Abstract
Objective:
Hip fractures have negative humanistic and economic consequences. Predictors and sub-groups of negative post-fracture outcomes (high costs and extensive healthcare utilization) were identified in patients with and without muscle atrophy/weakness (MAW).
Methods:
Truven Health MarketScan data identified patients ≥50 years old with inpatient hospitalizations for hip fracture. Patients had ≥12 months of continuous healthcare insurance prior to and following index hospitalization and no hip fracture diagnoses between 7 days and 1 year prior to admission. Predictors and sub-groups of negative outcomes were identified via multiple logistic regression analyses and classification and regression tree (CART) analyses, respectively.
Results:
Post-fracture 1-year all-cause healthcare costs (USD$31,430) were higher than costs for the prior year ($18,091; p < 0.0001). Patients with MAW had greater post-fracture healthcare utilization and costs than those without MAW (p < 0.05). Greater post-fracture costs were associated with a higher number of prior hospitalizations and emergency room visits, length of index hospitalization, Charlson Comorbidity Index (CCI), and discharge status; diagnosis of rheumatoid arthritis, osteoarthritis, or osteoporosis; and prior use of antidepressants, anticonvulsants, muscle relaxants, benzodiazepines, opioids, and oral corticosteroids (all p < 0.009). High-cost patient sub-groups included those with MAW and high CCI scores.
Conclusions:
Negative post-fracture outcomes were associated with MAW vs no MAW, prior hospitalizations, comorbidities, and medications.
Introduction
Hip fractures, a serious consequence of osteoporosis, are prevalent in the aging populationCitation1, with women over 50 years old accounting for 75.1% of all fracturesCitation2. More than 90% of all fractures are the result of fall injuries and occur most often in persons at least 65 years oldCitation3. In 2005, women and men residing in the US incurred nearly 300,000 hip fracturesCitation2. An extensive 60-year review of data from 63 countries revealed a 10-fold variation in hip fracture risk and probability between countries, with a moderate age-standardized incidence observed in women in the US (260/100,000)Citation4. The significance of hip fractures is growing, and the number of fractures in the US is expected to increase substantially, with an anticipated doubling of the older patient population (≥65 years) worldwide from 506 million in 2008 to 1.3 billion in 2040Citation1. Hip fracture-associated mortality secondary to osteoporosis may linger for up to a decade after the eventCitation1.
Aging individuals who experience a hip fracture face considerable disability and reduced quality and length of life. A population-based cohort study in community-dwelling patients who sustained a hip fracture found that 99% required hospitalization, 53% were discharged to an extended care facility, 8% entered an extended care facility within 1 year of the event, and 5% died before dischargeCitation5. Another study in elderly community-dwelling patients reported that ∼17% of a patient’s remaining life could require rehabilitation and extended long-term facility care following a hip fractureCitation6. Further, new deficits in activities of daily living (e.g., reduced mobility, bathing, dressing, eating) in the first 6 months following hip fracture accounted for at least half of the observed morbidity and mortality. Overall, hip fracture was associated with a reduced life expectancy of nearly 2 years, with short-term mortality occurring within 6 months of the event.
The economic burden of hip fracture is significant. In one study, the cost of ∼$27,000 (US 2001) was attributed to healthcare expenses accrued in the first 6 months, largely due to rehabilitation facility charges; the estimated lifetime attributable cost was over $81,000Citation6. Other studies corroborate the expense of caring for older patients with hip fractures based on the following: >340,000 emergency room (ER) visits in the US during 2008Citation7; direct hospitalization costs per patient ranging from $8,358–$32,195Citation8; and estimated direct care costs within the first year of occurrence at $30,000 per patientCitation1. Based on an estimated projection of >620,000 hip fractures in 2040, the lifetime cost of treating hip fractures in the US will likely exceed $47 billionCitation6.
