Qualitative exploration of factors influencing the plan selection process by Medicare beneficiaries

BACKGROUND: Because Medicare plan coverage and costs change annually and older adults, the major beneficiaries of Medicare, are faced with multiple health conditions and changing medical needs, Medicare beneficiaries should evaluate their options during open enrollment every year. However, because of the complexity of plan selection, it may be challenging for Medicare beneficiaries to make an appropriate decision from among competing options. OBJECTIVES: To (a) identify factors that beneficiaries consider having influenced their plan selection decision and (b) describe the decision-making process according to the consumer decision-making model (CDM). The 2 research questions guiding this study included (a) factors Medicare beneficiaries considered having influenced their Medicare plan selection decision and (b) characteristics of decision-making processes employed by Medicare beneficiaries. METHODS: This is a phenomenological qualitative study. Semistructured in-person or telephone interviews with Alabama residents who have Medicare as the sole insurance provider were conducted between June and August 2019. Participant recruitment continued until reaching the saturation point. Each interview session consisted of structured questions identifying characteristics of participants and open-ended questions used to elicit participant Medicare plan decisionmaking process and factors affecting their decision. Data were analyzed using content analysis with a process of qualitative inductive coding. RESULTS: Twenty participants were interviewed. Twenty codes were identified and categorized into 5 themes regarding the factors influencing plan selection decisions by beneficiaries. When making a plan selection, participants were influenced by plan attributes (including cost, coverage, access to doctors, region, quality rating, and transportation); information resources and personal assistance; knowledge about Medicare; status and changes in personal situation; and experience with Medicare. Additionally, we identified 7 codes relating to beneficiary characteristics during decision-making processes, including being proactive, setting priorities, limiting choices, evaluating plans against personal needs, acquiescing to recommendations, sticking to the status quo, and weighing trade-offs. We consulted the CDM and created a conceptual model demonstrating a 5-step Medicare plan selection decision-making process and the factors influencing that process. CONCLUSIONS: This study created a step-by-step decision flowchart of Medicare plan selection to illustrate the complexity of the plan selection that Medicare beneficiaries must use. We uncovered the plan selection decision-making process among Medicare beneficiaries and factors affecting that process. Drawing from the CDM and the study findings, we developed a conceptual model. Findings will help researchers and community agencies target Medicare beneficiaries with different needs for assistance and design decision-making interventions/tools to help beneficiaries make rational decisions when selecting Medicare plans. These findings suggest that health care professionals should be involved in assistance programs to maximize efficiency of Medicare plan selection and to improve monitoring and consulting mechanisms to ensure the reliability of assistance information and services.

What is already known about this subject • According to multiple survey and administrative data studies, plan selection decisions by Medicare beneficiaries are not always the most economical, and these decisions might be influenced by costs; coverage of a plan; and beneficiary age, knowledge level, and health status.
• Existing qualitative studies focused on the selection of original Medicare Part D plans, while the selection of Medicare Advantage plans was less studied.

