Abstract
Purpose:
Engagement of patients in their care can lead to better health outcomes, especially for adolescent and young adult (AYA) cancer survivors who experience mental and physical illnesses more often than noncancer adults. We examined how patient engagement in care influences health care expenses and use.
Methods:
AYA cancer survivors (n = 1162) and a comparison group of matched adults with no history of cancer (n = 2954) were identified from the 2011 to 2016 Medical Expenditure Panel Survey (MEPS) data. Medical expenditures and health care utilization associated with shared decision-making (SDM) measured by a self-administered questionnaire adapted from the Consumer Assessment of Healthcare Providers and Systems Clinician and Group (CAHPS-CG) survey were evaluated using multivariable regression models.
Results:
AYA cancer survivors were more likely to report poor SDM compared with adults with no history of cancer (odds ratio = 1.31, 95% confidence interval [CI]): 1.06 to 1.62). AYA cancer survivors with poor SDM were more likely to report poor mental and physical health compared with AYAs with good SDM. AYA cancer survivors with poor SDM had $3037 (CI: $110 to $7032) in additional annual medical expenses and 4.86 (CI: 2.00 to 8.52) in additional office visits compared with AYA cancer survivors with optimal SDM, even after adjusting for chronic conditions and psychological distress.
Conclusion:
Our results highlight the substantial economic burden associated with poor SDM in AYA cancer survivors. Our research suggests that interventions to improve SDM in AYA cancer survivors may contribute to patients' positive perception of their health and result in AYAs seeking fewer medical services resulting in lower medical expenses.
Introduction
Communication can impact patient engagement and adherence to treatment care plans, especially for adolescent and young adult (AYA) cancer survivors who suffer from an increased risk of physical and psychosocial late effects of cancer treatment 1 and are more likely to use health care services than adults without a history of cancer. 2
More generally, physician–patient communication during visits is an essential medium for delivering appropriate information about diagnoses and in making decisions about treatment plans. 3 Survivorship care plans are usually complex and include long-term follow-up, requiring additional time from doctors to communicate and deliver care. 4 Studies on cancer patients found a significant communication gap between physicians and patients. Cancer patients desired to be more actively involved in their care while getting evidence-based recommendations from their doctor in a way they can understand. 5
Shared decision-making (SDM) involves educating patients on the available evidence for care options and weighing this against their preferences. 6 SDM is a part of patient-centered care that improves doctor–patient communication and ensures that the patient's values lead all health care decisions instead of the physician dictating patient care. The SDM model provides an alternative model to the paternalistic informed models previously used that focus on one direction of information exchange from doctor to patient.7,8 A key characteristic of SDM is a stepwise process where both patient and physician share the information and agree on the treatment plan.9,26
Engagement of patients in their care can lead to improved health outcomes for the patients and greater cost efficiency for the health care system. 10 Cancer survivors have reported better quality of care when they are involved in decision-making. 11 Effective doctor–patient communication can also ensure receiving timely care for cancer patients. 12 Cancer patients were more likely to be dissatisfied with care if they did not get sufficient information about their treatment. 13
Moreover, patients were confused about their care decisions when there is a mismatch between the provided and the needed information. 5 Successful SDM can reduce unnecessary visits and increase patient adherence to medications and the treatment plans. 14 Moreover, patients who have been involved in decision-making have been found to spend $1800 less on health care annually than those who have not been involved. 10
Despite the Affordable Care Act (ACA) encouraging the use of SDM in health care in title III, through implementing new quality of care programs and grants that endorse “Patient Decision Aid” an educational tool to help patients communicate their beliefs and treatment options to their health care providers, 15 many health care providers do not adhere to the SDM principles. 16
Although studies have assessed the importance of SDM for improving quality of care and lowering health care expenditures in adults with specific chronic conditions, including diabetes and atherosclerosis,17,18 no studies have examined these associations in AYA cancer survivors, a group with a high prevalence of chronic conditions, psychological distress, and health care utilization,19,20 or a comparison population of adults without a history of cancer. Therefore, our aim is to examine access to care, health status, perceived health, and health behaviors associated with SDM, as well as the association of SDM with health care utilization and medical expenditures in AYA cancer survivors and how these associations might be different for the comparison population of adults without a history of cancer.
