Value-based health care aims to maximize patient outcomes while limiting cost.1 However, this model of health care delivery often focuses on outcome metrics such as readmission and hospital-acquired infection, as opposed to more personalized value-based health care measures.
Over an individual patient's cycle of care, the treating surgeon is faced with many treatment choices. Incorporating patient preference and patient values into the clinical decision-making process is a key component in the original concept of value-based health care and ultimately improves the quality of care provided from the patient's perspective.2
Organizations, such as the Agency for Healthcare Research and Quality and the Patient-Centered Outcomes Research Institute, have recently advocated for a more holistic view of patient care.3,4 This approach aims to improve health outcomes by considering the personal, social, and clinical context of a patient's life when assessing and addressing health care needs. Nonclinical interventions, offered in parallel with clinical care, may improve a nonclinical outcome that may be important to the patient and ultimately have an impact that improves the patient's clinical outcomes.5
The objective of this study was to determine the aspects of recovery that are most important to the patient in the subacute phase after a traumatic extremity fracture. The secondary objective was to determine which patient factors are associated with variations in recovery priorities.
In this cross-sectional study, we used a discrete choice experiment (DCE) to determine the relative importance of various recovery domains during the subacute phase after an appendicular fracture. Patients were enrolled at a major academic trauma center from June through December 2018. The institutional review board at the University of Maryland, Baltimore, approved the study.
Adult patients with an appendicular fracture treated at the study location were eligible for inclusion at their first postoperative clinical visit after discharge from their index hospitalization. We excluded patients with a severe traumatic brain injury, patients with spinal cord injuries, and patients who were non–English-speaking.
Discrete Choice Experiment
DCEs are an established method for eliciting patient preferences.6,7 With this technique, respondents are asked to compare 2 or more hypothetical options, called choice sets, based on described attributes. The pooled responses allow for the estimation of the relative importance of the included attributes and the utility associated with their respective levels.
A series of focus groups and semistructured interviews with members of the Trauma Survivors Network support group and patient stakeholders from 2 large pragmatic trials were used to determine and describe patient-important recovery domains and their respective levels for this DCE. Based on the interviews, 3 domains were determined to be of paramount importance: clinical recovery, work-related recovery, and disability benefit support. Clinical recovery was described based on possible outcomes, including amputation, 3 levels of pain, a nonunion, deep surgical site infection, superficial surgical site infection, and no complications. Work-related recovery was described based on preinjury employment, duties, and income. The final domain of disability benefit support was described as a binary attribute.
The Choice Modeling platform in JMP Version 14 (Cary, NC) was used to develop the choice sets in the DCE survey. Each choice set described 2 hypothetical outcomes, and patients were asked to select their preferred outcome within a 1-year time horizon (Fig. 1 for a sample choice set). Forty-eight choice sets were developed using a D-optimal, fractional factorial design with 4 blocks to maximize design efficiency. Each respondent was asked to complete 12 choice sets to minimize respondent burden. After informed consent, the patient was randomized to 1 of the 4 versions of the survey. Before completing the DCE, patients completed the PROMIS Global Health survey based on their preinjury health status. After the DCE, patients were administered questions relating to their preinjury socioeconomic status.
In the survey, we collected sociodemographic and clinical data that were hypothesized to be associated with heterogeneity in the relative importance on the outcome domains. Demographic characteristics included sex, being of a racial minority, and age stratified between those older or younger than 65 years. In addition, we included covariates for educational attainment, marital status, and having dependents. We asked enrolled patients if they were working before the injury, if they had a physical labor occupation, their preinjury income, and the source of the income. Low-income patients were defined based on the US Census Bureau poverty threshold for a 4-person family in 2018.8 Health insurance status was captured. The patients' home address was used to determine their Area Deprivation Index (ADI) that was then stratified into ranked quartiles.9 ADI was developed by the Health Resources and Services Administration and represents a geographic area–based measure of the socioeconomic deprivation experienced by a neighborhood as a composite score of 17 variables, including income disparity and percent of the population aged 25 years and older with at least a high school diploma. Potential clinical factors included those with lower extremity fracture, including pelvis and acetabulum fractures, and if at least 1 fracture was open. The PROMIS Global Health survey provides a preinjury physical and mental health score. Less than 10% of the personal income and ADI data were missing or not reported and were imputed using multiple imputation.10
Hierarchical Bayesian modeling was used to calculate patient-level utilities and relative importance based on the DCE responses. Three models were initially created with 3 different aspects of work-related recovery (employer, duties, and income). The model that included preinjury income, along with clinical recovery and access to disability benefits, provided the best model fit, based on the log-likelihood function, and was used for the final analysis. Relative importance was compared between the recovery domains using an analysis of variance test and the Tukey–Kramer test for the differences between paired domains.