What this study adds
• This study created a step-by-step decision flowchart of Medicare plan selection to illustrate the complexity of making that selection.
• The study also created a conceptual model illustrating the Medicare decision-making process and factors that influence the process based on a consumer decision-making model and qualitative study findings.
In 2020, Medicare beneficiaries have access to between 24 and 32 stand-alone prescription drug plans (PDPs) and an average of 28 available Medicare Advantage (MA) plans, with 24 of these MA plans including prescription drug coverage (MA-PD). 1,2 Because plan coverage and costs change from year to year, it is critical for Medicare beneficiaries to review options and make informed decisions every year. 3 The majority of Medicare beneficiaries (85%) are adults aged 65 years or older. 4 Because older adults face multiple health conditions and increasing medical needs, it is important for them to evaluate which plan best meets their multifaceted medical needs every year. 5,6 However, these evaluations and decisions are complex, as shown in the Medicare decision flowchart (Figure 1). Key influential factors affecting the plan selection process have been reported in published studies using surveys and administrative data. For example, lower costs are associated with higher beneficiary satisfaction with health plans. 7,8 However, this preference is heterogeneous, as some older adults may prefer plans with higher quality of care over plans with lower cost. 9 Further, increasing beneficiary age, worse personal health status, and less knowledge about a health plan have been associated with reduced willingness to switch plans. [10][11][12][13] According to decision-making studies, the characteristics of decision makers and the decision task interact to influence the quality of decision making. 14 As people age, older adults may rely more on rules developed through previous similar experience to make decisions and less on analytic decision making. 15,16 In addition, complexity and difficulty of a decision-making process rise severely with the increasing number of available choices and smaller differences between options. 17,18 The possible changes in decision-making strategies and complexity of plan selection may lead to a more difficult decision-making process for some Medicare beneficiaries. 14,[19][20][21] Qualitative studies exploring influential factors of the Medicare plan decision-making process have either excluded MA plan beneficiaries from study populations 22 or explored among a mixed convenience sample who had original Medicare, MA plans, Medicare with private insurance, Medicaid, or Veterans Health Administration benefits. 23 Because MA plans have gained popularity, 24 MA plans provide more options that need to be compared, and beneficiaries can switch between the original Medicare and MA plans, 1 we did not limit our study scope to Medicare Part D plans but chose to also include MA plan beneficiaries in this study.
This study consulted a 5-step consumer decision-making model (CDM) as guidance to interpret themes emerging from the qualitative data to understand characteristics of the process Medicare beneficiaries employ while making a plan selection decision. The CDM has been commonly used to explain how an individual's rational consumption decision making for a product or service occurs through a sequence of 5 steps: (1) problem/need recognition, (2) information search, (3) alternative evaluation, (4) purchase decision, and (5) postpurchase evaluation. 17 The model posits that when an individual is aware of the difference between the actual state (e.g., current Medicare plan not covering the desired services) and the desired state (e.g., having a plan that covers the services), a problem/ need arises as the first step of a consumption decision process. This problem/need recognition step then leads to the information search step in which the individual searches for information about available alternatives and important attributes to consider from internal knowledge or external resources or assistance.
In the alternative evaluation step, the individual compares and evaluates the alternatives on key attributes RESULTS: Twenty participants were interviewed. Twenty codes were identified and categorized into 5 themes regarding the factors influencing plan selection decisions by beneficiaries. When making a plan selection, participants were influenced by plan attributes (including cost, coverage, access to doctors, region, quality rating, and transportation); information resources and personal assistance; knowledge about Medicare; status and changes in personal situation; and experience with Medicare. Additionally, we identified 7 codes relating to beneficiary characteristics during decision-making processes, including being proactive, setting priorities, limiting choices, evaluating plans against personal needs, acquiescing to recommendations, sticking to the status quo, and weighing trade-offs. We consulted the CDM and created a conceptual model demonstrating a 5-step Medicare plan selection decision-making process and the factors influencing that process.

CONCLUSIONS:
This study created a step-by-step decision flowchart of Medicare plan selection to illustrate the complexity of the plan selection that Medicare beneficiaries must use. We uncovered the plan selection decision-making process among Medicare beneficiaries and factors affecting that process. Drawing from the CDM and the study findings, we developed a conceptual model. Findings will help researchers and community agencies target Medicare beneficiaries with different needs for assistance and design decision-making interventions/tools to help beneficiaries make rational decisions when selecting Medicare plans. These findings suggest that health care professionals should be involved in assistance programs to maximize efficiency of Medicare plan selection and to improve monitoring and consulting mechanisms to ensure the reliability of assistance information and services. Quality control procedures were implemented throughout data collection. For example, to gain background knowledge, the 3 interviewers participated in community Medicare plan selection assistance program training sessions. Moreover, the interviewers discussed lessons learned after each interview to ensure adherence to the interview protocol, consistency among interviewers, and that pertinent questions were explored and phrased in manner to best elicit participant responses.