Muscle atrophy/weakness (MAW), a sequelae of hip fracture, is largely due to the associated reduction in mobility. Loss of muscle mass, weakness, and low activity, which are three components of the geriatric frailty syndrome, increase the probability of a fallCitation9. In the elderly, lack of muscle movement may result in loss of muscle strength by as much as 10–15% per week, and long-term inactivity of unused muscles (∼2 months) may lead to substantial muscle atrophy (i.e., 50% reduction)Citation10. Development of MAW among hip fracture patients has been associated with poor physical functioningCitation11 and poor recovery of mobilityCitation11,Citation12. In a small study of 90 women ≥65 years old with hip fracture, Visser et al.Citation12 demonstrated that loss of muscle strength without loss of muscle mass was an independent predictor of poor functional recovery in the first year following a hip fracture.
This study identifies characteristics of hip fracture patients with or without MAW that predict negative post-fracture outcomes (e.g., high costs/utilization, longer hospitalization/rehabilitation, repeated surgery) and sub-groups of patients at increased risk for negative post-fracture outcomes using classification and regression tree (CART) analysis. It was hypothesized that patients with MAW would utilize more healthcare resources during and following recovery from hip fracture.
Patients and methods
Study design and data sources
Data for this retrospective analysis were derived using the Truven Health MarketScan database to identify US patients who had experienced a hip fracture and required an inpatient hospital stay (index hospitalization) between January 1, 2006 and September 30, 2010. Truven Health MarketScan (formerly Thomson Reuters) is one of the largest claims databases in the US (marketscan.truvenhealth.com). This database longitudinally captures detailed administrative medical and pharmacy claims including cost (payment) and healthcare utilization information for healthcare services performed in both inpatient and outpatient settings (e.g., person-specific clinical utilization, expenditures, prescription drug, and carve-out services) from ∼25 million individuals annually. The data are derived from a selection of large employers, health plans, and government and public organizations. For study populations included in the database, medical claims and encounter data are linked to detailed patient information across sites and types of providers. The annual medical databases include private-sector health data from more than 100 payers. Approximately 150 million unique members were represented in the database during 2004–2012. These data represent the medical experience of insured employees and their dependents for active employees, early retirees, Consolidated Omnibus Budget Reconciliation Act (COBRA) continuees, and Medicare-eligible retirees with employer-provided Medicare Supplemental plans. The Truven Health MarketScan is a rich database that provides representative inpatient, outpatient, prescription drug claims, and cost record data, thereby allowing payers to see how their populations are faring and where their healthcare dollars are being spent. MarketScan data are the basis of over 700 peer-reviewed articles published in leading journals since 1990.
The current study extracted patients from the database aged 50–64 years with commercial insurance (Commercial) or those at least 65 years old with Medicare supplemental insurance (Medicare) ().
Table 1. Attrition of patients hospitalized for hip fracture.
This retrospective claims database study is fully compliant with Health Insurance Portability and Accountability Act (HIPAA) privacy requirements and captures person-specific healthcare use, expenditures, and enrollment across inpatient, outpatient, and prescription drug services. Ethic committee approval was not required.
Study population
Study patients were identified using the following inclusion criteria: adults 50–64 years old for the Commercial population and at least 65 years of age for the Medicare population as of the index date (admission date of hospitalization for hip fracture between January 1, 2006 and September 30, 2010), with documented hip fracture identified per ICD-9 codes (820.xx and 733.14); continuous healthcare insurance enrollment for at least 12 months prior to the index hospitalization date, during index hospitalization, and at least 12 months following the index hospitalization discharge date; and no hip fracture diagnoses between 7 days and 1 year prior to the index hospitalization date. Selection of hospital stays due to hip fracture was based on primary and secondary diagnosis of hip fractures.
The hip fracture patient population was subsequently stratified into three cohorts as follows: those with no diagnosis of MAW (no-MAW) from the pre-period to the post-period; those diagnosed with MAW in the 1-year before the index date (pre-MAW); and those diagnosed with MAW during or after the index hospital stay, but no MAW claim in the pre-index period (post-MAW). MAW was identified using ICD-9 codes of 728.2x (disorders of muscle, ligament, and fascia; muscular wasting and disuse atrophy, not elsewhere classified) and 728.87 (disorders of muscle ligament and fascia; other disorders of muscle, ligament, and fascia; muscle weakness (generalized)Citation13.
