Trends in the Utilization and Outcomes of Medicare Patients Hospitalized for Hip Fracture,2000-2008

Abstract

Objective: This study examines temporal trends in hip fracture related utilization and outcomes among elderly fee-for-service Medicare beneficiaries. Method: The study uses claims data for a 5% sample of Medicare beneficiaries with an incident hip fracture hospitalization between 2000 and 2008. We present annual mean patient characteristics, health services utilization, and outcomes and use ordinary least squares regressions to examine adjusted trends in utilization and outcomes after controlling for changes in patient characteristics. Results: We observe a statistically significant temporal decline in inpatient acute days and a statistically significant increase in inpatient post-acute days following hip fractures. In models that control for patient characteristics, we observe statistically significant declines in 1-year hip fracture readmission and mortality rates. Rates of nursing home residence 1-year following fracture were unchanged and remain high. Discussion: Hip fractures remain highly debilitating events and pose significant challenges for the financing of public health insurance programs.

Keywords

Medicare hip fracture utilization debility

Introduction

The entry of the baby-boom cohorts into Medicare, alongside continued gains in life expectancy, will pose significant budgetary challenges over the coming decades. Between 2010 and 2030, the Medicare population is expected to grow from 47 million to nearly 80 million (Cubanski, Voris, Kitchman, Neuman, & Potetz, 2005). Changes in the prevalence and management of chronic diseases, such as osteoporosis, are expected to play a significant role in future Medicare expenditure growth (Thorpe, Ogden, & Galactionova, 2010).

Even ahead of these challenging demographic trends, the societal burden of osteoporosis is substantial (Becker, Kilgore, & Morrisey, 2011). A 2002 report from the National Osteoporosis Foundation (2012) estimated that 10 million Americans suffered from the disease, resulting in 1.5 to 2 million fragility fractures annually. A sizable literature has documented the significant costs (Burge et al., 2007; Kilgore et al., 2009; Lonnroos et al., 2009) and adverse health consequences (Bliuc et al., 2009; Braithwaite, Col, & Wong, 2003; Center, Nguyen, Schneider, Sambrook, & Eisman, 1999; Cooper, 1997; Dolan et al., 2000; Lenze et al., 2007) associated with osteoporosis and fragility fractures. Estimates of the annual direct medical costs of these fractures range from 14 to 20 billion dollars (Braithwaite et al., 2003; Ray, Chan, Thamer, & Melton, 1997). These fractures are associated with large and persistent increases in mortality, with prior research documenting age standardized mortality ratios between 2 to 4 in the 5- to 10-year period following hip and vertebral fractures (Bliuc et al., 2009; Braithwaite et al., 2003).

This article examines trends in the characteristics, treatments, and outcomes of elderly Medicare beneficiaries hospitalized with incident hip fractures between 2000 and 2008. Although other studies have examined trends in utilization during the initial hip fracture hospitalization (Gehlbach, Avrunin, & Puleo, 2007; Nguyen-Oghalai et al., 2008) and utilization associated with total hip arthroplasty (Cram et al., 2011), this report provides a more comprehensive assessment of secular trends in health services utilization and outcomes associated with hip fractures in the elderly. Our analysis explores more detailed measures of post-fracture utilization, and uses multivariable regressions to account for changes in patient characteristics in examining secular trends in hip fracture utilization and outcomes.

Most hip fractures among the elderly are caused by falls and declining bone density associated with osteoporosis (Cummings, Kelsey, Nevitt, & O’Dowd, 1985; Melton et al., 1997). The term hip fracture refers to a diverse set of fractures of the femoral neck (transcervical hip fractures) and upper femoral shaft (petrochanteric hip fractures). Hip fractures are usually repaired surgically using internal fixation, partial hip replacement, or full hip replacement depending on the location and severity of the fracture. The invasiveness of these surgical repairs typically requires lengthy inpatient hospitalizations and significant utilization of rehabilitative care.

Our analysis is motivated by two underlying trends in the treatment of osteoporosis and fractures. The first is the sharp increase in the use of oral bisphosphonate drugs during our study period. Previous research has documented declines in overall hip fracture incidence, but an increased proportion of subtrochanteric (atypical) fractures coincident with increased use of bisphosphonate therapy (Wang & Bhattacharyya, 2011). We hypothesize that the increased pharmacological management of osteoporosis will lead to changes in the composition of hip fracture patients over time, including increased patient age and baseline severity, and a rising proportion of atypical fractures. We account for these changes in patient characteristics in regression models which examine trends in post-fracture service utilization and adverse health outcomes. The second factor of interest is the continued decline in acute length of stays and the increased reliance of post-acute care. Although the incentives for shorter lengths of stay are unchanged since the adoption of the inpatient prospective payment system (PPS) in the early 1980s (DesHarnais et al., 1987; Feder, Hadley, & Zuckerman, 1987; Freiman, Ellis, & McGuire, 1989; Morrisey, Sloan, & Valvona, 1988; Newhouse & Byrne, 1988), inpatient stays have continued to trend shorter (Baker, Einstadter, Husak, & Cebul, 2004; Bueno et al., 2010). Increases in per-diem reimbursements for skilled nursing facility care (SNF) under a revision to the SNF PPS may have additional implications for the mix of post-fracture service utilization and health outcomes (Grabowski, Afendulis, & McGuire, 2011). We hypothesize that reductions in acute inpatient length of stays will lead to increases in post-fracture mortality and readmissions.