DATA ANALYSIS
Interviews were digitally audio-recorded, transcribed verbatim, and analyzed in ATLAS.ti version 8.4.2 qualitative data analysis software (ATLAS.ti Scientific Software Development GmbH, Berlin, Germany). For the first research question (i.e., factors influencing beneficiary plan selection decision), the interview transcripts were coded using an inductive process to identify themes using content analysis. 26 Three rounds of coding were conducted. First, 3 researchers individually used an open coding process to identify patterns within the data of 3 interviews to generate initial codes. Discrepancies in these codes among researchers were resolved via discourse and consensus. In the second round of coding, all transcripts were coded by 2 main coders, after which the researchers reevaluated the codes and identified additional codes. In the last round of coding, researchers consolidated several codes that were similar in nature but had low frequency ( f). Finalized codes were agreed upon by all researchers for a total of 20 codes, which were clustered into 5 broad themes.
For the second research question (i.e., beneficiary decision-making process characteristics according to the CDM), a mixed inductive and deductive approach was used. First, the same inductive content analysis procedure considered important according to the searched information (e.g., comparing attributes of more than 20 plans), which may result in a decision to purchase 1 of the alternatives evaluated (e.g., selecting a Medicare plan), which is referred to as the purchase decision step. The individual may continue evaluating the outcome of this decision as he or she uses the purchased product/service, and this postpurchase evaluation step loops the individual back to the first step, problem/need recognition, if dissatisfied.
However, as previously mentioned, not all beneficiary Medicare plan decisions are optimal, which means beneficiaries do not always thoroughly follow the 5-step process postulated by the CDM. 25 Many individual or environmental factors may interact with the complexity of Medicare options and plans, which can impact beneficiaries' ability or motivation to engage in 1 or more steps of the decisionmaking process, resulting in the decision-making process being less rational.
The purpose of this study was to identify such potential individual, environmental, or plan factors influencing the plan decision-making process by Medicare beneficiaries through a phenomenological investigation. This investigation was guided by 2 research questions:

STUDY DESIGN AND PARTICIPANTS
This phenomenological qualitative study was conducted using semistructured telephone or face-to-face interviews. Volunteering participants were eligible if they were Medicare beneficiaries, not covered by Medicaid or current/former employer-sponsored insurance in addition to Medicare, residents of Alabama, and able to read and write in the English language. Participants were recruited in May-July 2019 via a variety of methods, such as flyers, radio announcements, a study website, a press release by the authors' institution, and personal referrals by the Alabama State Health Insurance Assistance Program (SHIP) counselors. Written informed consent was obtained before conducting interviews. To improve recruitment rate and reduce interview "no-shows," a financial incentive in the form of a $25 electronic gift card was provided to each participant upon interview completion. All procedures were approved as an expedited review by the institutional review board of the authors' institution. involving 3 coding rounds was employed to identify and finalize the codes for participant decision-making characteristics that emerged from the data. This procedure resulted in 7 final codes under the single theme of decisionmaking process. Next, the 7 decision-making process codes were deductively reviewed against the 5 decision-making steps postulated by the CDM. Through discussion and negotiation, the researchers reached agreement on the primary CDM step that each decision-making characteristic code best represented, along with other CDM steps that were of secondary relevance to the code. In making the determination of secondary CDM steps for each code, the researchers considered which other CDM steps besides the primary step would either influence or be influenced by the characteristic addressed by the code.
For the 27 codes, Krippendorff's coefficient of 0.76 was obtained to establish the interrater reliability between the 2 main coders. 27 In addition, a final quality control review was completed by the third coder by reviewing randomly selected quotes to ensure that each quote was in congruence with the definition of the respective code.