For the second objective, separate models were developed for each of the 3 recovery domains. Given the large number of candidate covariates, we applied a double least absolute shrinkage and selection operator regression technique based on an Akaike information criterion validation. Double least absolute shrinkage and selection operator regression shrinks noninfluential coefficients to 0 and then scales the remaining covariates. This method of penalized regression is robust under conditions of nonnormality and minimized type I error. The absolute difference in the relative importance of the recovery domain associated with each remaining covariate is reported for each of the 3 recovery domain models. All statistical analyses were performed with JMP Version 14 (Cary, NC).
Two hundred nine eligible patients consented for the study. Eleven of the patients withdrew from the study before completing the questionnaire and were excluded from the final analysis. Of the 198 patients included in the analysis, the median age was 42 years (interquartile range: 29–58), 63% were male, and the median preinjury annual income was $35,000 (interquartile range: $15,000–$57,500) (Table 1).
The utility associated with the clinical outcomes ranged from amputation (mean: −2.3; SD: 0.5) to no complications (mean: 2.0; SD: 0.1) (see Appendix 1, Supplemental Digital Content 1, https://links.lww.com/JOT/A834). Work-related recovery utilities ranged from no income (mean: −1.1, SD: 0.2) to equivalent preinjury income (mean: 0.7; SD: 0.1). The mean utility for receiving a disability benefit was 0.4 (SD: 0.3) compared with a mean of −0.4 (SD: 0.3) for no other income support.
The DCE revealed that patients weighted their clinical recovery (mean: 62%, SD: 5.3) as most important. Work-related recovery (mean: 27%, SD: 3.9) and the receipt of other disability benefits (mean: 11%, SD: 6.4) were each of significantly less relative importance (see Appendix 2, Supplemental Digital Content 2, https://links.lww.com/JOT/A835). Heterogeneity was observed across these estimates (Table 2). Specifically, having a physically demanding preinjury occupation was associated with a 17% absolute decrease in the relative importance of clinical recovery and a 14% increase in the relative importance of receiving disability benefits. Having an open fracture was associated with a 15% increase in the relative importance of work-related recovery and a 10% decrease in the relative importance of disability benefits. Working before the injury increased the relative importance of work-related recovery by 7%. High preinjury physical health status was associated with a 10% increase in the relative importance of clinical recovery and a 12% decrease in the relative importance of disability benefits. The relative importance of disability benefits was 10% higher in patients with health insurance from a government provider, compared with those with private health insurance.
Patients prioritized clinical recovery as a clear focus during the subacute recovery phase. However, patients also valued access to disability benefits and resuming work. Significant heterogeneity was observed in patient recovery priorities based on the physical demands of preinjury employment, preinjury physical health, preinjury work status, health insurance type, and the severity of the fracture. Understanding the patient's recovery priorities early in the clinical care pathway will enable the development of multidisciplinary care plans that are responsive to these recovery priorities.
The findings of the study suggest that the income provided by employment is of greater importance to the patient than returning to their specific employer or their previous duties. The patient prioritization on income recovery after their fracture is well justified by recent literature on the high risk of catastrophic health expenditures faced by many trauma patients in the United States.11 Traumatic injuries are, by definition, unplanned. These sudden, unanticipated, and often high-cost treatments impose an unexpected financial strain on patients and their families. Even if the direct costs associated with the treatment are covered through an insurer, most patients incur substantial indirect costs related to time away from work and decreased productivity. A recent meta-analysis noted that fracture patients are absent from work for an average of 3 months after injury, and one-third of fracture patients do not return to work by 1 year after injury.12 Furthermore, withdrawal from the labor force is associated with increased rates of mortality, drug overdose, suicide, disability program enrollment, and incarceration.13
Several factors may explain the variation in patient recovery priorities. The increase in the relative importance of disability benefits for patients with physically demanding preinjury occupations was of similar magnitude to the decreased relative importance of their clinical recovery. These physically demanding occupations were mostly in construction or manufacturing sectors of the economy. The variation in recovery preferences for these patients may be more of a reflection of job satisfaction in those sectors, than specific to the physical demands of the occupation.14 The prioritization of disability benefits for patients in physically demanding occupations may also represent the patient's uncertainty regarding their ability to functionally return to physically demanding work after a fracture. Tversky and Kahneman's15 value function theory has 3 essential characteristics, reference dependence, loss aversion, and diminishing sensitivity, that may explain the variation in recovery preferences based on preinjury physical health, preinjury working status, severity of the injury, and health insurance status. Reference dependence hypothesizes that carriers of value are gains and losses defined relative to a reference point. Patients with severe open fractures may view a larger gap between work-related recovery and their current injured state, thus elevating its relative importance. Patients who receive health insurance through a government program may have a different reference for receiving disability benefits which are often administered by a government agency. Loss aversion suggests that losses are associated with a greater psychological burden than the benefits from corresponding gains. Patients who were working before their injury place a higher value on work-related recovery than patients who were not working before their injury place on gaining postinjury employment. Diminishing sensitivity proposes that the value of both gains and losses decreases as the distance from the reference point becomes larger. Based on this theory, patients with lower levels of preinjury physical health have a diminished value for clinical recovery from their current injury.