Method

Cohort Construction

This study utilizes a 5% random sample of Medicare claims and enrollment data obtained from the Chronic Condition Data Warehouse (CCW). The CCW data provide longitudinal claims for all Medicare Part A and Part B services, including inpatient, SNF, outpatient, physician, home health, and hospice care. The accompanying annual beneficiary summary files provide information on monthly enrollment status, patient demographic characteristics, and validated death dates.

We use the CCW data to construct annual cohorts of Medicare beneficiaries who were hospitalized with an “incident” diagnosis of hip fracture in 2000-2008. Our case definition requires an acute care hospitalization with a primary International Classification of Disease (Version 9) diagnosis code of hip fracture (codes 820.X). We define incident hip fractures by excluding patients who were hospitalized with a primary or secondary diagnosis of hip fracture in the prior 365 days. This case definition allows patients to have more than one incident hip fracture during the 9-year study period. We further restrict our sample by excluding non-elderly (under age 65), non-U.S. residents, and individuals who were not continuously enrolled in traditional fee-for-service Medicare (Parts A and B) in the 12 months prior and subsequent (or deceased) to their incident hip fracture. As such, our analysis is specific to beneficiaries enrolled in “traditional” Medicare and may or may not generalize to the trends among Medicare Advantage enrollees.

Patient Characteristics

From the beneficiary summary file, we construct a set of indictor variables denoting patient characteristics including categorical age (65-69, 70-74, 75-79, 80-84, 85+), female gender, race (White non-Hispanic, Black non-Hispanic, Asian, Hispanic, and Other) and urban residence. We use the claims data to construct several measures of baseline patient health status at the time of the hip fracture hospitalization. First, we use claims data from the prior year to construct the Charlson comorbidity index (Charlson, Pompei, Ales, & MacKenzie, 1987). We use primary and secondary diagnosis codes from the inpatient and outpatient claims, and the original Charlson weights, to construct our comorbidity index. Second, we construct an indicator variable denoting whether the patient was admitted to an acute care hospital in the prior year. Finally, we use a recently developed claims-based algorithm to determine “long-term” nursing home residency status at the time of the incident hip fracture (Yun et al., 2010). Based on the algorithm, a patient is classified as a long-term nursing home resident at the time of the fracture if they had a physician/outpatient claim with a “nursing facility” place of service code during any of the prior 12 months, but no SNF claims during the same month period. The algorithm relies upon the assumption that individuals remain nursing home residents in all months subsequent to the initial month containing evidence of long-term nursing home residence.

Utilization/Health Outcomes Measures

We use the claims data to study treatment and utilization patterns during both the initial hospital admission and the 180 day period subsequent to the hip fracture hospitalization. We construct a series of variables which characterize the initial hip fracture hospitalization, including the patient’s principal diagnosis code, the primary hip fracture repair procedure, length of stay, and the discharge disposition. We group the International Classification of Disease, Ninth Revision (ICD-9) procedure codes and categorize patients as having received total hip replacement, partial hip replacement, reduction/internal fixation, other non-specific repairs, and no evidence of a hip fracture repair. Discharge disposition from the index hip fracture claim is grouped into categories for home, home health care, hospice, transfer to acute inpatient, SNF, intermediate care facility, other inpatient facility (primary inpatient rehabilitation facilities), in-hospital death, and other. These variables are designed to examine broad changes in initial hip fracture treatment during our study period.

Next, we use the claims data to construct measures of aggregate health care utilization in the 180-day period following the index hip fracture admission date. The use of this follow-up period is motivated by prior work (Kilgore et al., 2009) showing that monthly health care expenditures returned to pre-fracture levels approximately 6 months after the index fracture date. We examine total acute and post-acute (SNF, rehabilitation hospital and long-term care hospital) days, home health hours, and outpatient physician visits (with evaluation and management codes) in the 180-day post-fracture period. This allows us to examine changes in the mix of inpatient and post-acute services used in the treatment of elderly hip fracture patients.

We also use the claims and beneficiary summary files to examine trends in mortality, readmissions, and nursing home residency status subsequent to the index hip fracture admission date. We construct a series of indicator variables denoting all-cause mortality, hip fracture readmissions, and all-cause readmissions within 30, 90, 180, and 365 days of the initial hip fracture admission date. In addition to mortality and readmissions, we also construct a variable denoting nursing home residence 1-year after the fracture hospitalization using the algorithm described above.²⁶

Statistical Methods

Our analysis is designed to provide a comprehensive overview of secular trends in the patient characteristics, utilization, and health outcomes of elderly FFS Medicare beneficiaries hospitalized for hip fracture. We begin by reporting annual unadjusted means in patient characteristics, utilization, and outcomes across our 9-year study period. The utilization and outcome measures reported at various post-fracture intervals are unconditional upon survival and therefore reflect the average experience of all patients hospitalized for hip fracture in a given calendar year. The unadjusted changes in utilization and outcomes may reflect either real changes in treatments and quality, or changes in the underlying severity of hip fracture patients during our study period.

To account for changes in patient characteristics over time, we estimate ordinary least squares regressions that examine trends in key utilization and health outcome variables including index length of stay, total acute days within 180 days of index admission, total post-acute days within 180 days of index admission, 1-year hip fracture readmission, 1-year all-cause readmission, and 1-year mortality. We estimate models with and without adjustment for patient characteristics, using two alternative specifications of the time trend. In the first, we control for a linear time trend using a continuous year variable, while in the second we allow for a non-linear time trend using year dummy variables. In our adjusted models, we control for categorical age, race, gender, urban residence, Charlson comorbidity index, and indicators denoting an acute hospitalization in the prior year and nursing home residency status at the time of the index hospitalization. This study was approved by our university’s Institutional Review Board and the Centers for Medicare and Medicaid Services Privacy Board. All analyses were conducted using SAS software version 9.1.