Analyses
Analyses were performed to assess differences in demographic characteristics, prior comorbidities, prior medication use, prior healthcare resource utilization and index stay, and pre- and post-index 1-year healthcare costs among the no-MAW, pre-MAW, and post-MAW cohorts. Patient characteristics included payer type (Commercial vs Medicare), age, gender, medication use, and comorbid conditions (e.g., osteoarthritis, diabetes, osteoporosis). Overall general health was approximated using the modified Charlson Comorbidity Index (CCI), which measures 17 different categories of comorbidities and generates a corresponding scoreCitation14. Prior resource use variables included any hospitalizations, any ER visits, and any outpatient visits. Index hospitalization was evaluated for length of stay, cost of hospitalization, and discharge status.
Pre- and post-index costs were calculated as total healthcare costs and sub-divided into inpatient, outpatient, and pharmacy costs. All costs were adjusted to 2011 US dollars using the medical care component of the Consumer Price IndexCitation15. For example, if the percentage change (PC) from 2010 to 2011 is PC11, the cost from 2010 will be adjusted as Cost2010*(1 + PC11). Similarly, the costs from 2006 will be adjusted as Cost2006*(1 + PC07)*(1 + PC08)*(1 + PC09)*(1 + PC10)*(1 + PC11).
Unadjusted differences between groups were compared using chi-square statistics for categorical variables (gender, region of residence, health plan type, medications, comorbidities, and proportion with resource use) and Student’s t-test for continuous variables (age, CCI, length of index hospital stay, and healthcare resource utilization, including total hospital days, ER visits, and healthcare costs in the pre- and post-index periods).
Logistic regression analyses
Patient characteristics associated with high post-fracture total healthcare costs were predicted using stepwise multiple logistic regression (MLR), controlling for age, gender, region of residence, health plan type, MAW status, discharge status, length of index stay, prior medication use (non-steroidal anti-inflammatory drugs [NSAIDs], antidepressants, anticonvulsants, muscle relaxants, benzodiazepines, opioids, oral corticosteroids, and bisphosphonates), CCI, comorbid medical conditions (e.g., osteoarthritis, rheumatoid arthritis, and osteoporosis), and pre-index resource use (e.g., number of hospital stays, fracture-related hospital stays, and ER visits). A binary variable for total healthcare costs with the median as the cut-off point was set as the response variable in the logistic regression model. The p-values for variables to enter and exit the models were 0.1 and 0.05, respectively. Odds ratios (ORs) and 95% confidence intervals (CIs) for each covariate in the final model were reported.
CART analyses
Introduced nearly three decades agoCitation16, CART analyses have recently been identified as preferred data mining tools, often for the purposes of increasing revenue, cutting costs, and determining factors that influence costCitation17. CART analyses depend on decision tree algorithms, which consider all input variables that are most strongly linked to the outcome under study and all subsequent splits of the data until no further significant split can be identified (terminal node), aiming to construct a tree with high predictive power. Accordingly, CART analyses were used in the current study to identify patient sub-groups at increased risk for negative post-fracture outcomes. Partitioning or splitting was stopped when two sub-groups resulting from any further partition were not statistically different using a chi-square test (for categorical variables) or F-test (for continuous variables; α = 0.05) or the node size was less than 75 patients (∼10% of the pre-MAW sample). The trees were trimmed by focusing on sub-groups with negative economic outcomes.
Table 2. Patient characteristics.
CART analyses were performed on a random sample from the study population and were based on 52 input variables, including demographics, comorbidities, prior medications, and prior healthcare utilization. Multiple random samples were tested (up to 20,000 times) to validate the results. Target outcome variables included post-1-year total cost, post-1-year discharge to skilled nursing facility (SNF), length of index stay, and post-1-year length of hospital stay. CART analyses were performed using SAS Enterprise Miner version 6.2 (SAS Institute Inc., Cary, NC), and the significance level was set a priori at p < 0.05.
Sensitivity analyses
All of the aforementioned analyses were also performed using a broader definition of MAW including the ICD-9 codes of 728.2x and 728.87 as well as 335.1x (anterior horn cell disease; spinal muscular atrophy), 335.21 (anterior horn cell disease; motor neuron disease; progressive muscular atrophy), and 359.xx (muscular dystrophies and other myopathies)Citation13.