Results
A total of 20 Medicare beneficiaries were interviewed, 19 via telephone and 1 in person during June-August 2019 ( Table 1). The average age of participants was 69.2 years (SD = 7.12). The majority of participants were White (80%), lived in urban areas (65%), and had monthly household income of $2,500 or less (55%). In terms of health care needs, most had good self-evaluated health status (75%), used 5 or fewer prescription medications (60%), and rated monthly medical expenses as a minor financial burden (80%).
We identified 20 codes, which we categorized into 5 themes addressing the first research question (i.e., factors influencing beneficiary plan selection decisions)-plan attributes, information sources, knowledge, personal situation, and experience (Table 2). We also used CDM as the guideline to interpret 7 codes related to the decision-making process theme, which answered the second research question (i.e., beneficiary characteristics during decision-making processes; Table 3). Using the findings obtained from the analyses, we also proposed a conceptual model displaying the Medicare decision-making process and influential factors affecting plan decisions. The themes and conceptual model are further discussed in the remaining sections.

INFLUENTIAL FACTORS IN MEDICARE DECISION MAKING
Plan Attributes. Plan attributes refer to characteristics of Medicare plans that influenced participant selection of their     setting priorities for what was important to participants. Most participants (85%) explained various characteristics of the plan and identified factors that were important. For example, 1 interviewee's priority was the ability to choose her own doctor, whereas another interviewee prioritized plans that covered her pricey medications. Finally, some interviewees (40%) expressed how they eliminated some Medicare program choices or plan options to narrow down their consideration set based on their priorities. For example, 1 interviewee said, "I usually try to look at what I think would pertain to me and go with that." Setting priorities, limiting choices based on the priorities, and evaluating alternatives against these personal priorities are characteristics of a rational decision maker in efforts to streamline the alternative evaluation step in a complex decision environment as in Medicare plan decisions. Finally, primarily related to the plan selection decision (PSD) step of the Medicare decision-making process, 3 (weighing trade-offs, sticking to the status quo, and acquiescence to recommendations) emerged. Despite many interviewees' showing potential rational decisionmaking characteristics relevant to the IS and APE steps as described above, only 20% indicated that their final PSD step involved weighing trade-offs among 2 or more key plan attributes to choose a plan among multiple options. The

DECISION-MAKING PROCESS
Seven associated codes constituted the decisionmaking process theme. These codes illustrate interviewees' self-reported cognitive or behavioral characteristics during their Medicare plan decision-making process. Table 3 presents the codes and how the CDM-based decisionmaking steps directly or indirectly related to each of the codes. One code (being proactive) was identified as primarily related to the information search (IS) step of the decision-making process. About half (55%) of the interviewees expressed some proactive actions they took in seeking information to help their decision making, which is a sign of a rational decision maker.
In addition, 3 codes (setting priorities, limiting choices, and evaluating plans against personal needs) emerged as primarily related to the alternative plan evaluation (APE) step of the decision-making process. The evaluation of a plan against personal needs was a key factor in terms of the code frequency ( f = 215), with all interviewees mentioning at least 1 quote that met this code definition. Plan evaluation happened when participants considered the fit between their health care needs and the plan characteristics. For example, 1 participant explained that he chose a plan that had a deductible because he was in good health. Next, the second most frequent code related to the APE step was "And that's what convinced me to take it. It was ... I mean they discovered it ... the coverage was so well, and they did it without no problems. So, I got a first-hand testimony, and that's what convinced me to go the way I went." (M3, 67, W) "I like the plan we have now, like I said the drug costs are not a great factor, because I've never had a problem with that, but the one that we have now seems to cover, you know, our needs, and meet our needs very well.  "Well, on this plan I got now, the biggest reason that I went to a plan that had a deductible was because I am in good health." (M3, 67, W) Acquiescence to recommendations PSD (PNR, IS, APE) 7 5 (25) "Well, you know, last year I become eligible for Medicare, so that's when I dropped < plan name > and went with < plan name >. And my mother, she has the same insurance. A friend of mine has got it, and they've both used it pretty extensively and they were well satisfied with it." (M3, 67, W) "... and then when it was said to me in the letter, they didn't give me much of a choice. They said, "This is the plan that we have.