Results

Table 1 presents annual baseline patient characteristics for the 2000 to 2008 hip fracture cohorts. Three notable trends emerge from the data. First, hip fractures are increasingly concentrated among the oldest old, with the percentage of patients over age 85 increasing from 44.4% in 2000 to 47.3% in 2008. Second, we observe a decrease in the percentage female over time, from 77.0% in 2000 to 74.4% in 2008. Finally, the percentage of hip fracture patients we classify as long-term nursing home residents at the time of their fracture declined sharply from 24.4% in 2000 to 18.9% in 2008. This finding is consistent with prior studies (Cutler, 2001; Feng, Fennell, Tyler, Clark, & Mor, 2011) and the growth in less intensive residential care options (e.g., assisted living facilities) for the elderly. Consistent with the trends in the age distribution of our cohorts, we observe modest increases in our measures of ex ante patient severity. The mean Charlson score increased from from 2.17 in 2000 to 2.55 in 2008.

Table 1.

Trends in Characteristics of FFS Medicare Beneficiaries Hospitalized for Hip Fracture (2000-2008).

	2000	2001	2002	2003	2004	2005	2006	2007	2008	Change 2000-2008
Age category (%)
Age 65-69	4.2	4.3	4.2	4.3	4.6	4.9	4.9	4.7	5.3	1.1
Age 70-74	9.7	8.9	9.0	9.1	9.1	9.0	8.9	8.7	8.5	−1.2
Age 75-79	17.4	16.9	17.2	16.8	16.9	17.0	16.5	15.8	14.9	−2.5
Age 80-84	24.3	24.5	24.8	24.8	25.3	26.0	24.9	24.4	23.9	−0.4
Age 85+	44.4	45.5	44.8	45.0	44.1	43.2	44.9	46.3	47.3	2.9
Female (%)	77.0	76.6	75.9	75.3	75.3	74.2	74.0	74.4	74.4	−2.6
Race/ethnicity (%)
White non-Hispanic	93.8	93.5	93.5	93.7	93.8	93.8	93.6	94.0	93.2	−0.6
Black non-Hispanic	3.6	3.7	3.7	3.7	3.5	3.5	3.4	3.0	3.5	−0.1
Asian	0.5	0.4	0.3	0.3	0.3	0.4	0.4	0.6	0.5	0.0
Hispanic	0.5	0.7	0.8	0.6	0.7	0.7	0.9	0.6	0.8	0.3
Other	0.8	1.0	1.0	1.1	1.1	1.1	1.2	0.8	1.2	0.4
Urban resident (%)	62.1	61.8	63.2	62.9	62.7	62.3	61.6	62.3	61.9	−0.2
Health status measures
Charlson index	2.17	2.17	2.27	2.29	2.38	2.38	2.53	2.57	2.55	0.38
Hospitalized prior 365d (%)	36.2	37.0	36.5	36.8	37.9	36.9	37.5	37.2	36.4	0.2
Nursing home resident (%)	24.4	23.7	22.7	22.4	20.8	19.9	20.3	19.6	18.9	−5.5
n	9,896	9,759	9,610	9,936	9,853	9,587	9,476	9,169	8,748

Note. We have excluded patients under 65 and patients who were not continuously enrolled in Medicare FFS (A + B) in the year prior to the hip fracture and non-decedents who did not remain continuously enrolled in Medicare FFS in the year following their hip fracture.

In Table 2, we report trends in a series of variables related to the initial hip fracture hospitalization. We are interested in whether changes in osteoporosis management, and the characteristics of hip fracture patients, are associated with changes in the incidence of various types of fractures, hip fracture repairs, length of stay, and discharge disposition. Consistent with prior work (Wang & Bhattacharyya, 2011), we document a relative decline in transcervical fractures (ICD-9 diagnosis codes 820.0x and 820.1x), and relative increases in unspecified (820.8 and 820.9) and petrochanteric fractures (820.2x and 820.3x), the latter of which is driven primarily by an increase in atypical fractures of the subtrochanteric region (ICD-9 diagnosis code 820.22). Although atypical fractures remain relatively rare, their share of total hip fractures increased from 3.2% in 2000 to 4.4% in 2008. We do not observe any clear changes in the distribution of hip fracture repair procedures over time, as the percentages of patients receiving partial hip replacements, total hip replacements, and reduction/internal fixation were roughly constant across our study period.

Table 2.

Trends in Diagnosis, Treatment, Length of Stay, and Discharge Disposition for Incident Hip Fracture Hospitalizations (2000-2008).