Innovative approaches to value-based care can support the recovery priorities of individual patients while providing financial benefits to the provider under a value-based payment model. A recent randomized controlled trial at an urban Chicago hospital demonstrated that hospitals that provided housing and a case management program for chronically ill homeless adults significantly reduced hospitalizations and emergency department visits.2 A recent report by the Commonwealth Fund recommends that Medicaid reimburse providers at higher payment rates if they invest in social interventions.16 These social interventions may include assistance in retaining or finding employment, peer support programs, stable housing programs, and linkages with programs that offer help with food assistance, rent, childcare costs, heating bills, and other major household expenses.
The results of this study provide insight into the clinical and social recovery priorities of orthopaedic trauma patients. However, the results must be interpreted within the limitations of the study design. The data were collected from a single trauma center. Variations in the demographics of the patient population and socioeconomic conditions in the region may reduce the generalizability of the results. In addition, the patients responded to hypothetical scenarios on their clinical and socioeconomic recovery. Five-percent of eligible patients withdrew from the study after initial consent was obtained. The recovery scenarios presented in the DCE may have been emotionally sensitive or cognitively challenging. It is possible that some patients completing the questionnaire experienced similar challenges with the study, and those challenges may have biased their responses. Although several factors were found to be associated with significant heterogeneity in recovery preferences, the independent variables included in the models accounted for less than 10% of the variability of the dependent variable. Finally, all responses were collected within the first 6 weeks after the index admission. It is possible that the relative importance of the recovery domains may vary with an increased time from injury and in response to future socioeconomic and clinical conditions.
Orthopaedic surgeons are trained to focus on and improve a patient's clinical recovery. Although clinical improvement was of foremost importance to the study patients during the subacute phase, surgeons may improve overall value by incorporating a broader definition of recovery into their care plan. This study is an essential first step to understanding the variation in recovery priorities for orthopaedic trauma patients and improving value-based orthopaedic care delivery. Further research is required to determine whether recovery preferences remain static after injury or change over time or in response to future circumstances.
The authors thank Ms Frances Grissom and the members of the Trauma Survivors Network who were instrumental in the design of the study. The authors thank patient stakeholders from the PREVENT CLOT and PREP-IT trials for invaluable feedback during the study design.
1. Porter ME. What is value in health care? N Engl J Med. 2010;363:2477–2481.
2. Sadowski LS, Kee RA, VanderWeele TJ, et al. Effect of a housing and case management program on emergency department visits and hospitalizations among chronically ill homeless adults: a randomized trial. JAMA. 2009;301:1771–1778.
3. Research Priorities. Agency for Healthcare Research and Quality. Available at: https://www.ahrq.gov/funding/policies/foaguidance/index.html#priorities
. Accessed May 5, 2019.
4. Frank L, Basch E, Selby JV. Patient-Centered Outcomes Research Institute. The PCORI perspective on patient-centered outcomes research. JAMA. 2014;312:1513–1514.
5. Taylor LA, Tan AX, Coyle CE, et al. Leveraging the social determinants of health: what works? PLoS One. 2016;11:e0160217.
6. McFadden D. Conditional logit analysis of discrete choice behavior. Front Econ. 1974;4:105–142.
7. Ryan M, Gerard K, Amaya-Amaya M, eds. Using Discrete Choice Experiments to Value Health and Health Care. Dordrecht, The Netherlands: Springer Science & Business Media; 2007.
9. Kind AJH, Buckingham WR. Making neighborhood-disadvantage metrics accessible—the neighborhood atlas. N Engl J Med. 2018;378:2456–2458.
10. Little RJ, Rubin DB. Statistical Inference with Missing Data. 2nd ed. New York, NY: Wiley; 2002.
11. Scott JW, Havens JM, Raykar N, et al. Cured into destitution: catastrophic health expenditure risk among uninsured trauma patients in the United States. Ann Surg. 2018;267:1093–1099.
12. O'Hara NN, Isaac M, Slobogean GP, et al. The Socioeconomic Impact of Orthopaedic Trauma: A Systematic Review and Meta-analysis. Washington, DC: Academy Health Annual Meeting; 2019.
13. Coile C, Duggan M. When Labor's Lost: Health, Family Life, Incarceration, and Education in a Time of Declining Economic Opportunity for Low-Skilled Men. National Bureau of Economic Research. Working Paper 25569.
14. Sousa-Poza A, Sousa-Poza AA. Well-being at work: a cross-national analysis of the levels and determinants of job satisfaction. The J Socio-Economics. 2000;29:517–538.
15. Tversky A, Kahneman D. Loss aversion in riskless choice: a reference-dependent model. Q J Econ. 1991;106:1039–1061.
16. Bachrach D, Guyer J, Meier S, et al. Enabling Sustainable Investment in Social Interventions: a Review of Medicaid Managed Care Rate-Setting Tools. New York, NY: The Commonwealth Fund; 2018. Available at: https://www.commonwealthfund.org/sites/default/files/documents/___media_files_publications_fund_report_2018_jan_bachrach_investment_social_interventions_medicaid_rate_setting.pdf
. Accessed May 7, 2019.