Results
A total of 34,993 patients who required inpatient hospitalization for hip fracture and satisfied study inclusion criteria were identified from the database. The population consisted primarily of Medicare (n = 29,643) vs Commercial (n = 5350) patients. The remainder of the analyses described herein were performed on the combined Medicare/Commercial population and stratified into cohorts based on the presence or absence of MAW: pre-MAW in 1214 (3.5%), post-MAW in 3948 (11.3%), and no-MAW in 29,831 (85.2%) patients.
Demographics, prior co-morbid conditions, and prior medications
Mean age for the overall population was 78 (SD 11) years (). Patients with pre- or post-MAW were slightly older (80 years) than those with no-MAW (78 years; p < 0.0001). As predicted, there was a predominance of females (71%).
Prior to hip fracture, the following co-morbid conditions were most prevalent: osteoarthritis (23.2%), diabetes (19.1%), osteoporosis (19.0%), chronic pulmonary disease (17.6%), cerebrovascular disease (16.2%), tumors (12.4%), and congestive heart failure (10.4%) (). Each of these conditions was reported at numerically higher rates in patients with pre- or post-MAW compared to those with no-MAW (each p ≤ 0.02), with the exception of osteoarthritis in post-MAW patients (p = 0.06). Mean overall general health score per CCI was 1.4 (SD = 1.8); patients with pre- or post-MAW had significantly higher CCI scores prior to their hip fracture compared to those with no-MAW (p < 0.0001). Pre-MAW patients also had significantly higher CCI scores than post-MAW patients prior to hip fracture (p < 0.0001).
Medication use prior to hip fracture was common, with the following drugs identified in >10% of patients: opioids (41%), antidepressants (32%), oral corticosteroids (22%), NSAIDs (21%), benzodiazepines and bisphosphonates (each 18%), and anticonvulsants (15%) (). Antidepressants and anticonvulsants were used numerically more often in patients with pre- or post-MAW compared to no-MAW (p ≤ 0.0005).
Resource use prior to, during index hospitalization, and following hip fracture surgery
Hospital admissions, ER visits, and outpatient visits within the year prior to hip fracture were common, with rates of 25.0%, 47.1%, and 74.5%, respectively ().
Table 3. Healthcare resource utilization over the 1-year pre- and post-index periods.
The rate of any hospital stay was significantly higher among pre-MAW patients (58.1%) vs post-MAW (27.9%) and no-MAW patients (23.2%) (each p < 0.0001). Corresponding rates of prior hospital admissions related to fracture other than hip fracture were also significantly higher in pre-MAW patients vs post-MAW and no-MAW patients (8.7%, 3.5%, and 2.6%, respectively; each p < 0.003).
Patients with pre-MAW also had significantly higher rates of ER and outpatient visits compared to those with post-MAW and no-MAW (). The mean number of ER visits was 1.0 (SD = 1.8) for the overall population, including 2.4 (SD = 3.3) for pre-MAW, 1.2 (SD = 2.1) for post-MAW, and 0.9 (SD = 1.7) for no-MAW (each p < 0.0001). The mean number of outpatient visits was 4.2 (SD = 8.6) for the overall population, including 8.3 (SD = 13.1) for pre-MAW, 4.5 (SD = 7.4) for post-MAW, and 4.0 (SD = 8.5) for no-MAW (each p ≤ 0.0003).
While the length of index hospitalization for hip fracture was slightly longer for the MAW cohorts, costs were similar between those with pre-MAW and no-MAW and numerically higher for those with post-MAW vs no-MAW (p = 0.0006) (). Mean index hospitalization duration was ∼6 days for the overall population; patients with pre- and post-MAW had numerically longer mean lengths of stay for hip fracture surgery (6.3–6.4 days) compared to those with no-MAW (5.8 days; p < 0.004). Mean index hospitalization costs were $20,328 for pre-MAW, $22,567 for post-MAW, and $20,964 for no-MAW (p ≤ 0.002 for comparisons of post-MAW to pre- or no-MAW). Overall, 41 (0.1%) patients had a discharge status of death following admission for their hip fracture. Regardless of MAW status, most patients went home (37.6%) or to a SNF (28.2%) upon discharge.