	2000	2001	2002	2003	2004	2005	2006	2007	2008	Change (%)
Hip fracture diagnosis type, N (%)
Transcervical (820.0x, 820.1x)	3,044 (30.8)	2,971 (30.4)	2,930 (30.5)	3,032 (30.5)	2,906 (29.5)	2,753 (28.7)	2,672 (28.2	2,618 (28.6)	2,307 (26.4)	−4.4
Petrochanteric (820.2x, 820.3x)	4,873 (49.2)	4,810 (49.3)	4,733 (49.3)	4,882 (49.1)	4,915 (49.9)	4,753 (49.6)	4,733 (49.9)	4,574 (49.9)	4,458 (51.0)	1.8
Other unspecified (820.8, 820.9)	1,979 (20.0)	1,978 (20.3)	1,947 (20.3)	2,022 (20.4)	2,032 (20.6)	2,081 (21.7)	2,071 (21.9)	1,977 (21.6)	1,983 (22.7)	2.7
Fracture repair type, N (%)
Total hip replacement	300 (3.0)	281 (2.9)	262 (2.7)	259 (2.6)	284 (2.9)	254 (2.6)	250 (2.6)	250 (2.7)	224 (2.6)	−0.4
Partial hip replacement	3,270 (33.0)	3,194 (32.7)	3,194 (33.2)	3,440 (34.6)	3,252 (33.0)	3,175 (33.1)	3,277 (34.6)	3,035 (33.1)	2,947 (33.7)	0.7
Reduction/internal fixation	5,817 (58.8)	5,780 (59.2)	5,634 (58.6)	5,660 (57.0)	5,752 (58.4)	5,613 (58.5)	5,444 (57.5)	5,375 (58.6)	5,083 (58.1)	−0.7
Other non-specific repair	151 (1.5)	140 (1.4)	156 (1.6)	154 (1.5)	167 (1.7)	160 (1.7)	151 (1.6)	147 (1.6)	132 (1.5)	0.0
No repair	358 (3.6)	364 (3.7)	364 (3.8)	423 (4.3)	398 (4.0)	385 (4.0)	354 (3.7)	362 (3.9)	362 (4.1)	0.5
Length of stay, mean days	6.52	6.51	6.57	6.54	6.39	6.22	6.22	6.20	6.06	−0.46
(SD)	(4.5)	(5.1)	(4.8)	(5.5)	(4.3)	(4.4)	(4.3)	(4.4)	(4.0)
Discharge disposition (%)
Home	7.8	6.7	5.4	4.0	2.8	2.5	2.8	3.2	2.8	−5.0
Home care	4.0	4.0	3.9	3.8	4.8	4.7	4.7	4.6	5.0	1.0
Hospice	0.1	0.2	0.4	0.5	0.8	0.8	1.0	1.1	1.5	1.4
Transfer (acute inpatient)	2.6	2.9	3.1	2.2	2.0	2.0	1.9	1.7	1.6	−1.0
SNF	55.2	54.7	55.6	57.3	59.8	59.9	60.3	60.3	61.4	6.2
ICF	5.8	5.9	4.2	3.2	1.4	1.3	1.2	1.1	1.0	−4.8
Other inpatient facility^a	21.4	22.4	23.6	25.3	25.2	25.7	24.9	24.7	23.7	2.3
In-hospital death	3.1	3.3	3.8	3.8	3.2	2.9	3.0	3.2	2.8	−0.3
Other	0.0	0.0	0.1	0.1	0.1	0.2	0.2	0.2	0.2	0.2

Note. SNF = skilled nursing facility; ICF = intermediate care facility.

Includes inpatient rehabilitation facility, long-term care hospital, federal hospital, psychiatric hospital, critical access hospital, and other inpatient facilities.

Table 2 also shows trends in length of stay and discharge disposition from the initial hip fracture hospitalization. There is a steady decline in mean length of stay from 6.52 days in 2000 to 6.06 days in 2008. Although hospitals have incentives to restrict length of stay under Medicare’s PPS, these incentives did not change appreciably during our study period. Instead, the declines in length of stay may reflect continued changes in patterns of post-acute care. Across the full 9-year study period, approximately 97% of hip fracture patients are discharged alive. The fraction of patients discharged to home without any formal post-acute care fell from 7.7% in 2000 to just 2.8% in 2008. This decline is partially offset by increases in the fraction discharged to home with home health or hospice care. An increasing majority of hip fracture patients are discharged to inpatient post-acute facilities—including SNFs, inpatient rehabilitation hospitals, long-term care hospitals, and swing beds. Between 2000 and 2008, the fraction of hip fracture patients discharged to SNF care (including swing beds) increased from 55.2% in 2000 to 61.4%. The fraction of patients discharged to other inpatient facilities (primarily inpatient rehabilitation hospitals) also increased from 21.4% to 23.7% during our study period. These increases were offset by a sharp decline in the percentage of patients discharged to intermediate care facilities from 5.8% in 2000 to 1.0% in 2008.

To assess changes in the mix of health care services associated with hip fractures, Table 3 presents trends in utilization during the 180-day period subsequent to the index fracture admission date. Average total inpatient acute days declined from 9.63 in 2000 to 9.22 days in 2008. This decline (−0.41) is smaller than the decrease during the index hip fracture hospitalization (−0.46), indicating that re-hospitalization rates increased during our study period. We find evidence suggesting a continued shift of post-fracture management from the inpatient acute setting to all forms of post-acute care. Total inpatient post-acute days increased from 30.6 in 2000 to 40.7 in 2008. This increase is driven by longer inpatient post-acute stays (the intensive margin), as the fraction of patients using inpatient post-acute care (the extensive margin) increased only slightly during our study period. The increase in post-acute days was concentrated in SNFs, as conditional mean days for both inpatient rehabilitation facilities and long-term care hospitals declined between 2000 and 2008. In addition to the increase in inpatient post-acute care we also observe increases in the use of home health care (13.7 hr in 2000 vs. 15.9 in 2008) and outpatient physician visits (6.2 in 2000 vs. 8.3 in 2008) in the 180-day period following hip fracture hospitalizations.

Table 3.

Trends in Total Utilization in the 180 day period subsequent to Hip Fracture (2000-2008).