CONCEPTUAL MODEL
Drawing from the CDM and the study findings, we developed a conceptual model illustrating a 5-step (PNR-IS-APE-PSD-PDE) Medicare plan selection decision-making process and the influential factors affecting the process ( Figure 2). 25,28 Similar to the CDM, after recognizing a need to select a plan (the PNR step), beneficiaries start searching for information/resources about Medicare plan options and attributes (the IS step). Given the information learned, alternative Medicare plans are evaluated and compared based on a set of decision preferences (the APE step), and finally a plan is selected (the PSD step). Depending on beneficiaries' level of involvement in each of these steps, they may follow a more or less rational decision-making route, resulting in staying in the current plan or switching to another plan, which may be more or less optimal. The utilization experience with a plan is evaluated (the PDE step), which may affect the beneficiaries' future plan selection decision. Further, the conceptual model incorporates the influential factors identified in this study, including plan attributes, information sources, individual factors of knowledge, compensatory decision-making process involving tradeoffs tends to be cognitively intensive but often believed to result in more "normatively accurate" decisions than do other heuristics-based decision-making processes. 40 However, most interviewees showed characteristics of bounded rationality, instead. A majority of the interviewees (65%) chose to stick to the status quo (keeping their existing plan) and 25% acquiesced to a recommendation by family/ friends or another source without a thorough alternative plan evaluation.
Some of the previously mentioned decision-making process codes may also be secondarily related to other CDM-based decision-making steps by influencing or being influenced by them. For example, although sticking to the status quo may be primarily a characteristic of the PSD step of the Medicare decision-making process, both the CDM and our data suggest that this characteristic may be influenced by beneficiaries' favorable assessment of their existing plan (the postdecision evaluation [PDE] step), which may prevent them from recognizing a problem or a need to switch plans (the problem/need recognition [PNR] step), thereby discouraging them from seeking additional information on other alternative plans (the IS step) or biasing their evaluation of alternative plans in favor of their existing plan (the APE step). This example illustrates

FIGURE 2
A would prevent interviewees from participating in the study. Instead, the status and changes in personal situations, such as anticipated health care utilization, residential location, and plan eligibility, affected participants' plan selection.
Additionally, plan selection assistance from others was critical among our participants. The Centers for Medicare & Medicaid Services disseminates plan information by interactive web page. Although delivering plan information via internet is efficient, it may have limited reach, especially to those living in rural areas and/or those who may not have access to the internet or may not own a computer. 35 For example, in Alabama, 55 of 67 counties are designated as rural areas and a large portion of the state is medically underserved. 36 The residents also have limited access to assistance from health care professionals, such as pharmacists, who are capable of helping with plan selection. 37 Therefore, it is not a surprise that these participants may reach out to SHIP, 38 insurance providers, or their friends and families for plan selection assistance.
We uncovered that the plan selection decision-making process among Medicare beneficiaries could be explained within the CDM's 5 steps of consumer decision-making process. 25 Information-processing capability and task environment interact to shape people's behaviors. 39 Individuals who are more capable of information processing "set priorities" and "weigh the trade-offs" when they undergo a rational decision-making process. 40 In contrast, the rationality of some participants in selecting a plan is limited to some extent by unknown future health status, an overload of information, or individuals' cognitive limitations; in such cases, beneficiaries may rely on schema-based mechanisms to make complex decisions, including generalizations from past experience, engaging in an unsystematic decisionmaking process, and spending less time thinking about all aspects of a given decision choice. 41 The difference between individuals' actions may also be due to different Medicare knowledge levels and health status, as people who are more knowledgeable and have better health status may be more involved in plan selection and switching. 11,13 Thus, it is desired that assistance/intervention programs are designed to take these varying factors and decision-making steps into consideration.
Furthermore, involvement of health care professionals in plan selection has been recommended by researchers, but has not been well implemented. 42,43 The majority (85%) of our participants selected plans that maintain their access to specific physicians, and more than half (55%) compared plans against their anticipated health care needs. Therefore, involving physicians in the plan selection process could help identify necessary health care needs and promote efficient plan selection. Though physicians and patients define "good personal situation, and experience, which are mapped around the 5-step decision-making process. These factors are likely to exert influences, individually or interactively, on various characteristics that beneficiaries manifest across the 5 steps of decision making. When making a plan selection decision, beneficiaries assess plan attributes such as cost, coverage, and physician access. Additionally, elements of decision making, such as the degree of the problem/ need recognized, the extent of information searched, the diversity of plans evaluated, and the decision strategy used by beneficiaries may be affected by individual predisposing factors including their knowledge of Medicare and its plans, experience with Medicare plans, and personal situation such as health status. Beneficiaries' information sources also influence their choices and the final plan selection.
The model provides essential elements for researchers to consider in examining beneficiary decision making and stimulate research in health care or decision-making areas. The model could also offer guidance for designing assistance interventions targeting beneficiaries at different stages through the decision-making process.