Re-hospitalizations, ER visits, and outpatient visits were common (each > 30%) in the year following index hospitalization in patients who had sustained a hip fracture (). The rate of any hospital stay was significantly higher among pre-MAW (40.3%) and post-MAW (43.0%) vs no-MAW patients (29.1%) (each p < 0.0001). Corresponding rates of hospital admissions related to the hip fracture were also significantly higher in pre-MAW and post-MAW patients vs no-MAW (11.4%, 12.2%, and 7.5%, respectively; each p < 0.0001). The mean number of re-hospitalization days, related or unrelated to the fracture, was ∼2-fold higher for patients with pre- or post-MAW compared to no-MAW (each p < 0.001). Numerically higher rates of ER (∼2 vs 1) and outpatient visits (∼8 vs 6) were also observed for patients with pre- and post-MAW compared to the no-MAW cohort (each p < 0.05).
One-year healthcare costs pre- and post-index hospitalization
Overall, mean post-fracture 1-year all-cause healthcare costs ($31,430 [SD = $47,279]) were significantly higher than costs for the 1-year period prior to index hospitalization ($18,091 [SD = $53,668]; p < 0.0001) (). Inpatient and outpatient costs were significantly higher in the 1-year post-index period compared to the year prior (p < 0.0001). Patients with pre- and post-MAW generally had greater mean pre- and post-1-year total costs (each p < 0.0001) compared to no-MAW patients.
Table 4. One-year healthcare costs pre- and post-index hospitalization.
Identifiers of high post-fracture healthcare costs
Patient characteristics predictive of negative outcomes (i.e., high costs, high healthcare utilization, long hospitalization) during the 1-year following index admission for hip fracture were identified via MLR analyses. Greater total post-fracture costs were associated with the presence of MAW; higher CCI score; a higher number of prior hospitalizations and ER visits, length of index hospital stay, and discharge status to a facility; diagnosis of rheumatoid arthritis, osteoarthritis, or osteoporosis; and prior use of medications (all p < 0.009) ().
Figure 1. Demographic and clinical predictors of greater total healthcare costs in 34,993 patients during 1-year following hip fracture. *For age, the reference is 50–54 years. †For medications and comorbidities, the reference level is ‘no’. ‡For index hospital discharge status, the reference level is ‘home’.
Patients with pre-MAW were more likely to have greater post-1-year healthcare costs compared with those with no pre-MAW (OR = 1.27; 95% CI = 1.11–1.46). Higher post-fracture costs were also linked to the total CCI score (OR = 1.21; 95% CI = 1.19–1.23), as well as the number of hospital admissions (OR = 1.12; 95% CI = 1.08–1.17) and ER visits (OR = 1.07; 95% CI = 1.05–1.09) in the year prior to index hospitalization. Longer index hospitalization was more likely to lead to greater post-1-year healthcare costs (OR = 1.03; 95% CI = 1.02–1.03). Not unexpectedly, discharge to either a SNF (OR = 2.39; 95% CI = 2.25–2.54) or an inpatient rehabilitation facility (OR = 1.37; 95% CI = 1.26–1.48) substantially increased the likelihood of having greater post-1-year healthcare costs compared to patients who went home. A diagnosis of rheumatoid arthritis also was more likely to result in higher post-fracture costs (OR = 1.52; 95% CI = 1.32–1.75), and osteoarthritis (OR = 1.12; 95% CI = 1.06–1.18), or osteoporosis (OR = 1.10; 95% CI = 1.04–1.17) were also associated with a relatively higher risk of higher total healthcare costs in the year immediately following hip fracture. Patients with a prior history of using antidepressants (OR = 1.46; 95% CI = 1.39–1.54), anticonvulsants (OR = 1.44; 95% CI = 1.35–1.54), muscle relaxants (OR = 1.20; 95% CI = 1.10–1.31), opioids (OR = 1.20; 95% CI = 1.14–1.26), oral corticosteroids (OR = 1.14; 95% CI = 1.08–1.21), and benzodiazepines (OR = 1.09; 95% CI = 1.02–1.16) had escalated costs following hip fracture.
Negative outcome patient sub-groups: CART analyses
CART analyses were performed to identify patient sub-groups associated with post-1-year total healthcare costs, SNF discharge, and length of index and post-1-year hospital stays.