	2000	2001	2002	2003	2004	2005	2006	2007	2008	2000-2008 change
Acute length of stay, mean days	9.6	9.5	9.7	9.8	9.6	9.5	9.5	9.5	9.2	−0.4
(SD)	(9.1)	(9.6)	(9.6)	(9.6)	(9.5)	(9.3)	(9.3)	(9.8)	(9.2)
Post-acute length of stay, mean	30.6	31.9	33.4	34.8	35.4	35.9	38.5	40.8	40.7	10.3
(SD)	(31.0)	(31.5)	(32.5)	(33.8)	(34.0)	(34.0)	(37.5)	(41.0)	(40.3)
SNF days, mean	26.8	28.1	29.5	30.8	31.4	31.8	34.7	37.0	37.1	10.3
(SD)	(31.0)	(31.7)	(32.8)	(34.0)	(34.2)	(34.4)	(37.9)	(41.4)	(40.6)
IRF days, mean	3.2	3.2	3.2	3.2	3.4	3.4	3.2	3.2	3.1	0.1
(SD)	(7.7)	(7.3)	(7.0)	(7.0)	(7.0)	(7.0)	(6.6)	(6.6)	(6.5)
Other post-acute days, mean	0.6	0.6	0.7	0.7	0.6	0.6	0.5	0.6	0.6	0.0
(SD)	(5.7)	(5.6)	(5.3)	(5.4)	(4.8)	(5.0)	(4.5)	(4.8)	(4.6)
Home health hours, mean	13.7	12.0	13.2	12.6	14.0	14.0	14.3	15.5	15.9	2.2
(SD)	(26.5)	(21.7)	(23.6)	(22.6)	(24.0)	(23.9)	(24.8)	(29.5)	(26.9)
Outpatient physician visits	6.2	6.4	6.7	7.0	7.4	7.6	7.9	8.1	8.3	2.1
(SD)	(5.5)	(5.5)	(5.9)	(6.1)	(6.4)	(6.7)	(7.1)	(7.1)	(7.2)
N of hospitalizations	9,896	9,759	9,610	9,936	9,853	9,587	9,476	9,169	8,748

Note. SNF = skilled nursing facility; IRF = intermediate scan facility.

Table 4 presents trends in health outcomes at various intervals subsequent to the initial hip fracture hospitalization. Mortality at 30, 90, and 180 days is essentially unchanged between 2000 and 2008. Although, 1-year mortality is 1.3 percentage points lower in 2008 than in 2000, there is no evidence of a consistent downward trend over our study period. We observe more pronounced trends in readmission rates after hip fractures. In all time periods subsequent to the initial hospitalization, we observe declining rates of readmission with a primary diagnosis of hip fracture. Hip fracture readmissions fall by 0.5 percentage points at 30-days (from 1.2% to 0.7%) and by 1.4 percentage points at 1-year (8.4% to 7.0%). In contrast, all-cause readmission rates increased throughout our study period. The increase was driven primarily by readmissions during the first 30 days, where the all-cause readmission rate increased from 11.3% in 2000 to 12.4% in 2008. The 1-year cumulative readmission rate rose from 43.2% in 2000 to 44.2% in 2008. Despite the secular decline in the fraction of patients who resided in a nursing home at the time of their incident hip fracture (Table 1), the percentage of patients who were long-term nursing home residents 1-year after their fracture remained roughly constant over the course of our study period (34.8% in 2008 vs. 35.2% in 2000).

Table 4.

Trends in Mortality, Readmissions and Nursing Home Residency Subsequent to Hip Fractures (2000-2008).

	2000	2001	2002	2003	2004	2005	2006	2007	2008	2000-2008 change
Mortality (%)
30 days	7.9	8.1	8.9	8.9	7.9	7.9	7.9	8.0	8.1	0.2
90 days	15.3	15.9	17.0	16.9	15.2	15.8	15.8	15.8	15.2	−0.1
180 days	20.8	21.2	22.8	22.6	21.1	21.4	21.8	21.8	20.8	0.0
365 days	29.4	28.4	30.2	30.2	28.4	28.5	28.8	29.2	28.1	−1.3
Hip fracture readmission^a (%)
30 days cumulative	1.2	0.8	1.0	1.1	0.9	0.8	0.7	0.7	0.7	−0.5
90 days cumulative	3.1	2.4	2.9	2.9	2.6	2.1	2.4	2.2	2.0	−1.1
180 days cumulative	5.4	4.9	5.4	5.3	4.9	4.2	4.2	4.5	4.1	−1.3
365 days cumulative	8.4	8.0	8.7	8.7	8.2	7.1	7.4	7.1	7.0	−1.4
All-cause readmission^a (%)
30 days cumulative	11.3	10.5	11.4	12.3	11.8	12.3	12.4	12.7	12.4	1.1
90 days cumulative	23.4	23.0	24.3	24.8	24.6	24.9	25.9	25.1	24.7	1.3
180 days cumulative	32.1	31.5	33.2	33.8	33.4	34.1	34.5	33.8	33.9	1.8
365 days cumulative	43.2	43.0	44.2	45.0	44.8	45.2	46.0	44.7	44.2	1.0
Nursing home residency (%)
1-year post-fracture	35.2	34.6	33.7	34.3	34.5	34.4	35.2	35.0	34.8	−0.4
n of hospitalizations	9,896	9,759	9,610	9,936	9,853	9,587	9,476	9,169	8,748

Readmission variables are cumulative and require admission to an acute care hospital in the period specified.