Methods
Data source
We used the household component of the 2011–2016 Medical Expenditure Panel Survey (MEPS), 21 which collected data from a nationally representative sample of U.S. noninstitutionalized civilians. The final response rate ranged from 53.5% to 59.3%. The sample is taken from the panel of individuals who completed the National Health Interview Survey. The panel is surveyed for five rounds of in-person interviews and self-administered questionnaires over a period of 2 years. MEPS gathers comprehensive information on sociodemographic characteristics, health status, medical conditions, and health care expenditures. 22
Study population
We identified participants who reported that they were first diagnosed with cancer between 15 and 39 years of age, as done previously. 19 This may capture participants who were diagnosed with cancer before participating in MEPS. We considered all types of cancer, except nonmelanoma skin cancers, consistent with previous studies that have not classified people with nonmelanoma skin cancer as cancer survivors.20,23
We identified 2431 AYA cancer survivors. As done previously, 18 we excluded participants with no usual source of care, body mass index (BMI) <18.5, and those with person survey weight = 0, resulting in 1162 AYA cancer survivors. For the comparison group, we used propensity score matching methods to identify our population of adults with no history of cancer. There were 118,564 adults without a history of cancer who were eligible. We matched on age ±2 years, sex (male, female), and race/ethnicity (Hispanic of any race, non-Hispanic white, non-Hispanic black, non-Hispanic Asian, other).
We conducted unconditional logistic regression to obtain the propensity score, which is the predicted conditional probability of being an AYA cancer survivor based on the matched covariates. We used the propensity score to identify three matched adults with no history of cancer for each AYA cancer survivor using the greedy matching process in SAS 24 resulting in a matched sample of 2954 adults with no history of cancer.
Measures
Outcomes
Medical expenditures were calculated based on the reported health care use as dollars spent on ambulatory visits, inpatient visits, emergency visits, and prescription medications per year. Additional medical expenses reflected additional expenditures above those for AYA cancer survivors with an optimal shared decision. Health care utilization was measured as the number of ambulatory visits, inpatient visits, emergency visits, and any prescribed medications, including refills, per year.
SDM measures
SDM was measured using a self-administered questionnaire adapted from the Consumer Assessment of Healthcare Providers and Systems Clinician and Group (CAHPS-CG) survey, which is a widely used survey to assess patient experiences in health care. A total score out of 12 was given to each participant. The score was based on four questions about doctor–patient SDM. Questions include the following: (1) how often health care providers explained things in a way that was easy to understand? (2) How often providers showed respect for what you had to say? (3) How often providers spent enough time with you? (4) How often providers listened carefully to you? The responses were provided on a 4-point Likert scale: 1—never, 2—sometimes, 3—usually, and 4—always. We combined never and sometimes based on the method used by previous studies. 6
The final composite SDM scores ranged from 4 to 12, which were used to categorize an overall assessment of SDM, classified as “poor” (4–7 points), “average” (8–11 points), and “optimal” (12 points).
Covariates
Sociodemographic characteristics included age at the survey, sex, education, marital status, race/ethnicity, health insurance based on participant self-report, and family income based on the poverty statistics developed by the Current Population Survey 25 as a percentage of the applicable poverty line (based on family size and composition). Income was categorized as poor (<100%), near-poor (100% to <125%), middle income (125% to <200%), middle income (200% to <400%), and high income (≥400%).
Time since diagnosis was calculated based on the difference between age at survey and age of cancer diagnosis. Elevated BMI was defined as >25 kg/m2 and health behaviors included current smoking and not exercising regularly, defined as not meeting the guidelines of 150 minutes/week. Access to health care included health insurance type (private, public, or uninsured) and unable to get needed care (unable to receive or delay in receiving needed treatment).
Chronic conditions were identified by a series of survey questions asking the participants whether a physician or other health care professional ever told them that they had coronary heart disease, angina, myocardial infarction, other unspecified heart diseases, high blood pressure, high cholesterol, diabetes, asthma, chronic bronchitis, emphysema, stroke, or arthritis. We classified participants as having at least one or more of these chronic conditions versus none. Psychological distress was measured using Kessler (K6) item questionnaire. 26 It shows consistency when measuring distress across multiple sociodemographic populations and has been validated as a screening tool for clinically significant psychological distress.27,28
The participants were asked how often they felt so sad nothing could cheer them up, nervous, restless, fidgety, hopeless, that everything was an effort, or felt worthless in the past 30 days (all of the time, most of the time, some of the time, a little of the time, or none of the time). Scores were summed to generate a total symptom score, with scores ≥13 indicating clinically significant distress. Perceived mental health and physical health were measured by asking each person to rate his or her health (excellent, very good, good, fair, poor) and categorized as poor/fair and at least good (good, very good, and excellent).