Overall, patients with MAW (pre- and post-) and higher CCI scores (≥4) had the greatest post-fracture total healthcare costs (). Specifically, patients with MAW had a mean total cost of ∼$20,000 more than the total cost incurred in patients with no-MAW during the 1-year post-fracture surveillance period. A CCI score ≥4 incurred nearly double the total 1-year post-fracture cost compared to a score <4, regardless of MAW status. Other factors, such as relative younger age and increased number of outpatient visits in the year prior to index hospitalization, were also associated with increased total costs, although not uniformly across the three cohorts.
Figure 2. Classification and regression tree analysis for identifying clinical factors that predict post-1-year total costs.
Discharge to a SNF following index hospitalization for hip fracture was more common in patients with MAW (pre- and post-) and advanced age (≥63.5 years with pre-MAW, ≥64 years post-MAW, or ≥76.5 years no-MAW) (). Older patients (≥64 years) with and without MAW were 2-times more likely to be discharged to a SNF.
Figure 3. Classification and regression tree analysis for identifying clinical factors that predict discharge to skilled nursing facility.
Extended index hospitalization for hip fracture was linked to patients with MAW and those with longer prior 1-year acute hospitalizations (). Patients with MAW had a mean 0.7-day longer hospital stay for their fracture compared to those with no-MAW. Mean index length of hospitalization was 9.1 and 13.4 days in patients with pre- or post-MAW and a prior 1-year hospital length of stay of at least 13–14 days vs ∼6 days in those with shorter prior hospitalizations (<13–14 days). In patients with no-MAW, mean index hospitalization length was ∼2 days longer in those with extended (≥3.5 days) vs shorter prior hospitalizations (<3.5 days).
Figure 4. Classification and regression tree analysis for identifying clinical factors that predict length of index stay.
For patients requiring re-hospitalization during the post-1-year period, sub-groups of patients with longer post-1-year hospital stays included those with MAW (pre- and post-) and longer pre-1-year acute hospital lengths of stay (pre-MAW, post-MAW and no-MAW cohorts) of ≥19.5, ≥14.5, and ≥11.5 days, respectively (). On average, the length of re-hospitalization in patients with MAW during the 1-year post-fracture period was approximately double that of patients with no-MAW (6.3 vs 3.0 days).
Figure 5. Classification and regression tree analysis for identifying clinical factors that predict post-1-year length of hospital stay.
Sensitivity analyses
Sensitivity analysis using a broader definition of MAWCitation13 identified 57 (4.7%) more patients with pre-MAW and 31 (0.8%) more patients with post-MAW. All of these 88 patients were previously grouped in the no-MAW cohort. The results from the sensitivity analysis, including patient characteristics, healthcare resource utilization and costs over the 1-year pre- and post-index periods, the predictors of greater healthcare costs from multiple logistic regression analysis, and CART sub-groups, were very similar to the original analysis, with slight numerical variations.
Discussion
Because the incidence of hip fractures is currently high and is expected to substantially increase (3–8-fold) over the next several decadesCitation5, identifying patient sub-types at higher risk for negative outcomes could be beneficial, especially if effective interventions can be instituted to reduce overall clinical and economic consequences.