Figure 1 provides a graphical summary of the trends in our primary utilization and health outcome measures. Panel A highlights the 0.46 day decline in the index length of stay between 2000 and 2008. Panel B shows evidence of a similar decline in mean acute days within 180 days of the index hospitalization, but a large increase in mean post-acute days (10 days) over the course of our study period. In Panel C, we observe the divergent trends in hip fracture and all-cause readmissions with rates of hip fracture readmissions declining by over 1 percentage point while all-cause readmissions increased by a approximately 1 percentage point. Finally, Panel D demonstrates that patient disposition following hip fractures was little changed during our study period. We observe a modest reduction in 1-year mortality despite an increase in patient severity over time.

Figure 1.

Trends in key outcome variables (2000-2008).

In Table 5, we examine the role of patient characteristics in our observed trends in post-fracture utilization and health outcomes. The left-panel of the table presents the results from unadjusted regressions using two alternative specifications of the time trend. In Model 1, with a continuous year variable, we observe statistically significant declines in index length of stay, 180-day acute days, 1-year hip fracture readmission and 1-year mortality and statistically significant increases in 180-day post-acute days and 1-year all-cause readmission. Model 2, with year dummy variables reveals evidence of the non-linearity in these trends that was evident in Figure 1. With the exception of post-acute days which increase consistently throughout the study period, the trends reported in model 1 are non-linear and are primarily driven by changes in 2004 and beyond. The adjusted regressions, which include controls for patient characteristics, yield relatively similar results. In the linear specification, controlling for the worsening baseline health of hip fracture patients yields evidence of larger declines in index length of stay, 180-day acute inpatient days, and 1-year mortality. The increase in all-cause readmission observed in the unadjusted models is no longer statistically significant after controlling for patient characteristics. As in the adjusted models our trends in utilization and outcomes are more pronounced in the latter half of our study period.

Table 5.

Results From Regressions Examining Trends in Select Utilization and Outcome Measures.

	Unadjusted models						Adjusted models
	Index LOS	180d acute days	180d post-acute days	1-year hip fracture readmit	1-year all-cause readmit	1-year mortality	Index LOS	180d acute days	180d post-acute days	1-year hip fracture readmit	1-year all-cause readmit	1-year mortality
Model 1: Linear trend
Year	−0.063**	−0.040**	1.309**	−0.204**	0.227**	−0.119*	−0.076**	−0.074**	1.365**	−0.199**	0.088	−0.283**
	(0.006)	(0.012)	(0.047)	(0.060)	(0.066)	(0.060)	(0.006)	(0.012)	(0.046)	(0.060)	(0.066)	(0.057)
Model 2: Nonlinear time trend
Year 2001	−0.094	−0.130	1.350**	−0.405	−0.111	−1.011	−0.023	−0.164	1.299**	−0.415	−0.314	−1.201
	(0.066)	(0.128)	(0.500)	(0.643)	(0.709)	(0.647)	(0.065)	(0.125)	(0.492)	(0.643)	(0.699)	(0.613)
Year 2002	0.054	0.074	2.861**	0.374	1.003	0.863	0.016	−0.019	2.923**	0.380	0.620	0.504
	(0.066)	(0.128)	(0.502)	(0.646)	(0.712)	(0.650)	(0.065)	(0.126)	(0.494)	(0.646)	(0.702)	(0.615)
Year 2003	0.019	0.175	4.199**	0.309	1.799**	0.807	−0.027	0.061	4.298**	0.320	1.326*	0.321
	(0.065)	(0.127)	(0.498)	(0.640)	(0.706)	(0.645)	(0.065)	(0.125)	(0.490)	(0.640)	(0.697)	(0.610)
Year 2004	−0.131*	−0.033	4.876**	−0.197	1.670**	−0.958	−0.199*	−0.213*	5.153**	−0.209	0.858	−1.469**
	(0.066)	(0.128)	(0.499)	(0.642)	(0.707)	(0.646)	(0.065)	(0.125)	(0.491)	(0.642)	(0.698)	(0.612)
Year 2005	−0.297**	−0.172	5.335**	−1.243*	1.996**	−0.837	−0.371**	−0.357**	5.797**	−1.244*	1.228*	−1.177*
	(0.066)	(0.128)	(0.502)	(0.646)	(0.712)	(0.650)	(0.065)	(0.126)	(0.495)	(0.646)	(0.703)	(0.616)
Year 2006	−0.293**	−0.129	7.928**	−0.959	2.873**	−0.608	−0.389**	−0.371**	8.263**	−0.938	1.842**	−1.753**
	(0.066)	(0.129)	(0.504)	(0.648)	(0.714)	(0.652)	(0.066)	(0.126)	(0.496)	(0.649)	(0.705)	(0.618)
Year 2007	−0.320**	−0.155	10.227**	−1.245*	1.568**	−0.212	−0.420**	−0.402**	10.522**	−1.225*	0.504**	−1.528**
	(0.067)	(0.130)	(0.508)	(0.654)	(0.720)	(0.658)	(0.066)	(0.128)	(0.501)	(0.654)	(0.712)	(0.624)
Year 2008	−0.459**	−0.415**	10.182**	−1.394**	1.023	−1.266**	−0.564**	−0.666**	10.520**	−1.349**	0.000	−2.539**
	(0.068)	(0.132)	(0.515)	(0.662)	(0.729)	(0.666)	(0.067)	(0.129)	(0.507)	(0.663)	(0.721)	(0.631)

Note. All models are estimated using ordinary least squares. Coefficients from readmission and mortality models are multiplied by 100 to yield a percentage point interpretation.

Denotes statistical significance at the 5% level. **Denotes statistical significance at the 1% level.

