coming to light?	• This HTA is reliable: the analysis stopped only once data saturation was achieved and two coders participated in the analysis
Generalizable	• Did the analyst explicitly describe the systems aspects to which the analytic results generalize, and support that argument by making links from the participants and settings in which the analysis took place, to those of interest?	• This HTA is generalizable to older adults that live in this region and shop for OTC medications in a community pharmacy chain. Future work is needed to apply this HTA to other groups of older adults to ensure this output is generalizable to all older adults.
Scope	• Is scope of the analysis appropriate for the design goals?	• The scope is appropriate: the HTA covered the entire decision-making process from experience symptoms to selecting an OTC to treat the medication. We are able to generalize the scope of the HTA diagram to the entire shopping experience surrounding the Senior Section.
Complete	• It is usually impossible to guarantee completeness of coverage – not every variable, function, task, component, decision, etc. is going to be identified or addressed. • Consider the constraints and challenges inherent in the work domain, and the knowledge, skills, and strategies that expert practitioners bring to bear on those problems.	This HTA output is complete: variety of goals and subgoals have been identified, recognizing that some goals may have been missed because we did not recruit participants that consider alternative decision processes (e.g., demographically different participants than that of our study)
Useful	• Did the analysis result in information that is useful for the design of a new decision aid or training program? • Does this analysis support knowledge elicitation from and participation of subject matter experts in the analysis process?	This HTA is useful: used to evaluate the impact of the Senior Section intervention on older adult OTC medication decision making and can be applied prospectively to future iterations of the Senior Section and its implementation in pharmacy store chains.

A number of limitations also require consideration. First, each older adult selected one of three hypothetical symptom scenarios that were most applicable to them, to trigger appropriate decision processes for older adults’ mental models when selecting an OTC medication. In some instances, however, it became clear during the interview that another scenario was more relevant. When this occurred, the decision-making was captured for that alternative scenario only. Second, this analysis was conducted from data obtained for a large, multi-faceted, project, which yielded interviews with varying degrees of interviewer bias. Since HTA validity depends on natural decision-making processes that are free from bias, interview transcripts were purposefully selected that were most conducive to the HTA process; previous research demonstrates that the resulting post-implementation sample size is sufficiently powered for a CTA.^30–32 Third, an HTA approach is resource intensive, but generates output that has utility for future research on this topic. Finally, CTA methods do not always capture non-cognitive attributes for accomplishing results (e.g., physical capabilities, access to resources, relationships).⁵³ Although video recordings were reviewed to identify older adult physical limitations, older adult/pharmacist relationships were not observable and were coded only when explicitly stated.^51,54

CONCLUSION

HTA was shown to be a valuable method for classifying a variety of factors that are important for older adults’ decision-making about OTC medication selection. Findings from this approach can be used to help the pharmacy profession identify and understand older adult OTC health needs that can benefit from pharmacist support. In addition, the diagram output can inform future intervention design to improve OTC medication safety in the community pharmacy space. This study also used HTA to evaluate the Senior Section’s impact on older adult decision-making, which innovatively focuses its evaluation on the older adult as a patient and consumer shopping the community pharmacy OTC aisles. As a result of this process, older adults provided unique perspectives about the extent that a pharmacy-based intervention can alter their mental model towards safer OTC medication use.