This large retrospective database study in nearly 35,000 patients, which identified patients ≥50 years old with inpatient hospitalizations for hip fractures using US Commercial and Medicare claims, provides interesting new insights into the profile of the older adult patient who is likely to experience a negative outcome following a hip fracture (i.e., high costs, high healthcare utilization, long hospitalizations, and discharge to an extended care facility). Overall, the vast majority of patients included in the analysis were on Medicare (84.7%) as expected, with a hip fracture target population ≥65 years (mean 78.1 years), with females comprising more than two-thirds of the population. In the year prior to hip fracture, osteoarthritis and osteoporosis (23% and 19%, respectively) as well as high use of concomitant medications such as opioids, antidepressants, oral corticosteroids, and NSAIDs (each >20%) were commonly reported. A diagnosis of MAW (either prior to or in the first year following hip fracture) was confirmed in 5162 (15%) patients. These demographic and medical attributes, particularly older age, female gender, underlying osteoporosis, and muscle weakness, are comparable with other studies that describe patients who have sustained a hip fractureCitation2,Citation4,Citation5,Citation7,Citation11,Citation18,Citation19. In the current study, increased healthcare resource use and greater expenditures in the year following hip fracture were initially identified using MLR analyses. Predictors of negative post-fracture outcomes included: MAW, an increased CCI score, greater number of prior hospitalizations and ER visits, increased length of index hospital stay, and discharge to a facility as well as a diagnosis of rheumatoid arthritis, osteoarthritis, or osteoporosis and prior use of medications including opioids, antidepressants, oral corticosteroids, and NSAIDs. Using CART analyses, we identified that patients with pre- or post-MAW and higher CCI scores (≥4) predicted negative outcomes. Notably, the mean cost of an average index stay (6 days) for a hip fracture was ∼$21,000, which is aligned with recently reported direct costs for treatment of osteoporosis-related hip fractures in the USCitation8. Yet, additional 1-year pre- and post-fracture healthcare utilization costs of $18,100 and $31,400, respectively, raised the overall cost to over $70,000 per patient for the 2-year observational period. Patients with MAW (pre- and post) had even higher total 2-year costs of ∼$111,100 and $89,900, respectively, compared with $66,400 for the no-MAW cohort, which was largely driven by discharge to an extended care facility and the need for additional hospitalization in the year following the event. Notably, the findings from this study were not sensitive to a broader definition of MAW.
Although it is not unexpected that the presence of MAW was a strong predictor of greater healthcare utilization and economic burden after hip fracture, it is noteworthy that MAW prior to hip fracture and development of MAW in the year following the event were associated with negative outcomes. Age-related decrease in skeletal muscle mass coupled with low muscle strength or performance has been associated with many physical limitations, chronic diseases, and reduced quality-of-lifeCitation20. Interventions that increase both muscle strength and mass may prove beneficial to older hip fracture patients, including a shortened rehabilitation time and perhaps reduced economic burden.
Our study confirmed that patients with osteoporosis were pre-disposed to hip fracture, as well as negative post-fracture outcomes (i.e., higher healthcare utilization and costs). The role of osteoporosis in the genesis of hip fractures is well described, particularly in post-menopausal women and the elderlyCitation21. Fragility fractures commonly arise following a fall, secondary to decreased bone mass and deterioration of bone tissue, and are frequently the consequence of age-related increased bone resorption. Current available treatments for osteoporosis do not guarantee the prevention of future fractures. The majority of hip fracture cases in our study population were due to pathological bone; only a very small portion was due to trauma (<5%, data not shown).
Several limitations should be considered when interpreting the results of these analyses. Patients were identified by diagnostic codes, which may not be applied uniformly across providers, therefore may result in under-reporting of MAW. The analysis only included patients with healthcare benefit coverage without prior hip fractures and those who survived ≥1 year post-fracture. As such, the healthcare costs derived in these analyses cannot be applied to patients who had previous events or who died due to the initial hip fracture. Medical claims data did not include important outcomes such as quality-of-life, indirect costs, or baseline severity of co-morbid conditions. The low prevalence of MAW observed in this database potentially under-estimates the findings of this study. Finally, the associations demonstrated in this analysis are not necessarily causal in nature.
In conclusion, this study extends current knowledge on total hip fracture costs by examining the specific impact of identified MAW on healthcare utilization and expenses before and following a hip fracture event using recent real-world data. Specifically, the findings from this study indicate that healthcare costs related to a hip fracture extend beyond the index hospitalization, with potential costs exceeding $70,000 per patient when the 1-year periods prior to and following the event are included. Presence of MAW was found to be a strong predictor of total cost related to hip fracture in this aging population. Identification, monitoring, and management of older patients with MAW are likely to reduce severity of the condition, improve health outcomes, and reduce costs. Whether interventions such as improved diet, increased physical activity, and specific novel treatment modalities can manage or prevent MAW and reduce the incidence of hip fractures in older patients will require further study.
Transparency
Declaration of personal interests
ZC, RB, MS, & EB are all employees and stockholders of Eli Lilly and Company. YC was an employee of InVentiv Health clinical at the time of the study.