Discussion

This article provides a comprehensive assessment of trends in the characteristics, health services utilization and outcomes of traditional FFS Medicare beneficiaries hospitalized for hip fractures. The proportion of hip fracture patients who were older (above age 85), male and community dwelling all increased during our study period. This changing profile of hip fracture patients likely reflects some combination of increased longevity, improved osteoporosis management among women, and the growth of Medicare Advantage enrollment during our study period (Gold, Phelps, Neuman, & Jacobson, 2010). Consistent with previous work, we observe an increase—both in a relative and absolute sense—in subtrochanteric fractures.

During our study period, the mix of health care services associated with hip fractures continued to shift from acute inpatient to post-acute care. Lengths of stay for the initial hip fracture hospitalization declined, and the fraction of patients discharged to SNFs and other inpatient post-acute care increased. We observe similar trends in the 6-month period subsequent to the index hip fracture admission date, with declines in total inpatient acute days, but increases in post-acute inpatient days, home health care, and outpatient physician visits. The increased reliance of post-acute care may be related to the declining share of hip fracture patients who resided in nursing homes at the time of their fracture. These trends are likely to continue with the entry of the baby-boom cohorts into Medicare and continued declines in traditional nursing-home residency among Medicare beneficiaries.

We find some evidence of improved hip fracture outcomes during our study period. Although 1-year mortality declined between 2000 and 2008, there was considerable year-to-year variability in our estimated mortality rates. We observe a steady decline in hip fracture readmission rates in all periods subsequent to the index admission. However, all-cause readmission rates actually increased over our 9-year study period, with the majority of this increase occurring within 30 days of the initial hip fracture hospitalization. Using a recently developed claims-based algorithm for identifying long-term nursing home residence we also examine changes in patient disposition status 1 year after the index hip fracture hospitalization. Despite an increase in the fraction of patients living in the community at the time of their index hip fracture, we do no observe a significant change in the fraction of patients classified as long-term nursing home residents 1 year after their hip fracture. This finding suggests that despite declines in traditional nursing home residency among Medicare beneficiaries, hip fractures remain highly debilitating events. The aging of the baby boomers and resultant growth of the Medicare population will pose enormous challenges for public health care financing, particularly in the area of long-term care.

There are several important limitations to this study. First, our analysis is limited to patients enrolled in traditional FFS Medicare. Medicare Advantage enrollment increased significantly during the latter half of our study period as a result of changes in payment generosity and risk adjustment. The fraction of beneficiaries enrolled in Medicare Advantage plans which had declined from 17% in 2000 to 13% in 2003, increased to 22% in 2008 (Gold et al., 2010). Although the recent changes in Medicare policy were designed to reduce the degree of favorable selection into these plans, work by Morrisey, Kilgore, Becker, Smith, and Delzell (2013) finds only small, nonstatistically significant reductions in favorable selection. A second limitation is that our analysis is reliant upon the 5% sample of Medicare claims. The claims-based algorithm we use to determine nursing home residency algorithm has been externally validated against the Medicare Current Beneficiary Survey but is inferior to more sophisticated methods which augment claims with information from the Minimum Data Set (Intrator, Hiris, Berg, Miller, & Mor, 2011). The limited sample sizes in the 5% data also restrict our ability to establish definitive trends in some important post-fracture outcomes, including mortality rates. Finally, our analysis of hip fracture utilization focuses exclusively on Medicare covered services. Although we explore transitions to long-term nursing home residency subsequent to fractures, we do not address the additional public sector burden associated with long-term care.

This study was designed to provide a descriptive overview of trends in hip fracture treatments among traditional FFS Medicare beneficiaries. Although a range of interesting questions surround the treatment of osteoporosis and hip fractures, our analysis suggests two specific areas where additional research could be particularly valuable. First, echoing recent work on other acute conditions in Medicare (Bueno et al., 2010), we observe a trend toward shorter acute lengths of stay and increased all-cause readmission rates, particularly in the 30-days following the hip fracture hospitalization. The rise in all-cause readmissions, despite a decline in the hip fracture specific readmission rate, raises interesting questions. What types of hospitalizations are driving this increase in the all-cause readmission rate? Is this shift in away from hip fracture readmissions the result of actual changes in post-fracture complications, or changes in coding behavior on the part of providers? Second, although hip fracture patients have become less likely to reside in nursing homes at the time of the fracture, rates of nursing home residence subsequent to hip fractures have remained high. The costs of long-term care represent a critical component of the burden of osteoporosis, and additional work is necessary to assess the public sector costs associated with the debility and destitution associated with these fractures.

Footnotes

Declaration of Conflicting Interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported through a research contract with Amgen. All coauthors supported through a research contract with Amgen. Additional financial involvements: Kilgore: Consultant (Eli Lilly), Curtis: Research contracts (Eli Lilly, Merck), Consultancy (Amgen, Eli Lilly, Merck) Saag: Research contracts (Eli Lilly, Merck), Consultancy (Amgen, Eli Lilly, Merck).

References

Baker

D. W.

Einstadter

Husak

S. S.

Cebul

R. D.

(2004). Trends in postdischarge mortality and readmissions: Has length of stay declined too far? Archives of Internal Medicine, 164, 538-544.

Becker

D. J.

Kilgore

M. L.

Morrisey

M. A.

(2011). The societal burden of osteoporosis. Current Rheumatology Reports, 12, 186-191.

Bliuc

Nguyen

N. D.

Milch

V. E.

Nguyen

T. V.

Eisman

J. A.

Center

J. R.

(2009). Mortality risk associated with low-trauma osteoporotic fracture and subsequent fracture in men and women. Journal of the American Medical Association, 301, 513-521.

Braithwaite

R. S.

Col

N. F.

Wong

J. B.

(2003). Estimating hip fracture morbidity, mortality and costs. Journal of the American Geriatrics Society, 51, 364-370.