Statistical analysis
Chi-square tests were used to compare the characteristics of AYAs with optimal versus average versus poor SDM. Significant differences observed across groups were followed by pairwise chi-square tests to determine differences within these groups. Then, we used multinomial logistic regression to estimate bivariate and multivariable associations between having poor and average SDM (versus optimal) with health behaviors and health care access.
Multivariable regression models with gamma distribution and log link were used to estimate medical expenditures. As done previously, 23 multivariable regression models with negative binomial distribution and log link were used to estimate office-based visits and prescription medication use, including refills. Two-part zero inflation-negative binomial models were used to estimate the number of inpatient visits and emergency room visits because a large percentage of AYAs did not use these services.
We matched on age ±2 years, sex (male, female), and race/ethnicity (Hispanic of any race, non-Hispanic white, non-Hispanic black, non-Hispanic Asian, other), and in addition all models were adjusted for chronic conditions, age at the survey, sex, race/ethnicity, education, marital status, income, insurance, exercise, BMI, and smoking status. All descriptive statistics and regression models were conducted separately for AYA cancer survivors and on a matched sample of adults without a history of cancer and were adjusted for inflation up to the year 2016 with the Personal Health Care Expenditure Price Index. Analyses were conducted using SAS statistical software (version 9.4). The study was exempt for UC Davis IRB.
Results
Characteristics of AYA cancer survivors and adults with no history of cancer
Table 1 shows that AYA cancer survivors who reported poor SDM were more likely to be uninsured or have public health insurance versus those with optimal SDM (8% and 47% vs. 4% and 34%; p < 0.001). In addition, those who reported poor versus optimal SDM were more likely to smoke (29% vs. 22%; p = 0.08), report not getting care when needed (1% vs. 4%; p < 0.001), have psychological distress (25% vs. 12%; p < 0.001), and perceive their mental (36% vs. 16%; p < 0.001) and physical health (51% vs. 30%; p < 0.001) as poor/fair. Similar results were found in the matched sample of adults with no history of cancer. Furthermore, in adults without a history of cancer, those with poor SDM were less likely to exercise regularly, more likely to be of Hispanic race/ethnicity, and more likely to have lower income compared with those with optimal SDM.
Characteristics of Adolescent and Young Adult Cancer Survivors and (1:3) Matched Sample of Adults with No History of Cancer by Shared Decision-Making, Medical Expenditure Panel Survey 2011–2016
AYA, adolescent and young adult; BMI, body mass index; SDM, shared decision-making.
SDM in AYA cancer survivors
Compared with matched adults without a history of cancer (11.44%), AYA cancer survivors were more likely to report poor SDM (14.11%, Fig. 1). In a logistic regression model comparing AYA cancer survivors with adults without a history of cancer, the odds of reporting poor SDM were 1.31 times higher in AYA cancer survivors after adjusting for age, sex, race/ethnicity, education, and income (odds ratio [OR] 1.31; 95% confidence interval [CI]: 1.06 to 1.62; data not shown in Tables).
Cancer status by SDM using a 1:3 matched sample, Medical Expenditure Panel Survey 2011–2016. SDM, shared decision-making.
Association between SDM and health status
In multivariable models, the odds of reporting poor SDM (vs. optimal) were higher in AYA cancer survivors with psychological distress than those without psychological distress (OR = 3.10, 95% CI: 1.44 to 6.72) (Table 2). Moreover, AYA cancer survivors who perceived their physical health as at least good were less likely to report poor SDM (OR = 0.52, 95% CI: 0.27 to 0.97). Among adults without a history of cancer, results were similar to cancer survivors, with higher odds of psychological distress and perceived physical health observed among those with poor SDM (Table 3). In addition, those who were unable to get care when needed were more likely to report poor SDM and those who perceived their mental health as at least good were less likely to report poor SDM.