Declaration of funding interests
This study and the preparation of this paper were funded in full by Eli Lilly and Company. Data analyses were performed, in part, by InVentiv Health clinical, Indianapolis, IN. Writing support was provided by Teresa Tartaglione, PharmD of ClinGenuity, LLC, Cincinnati, OH. JME peer reviewers on this manuscript have no relevant financial or other relationships to disclose.
Acknowledgments
The authors thank Teresa Tartaglione, PharmD (ClinGenuity, LLC, Cincinnati, OH) for medical writing support and assistance with preparation and submission of this paper.
References
- Cauley JA. Public health impact of osteoporosis. J Gerontol A Biol Sci Med Sci 2013;68:1243-51
- Burge R, Dawson-Hughes B, Solomon DH, et al. Incidence and economic burden of osteoporosis-related fractures in the United States, 2005–2025. J Bone Miner Res 2007;22:465-75
- Bischoff-Ferrari HA. The role of falls in fracture prediction. Curr Osteoporos Rep 2011;9:116-21
- Kanis JA, Oden A, McCloskey EV, et al. A systematic review of hip fracture incidence and probability of fracture worldwide. Osteoporos Intl 2012;23:2239-56
- Leibson CL, Toteson ANA, Gabriel SE, et al. Mortality, disability, and nursing home use for persons with and without hip fracture: a population-based study. J Am Geriatr Soc 2002;50:1644-50
- Braithwaite RS, Col NF, Wong JB. Estimating hip fracture morbidity, mortality, and costs. J Am Geriatr Soc 2003;51:364-70
- Kim SH, Meehan JP, Blumenfeld T, et al. Hip fractures in the United States: Nationwide emergency department sample, 2008. Arthritis Care Res 2012;64:751-7
- Budhia S, Mikyas Y, Tang M, et al. Osteoporotic fractures: a systematic review of US healthcare costs and resource utilization. Pharmacoeconomics 2012;30:147-70
- Fried LP, Tangen CM, Walston J, et al. Frailty in older adults: evidence for a phenotype. J Gerontol A Biol Sci Med Sci 2001;56:M146-56
- Rosseau P. Immobility in the aged. Arch Fam Med 1993;2:169-78
- Bhave A, Marker DR, Seyler TM, et al. Functional problems and treatment solutions after total hip arthroplasty. J Arthroplasty 2007;22(6 Suppl 2):116-24
- Visser M, Harris TB, Fox KM, et al. Change in muscle mass and muscle strength after a hip fracture: relationship to mobility recovery. J Gerontol A Biol Sci Med Sci 2000;55:M434-40
- Wu N, Chen SY, Lee YC, et al. Demographics and clinical and economic characteristics of patients receiving total hip arthroplasty with and without muscle atrophy/weakness. Clinicoecon Outcomes Res 2013;5:271-80
- Deyo RA, Cherkin DC, Ciol MA. Adapting a clinical comorbidity index for use with ICD-9-CM administrative databases. J Clin Epidemiol 1992;45:613-19
- US Bureau of Labor Statistics. Consumer Price Index Detailed Report Tables, Annual Averages 2011. http://www.bls.gov/cpi. Accessed October 29, 2013
- Breiman L, Freidman JH, Olshen RA, et al. Classification and regression trees. New York: Chapman & Hall (Wadsworth, Inc.), 1984
- Wu X, Kumar V, Quinlan JR, et al. Top 10 algorithms in data mining. Knowledge and Information Systems 2008;14:1-37
- Pike CT, Birnbaum HG, Schiller M, et al. Prevalence and costs of osteoporotic patients with subsequent non-vertebral fractures in the US. Osteoporos Int 2011;22:2611-21
- Pike C, Birnbaum HG, Schiller M, et al. Direct and indirect costs of non-vertebral fracture patients with osteoporosis in the US. Pharmacoeconomics 2010;28:395-409
- Rizzoli R, Reginster JY, Arnal JF, et al. Quality of life in sarcopenia and frailty. Calcif Tissue Int 2013;93:101-20
- ESHRE Capri Workshop Group. Bone fractures after menopause. Hum Reprod Update 2010;16:761-73