Discussion
Existing literature has defined optimal plan selection as selecting the plan with the lowest estimated out-of-pocket costs. [19][20][21]29 However, beneficiaries did not consistently select the most economical plan. 19,30 Similarly, in our study, although cost was mentioned by all study participants, many other plan attributes have critical impact on their plan selection decision. Therefore, we argue that an optimal plan is not always the most economical option. Rather, it is a plan that the participant is satisfied with after engaging in a thorough patient-centric decision-making process. To help Medicare beneficiaries make thorough patient-centric decisions, it is necessary for people who assist Medicare beneficiaries in plan selection to educate beneficiaries about the importance of evaluating Medicare plans and inform them of diverse plan attributes to consider when selecting Medicare plans. 13 Beneficiaries' individual characteristics, including Medicare knowledge, personal situation, and plan experience were cited as influential factors in existing literature. For example, some older adults are vulnerable to suboptimal plan selection due to possible changes in cognitive function and limited numeracy and health literacy skills, compounded by a lack of awareness regarding cost-saving Medicare subsidy programs. 20,21,[31][32][33][34] Although our study did not measure cognitive function, numerical literacy, or health literacy, the study team did not notice significantly declined cognitive function or limited health literacy that treatments" in different ways due to discordance in goals and prioritization of care, 44,45 shared decision making may increase communication to reach consensus in terms of preferred treatments and achieve better treatment outcomes. [45][46][47] Moreover, pharmacists could assist in multiple steps through the decision-making process. Because pharmacists frequently encounter patients who need financial support for their medications, 48,49 they are in a unique position to help patients recognize the need to switch plans in a timely fashion, navigate plan alternatives, and provide medication cost-saving strategies. 50 Finally, addressing accuracy of information provided by people who assist Medicare beneficiaries in plan selection is critical. Several participants of our study reported they had been contacted by "Medicare people," who introduced available plans and led the beneficiaries in deciding between limited options offered by 1 commercial company. None of our participants suspected the legitimacy of these phone calls or were aware of the need to check the legitimacy of these companies. Selecting a plan recommended in this manner may contribute to inappropriate plan selection. Though SHIP offices in each state provide free and unbiased assistance and counseling and Senior Medicare Patrol educates communities to promote preventative awareness of health care fraud, 51 these programs have limited reach. Future efforts should build a systematic monitoring, consulting, and feedback mechanism to provide reliable protection for Medicare beneficiaries.

LIMITATIONS
This study has some limitations to consider. It was conducted by 3 interviewers, which raises the possibility of inconsistency among interviewers. 52 Additionally, the study sample had a high proportion of highly educated