Association of Average and Poor (vs. Optimal) Shared Decision-Making with Access to Health Care and Health Factors Among Adolescent and Young Adult Cancer Survivors: Medical Expenditure Panel Survey, 2011 to 2016
Indicates statistical significance of p ≤ 0.05.
CI, confidence interval; OR, odds ratio.
Association of Average and Poor (vs. Optimal) Shared Decision-Making with Access to Health Care and Health Factors Among (1:3) Matched Adults Without a History of Cancer: Medical Expenditure Panel Survey, 2011 to 2016
Indicates statistical significance of p ≤ 0.05.
SDM, medical expenditures, and health care utilization
AYA cancer survivors with optimal SDM had an average of $7896 (95% CI: $5707 to $10,925) in annual medical expenditures and 8.36 (95% CI: 6.60 to 10.58) annual office visits (Table 4). Additional health care expenditures of $3037 (95% CI: $110 to $7032) and office visits of 4.86 (95% CI: 2.00 to 8.52) were found in those who reported poor (vs. optimal) SDM. Adults without a history of cancer had an average of $3726 (95% CI: $2889 to $4807) in annual medical expenditures and 5.90 (95% CI: 4.88 to 7.13) annual office visits (Table 3).
Additional Medical Expenses, Health Care Utilization, and Medications for Shared Decision-Making in Adolescent and Young Adult Cancer Survivors: Medical Expenditure Panel Survey, 2011 to 2016
Indicates statistical significance of p ≤ 0.05.
Similar to AYA cancer survivors, higher medical expenditures and increased office visits were observed among those without a history of cancer who reported poor SDM, and we found that they have an additional $2242 (95% CI: $1036 to $3753) for poor SDM than those with optimal SDM and use more 2.38 (95% CI: 1.08 to 3.93) office visits (Table 5).
Additional Medical Expenses, Health Care Utilization, and Medications for Shared Decision-Making in Matched (1:3) Adults Without a History of Cancer: Medical Expenditure Panel Survey, 2011 to 2016
Indicates statistical significance of p ≤ 0.05.
Discussion
Our study used a nationally representative sample of AYA survivors provided by MEPS. We found that poor SDM was higher in AYA cancer survivors compared with adults without a history of cancer. AYA cancer survivors with psychological distress and those who perceive their physical health as poor/fair were more likely to report poor SDM, findings that were similar to our analyses of adults without a history of cancer. In addition, medical expenditures and office visits among AYA cancer survivors with poor SDM were substantial.
Specifically, AYA cancer survivors with poor SDM spent an additional $3000 and had five additional office visits per year compared with those with optimal SDM. Our findings are consistent with previous studies that identified good doctor–patient communication to reduce referrals and diagnostic tests, improve patients' perception about their health, 29 and lower the cost of care. 16
We found a higher prevalence of poor SDM in AYA cancer survivors compared with adults without a history of cancer, findings that are consistent with previous studies on cancer patients of all ages. 30 One explanation for this poorer SDM is that cancer survivors are more likely to have physical and psychological health problems that need more time, explanation, and attention compared with adults without a history of cancer. However, after adjusting for both chronic conditions and psychological distress, poor SDM remained higher in AYA cancer survivors, highlighting the need to improve SDM in this population.
We observed that poor SDM was associated with increased medical expenditures and office visits among AYA cancer survivors in our study. Furthermore, for those reporting poor SDM, AYA expenditures and health care office visits were higher compared with adults without a history of cancer.
These findings are supported by a previous study that found poor SDM associated with higher health care utilization and health care expenditures among patients who visited a primary care physician. 31
The Institute of Medicine has emphasized the importance of SDM in cancer patients as a part of patient-centered care due to the complexity of treatment plans, limited evidence for many treatment options, and trade-offs of risks and benefits that can greatly change patient preferences and values. 32 Moreover, AYA cancer patients face long-term care that includes multiple treatments and screenings that require a series of decisions. 4
The bidirectionality and the stepwise nature of the SDM model can help untangle the complexity of the treatment plans, make sure patients understand the risks and benefits of treatment options, eliminate confusion about information, 5 increase physical and psychological outcomes, 33 and improve patient satisfaction, 10 For example, patient satisfaction with radiation treatment was found to be highly correlated with perceived SDM among cancer patients. 34
We found that having psychological distress was highly associated with reporting poor SDM in AYA cancer survivors. This finding may relate to patient satisfaction, as studies have identified psychological factors, specifically anxiety, to be strong determinants of patient satisfaction in patients with atraumatic painful upper extremity 14 or among cancer patients. 30 This result is consistent with a systematic review of 86 studies examining 75 patient experience measures that suggested accounting for psychological distress when collecting patient-reported measures of care satisfaction. 35
In our study, those who perceived that they are in at least good physical health were less likely to report poor SDM. These findings are consistent with a study that found involvement in treatment decisions to be associated with higher social and physical functioning among women with breast cancer. 36 Moreover, effective doctor–patient communication was associated with better emotional regulation of patient, 37 perceived satisfaction, better recovery, 38 and reduction in patient anxiety and need for future visits or unnecessary tests. 29 However, to date, interventions among physicians and patients to increase SDM have not shown any promising improvement in the reported and measured use of SDM in health care settings. 39
Compared with adults without a history of cancer, AYA cancer survivors were more likely to have higher medical expenditures and more office visits even with optimal SDM. This may be explained by AYA cancer survivors being more likely to have serious and multiple chronic conditions 19 and/or more severe psychological distress 40 than adults without a history of cancer, which can increase medical expenses and health care use. 41 Long-term chronic conditions have been significantly correlated with higher expenses and office visits in AYA cancer survivors. 2 Because of high health care costs, a previous study found that AYA cancer survivors are more likely to forgo medical care compared with those with no history of cancer. 20
Some limitations to this study include that this study is cross-sectional, and so, we were not able to confirm the direction of causality between exposures and outcomes. Instead, future longitudinal studies are needed to elucidate the causal relationships. SDM and other measures in this study were based on self-report, which might be subject to recall and information bias and driven by one bad experience. However, self-reported measures have been validated as a measure of patient experience in the health care. 42 Even though SDM is frequently reported in the literature, there is no standard or unified measure across studies. The self-reported measure also might be affected by the participants' psychological status, which we attempted to minimize by considering psychological distress in our analysis.
The prevalence of chronic conditions was higher among AYA cancer survivors than adults without a history of cancer, which could impact SDM in each population. Although we included chronic conditions in the multivariable models, it is possible that we were not able to fully account for the impact of chronic conditions on SDM. Despite these limitations, our study used MEPS data, which are population-based and provide a representative sample of the population. It also provides measurable insight into the potential economic outcome of applying SDM in medical settings and included a comparison population of matched adults without a history of cancer.
Conclusion
The prevalence of poor SDM is higher in AYA cancer survivors than in adults without a history of cancer and is associated with higher medical expenditures and more office visits even after adjusting for chronic conditions and psychological distress. In addition, those with at least good mental and physical health were less likely to report poor SDM. These findings suggest that interventions to improve SDM among AYA cancer survivors may contribute to patients' positive perception of their mental and physical health and may result in AYAs seeking fewer medical services resulting in lower medical expenses. This study highlights the importance and the potential economic and health benefits of adapting optimal SDM while caring for AYA cancer survivors.
Data Availability
The data underlying this article were provided by the Agency of Health Research and Quality under license and by permission. Data will be shared on request to the corresponding author with the permission of the Agency of Health Research and Quality.
Novelty and Impact
We estimated the association between patients' involvement in care and medical expenditures in young cancer survivors. Our research suggests that improved shared decision-making in young cancer survivors may contribute to patients' positive perception of their health and result in survivors seeking fewer medical services resulting in lower medical expenditures.
Footnotes
Authors' Contributions
O.A.A.: Conceptualization, data curation, formal analysis, funding acquisition, investigation, methodology, project administration, resources, software, supervision, validation, visualization, writing—original draft, and writing—review and editing. J.J.: Supervision, visualization, and writing—review and editing. B.H.P.: Supervision, visualization, and writing—review and editing. T.H.M.K.: Conceptualization, investigation, methodology, project administration, funding acquisition, resources, supervision, validation, visualization, and writing—review and editing.
Author Disclosure Statement
No competing financial interests exist.
Funding Information
No funding was received for this article.