Bueno

Ross

J. S.

Wang

Chen

Vidan

M. T.

Normand

S. L.

. . . Krumholz

H. M.

(2010). Trends in length of stay and short-term outcomes among Medicare patients hospitalized for heart failure, 1993-2006. Journal of the American Medical Association, 303, 2141-2147.

Burge

Dawson-Hughes

Solomon

D. H.

Wong

J. B.

King

Tosteson

(2007). Incidence and economic burden of osteoporosis-related fractures in the United States, 2005-2025. Journal of Bone and Mineral Research, 22, 465-475.

Center

J. R.

Nguyen

T. V.

Schneider

Sambrook

P. N.

Eisman

J. A.

(1999). Mortality after all major types of osteoporotic fracture in men and women: An observational study. Lancet, 353, 878-882.

Charlson

M. E.

Pompei

Ales

K. L.

MacKenzie

C. R.

(1987). A new method of classifying prognostic comorbidity in longitudinal studies: Development and validation. Journal of Chronic Diseases, 40, 373-383.

Cooper

(1997). The crippling consequences of fractures and their impact on quality of life. American Journal of Medicine, 103, 12S-17S.

10.

Cram

Kaboli

P. J.

Vaughan-Sarrazin

M. S.

Cai

Wolf

B. R.

(2011). Clinical characteristics and outcomes of Medicare patients undergoing total hip arthroplasty, 1991-2008. Journal of the American Medical Association, 305, 1560-1567.

11.

Cubanski

Voris

Kitchman

Neuman

Potetz

(2005). Medicare chartbook (3rd ed., The Henry J. Kaiser Family Foundation). Retrieved from http://www.kff.org/medicare/upload/Medicare-Chart-Book-3rd-Edition-Summer-2005-Report.pdf

12.

Cummings

S. R.

Kelsey

J. L.

Nevitt

M. C.

O’Dowd

K. J.

(1985). Epidemiology of osteoporosis and osteoporotic fractures. Epidemiology Review, 7, 178-208.

13.

Cutler

D. M.

(2001). Declining disability among the elderly. Health Affairs, 20, 11-27.

14.

DesHarnais

Kobrinski

Chesney

Long

Ament

Fleming

(1987). The early effects of the prospective payment system on inpatient utilization and the quality of care. Inquiry, 24, 7-16.

15.

Dolan

M. M.

Hawkes

W. G.

Zimmerman

S. I.

Morrison

R. S.

Gruber-Baldini

A. L.

Hebel

J. R.

Magaziner

(2000). Delirium on hospital admission in aged hip fracture patients: Prediction of mortality and 2-year functional outcomes. Journals of Gerontology Series A: Biological and Medical Sciences, 55, M527-M534.

16.

Feder

Hadley

Zuckerman

(1987). How did Medicare’s prospective payment system affect hospitals? New England Journal of Medicine, 317, 867-873.

17.

Feng

Fennell

M. L.

Tyler

D. A.

Clark

Mor

(2011). Growth of racial and ethnic minorities in US nursing homes driven by demographics and possible disparities in options. Health Affairs, 30, 1358-1365.

18.

Freiman

M. P.

Ellis

R. P.

McGuire

T. G.

(1989). Provider response to Medicare’s PPS: Reductions in length of stay for psychiatric patients treated in scatter beds. Inquiry, 26, 192-201.

19.

Gehlbach

S. H.

Avrunin

J. S.

Puleo

(2007). Trends in hospital care for hip fractures. Osteoporosis International, 18, 585-591.

20.

Gold

Phelps

Neuman

Jacobson

(2010). Medicare advantage 2010 data spotlight: Plan enrollment patterns and trends. Washington, DC: Kaiser Family Foundation. Retrieved at http://www.kff.org/medicare/upload/8080.pdf

21.

Grabowski

D. C.

Afendulis

C. C.

McGuire

T. C.

(2011). Medicare prospective payment and the volume and intensity of skilled nursing facility services. Journal of Health Economics, 30, 675-684.

22.

Intrator

Hiris

Berg

Miller

S. C.

Mor

(2011). The residential history file: Studying nursing home residents’ long-term care histories. Health Services Research, 46, 120-137.

23.

Kilgore

M. L.

Morrisey

M. A.

Becker

D. J.

Gary

L. C.

Curtis

J. R.

Saag

K. G.

… Delzell

(2009). Patterns of health care expenditures relative to skeletal fractures among Medicare beneficiaries, 1999-2005. Journal of Bone and Mineral Research, 24, 2050-2055.

24.

Lenze

E. J.

Munin

M. C.

Skidmore

E. R.

Dew

M. A.

Rogers

J. C.

Whyte

E. M.

. . . Reynolds

C. F.

III . (2007). Onset of depression in elderly persons after hip fracture: Implications for prevention and early intervention of late-life depression. Journal of the American Geriatrics Society, 55, 81-86.

25.

Lonnroos

Kautiainen

Sund

Karppi

Hartikainen

Kiviranta

Sulkava

(2009). Utilization of inpatient care before and after hip fracture: A population-based study. Osteoporosis International, 20, 937-941.

26.

Melton

L. J.

III Thamer

Ray

N. F.

Chesnut

C. H.

III Einhorn

T. A.

Johnston

C. C.

. . . Siris

E. S.

(1997). Fractures attributable to osteoporosis: Report from the national osteoporosis foundation. Journal of Bone and Mineral Research, 12, 16-23.

27.

Morrisey

M. A.

Kilgore

M. L.

Becker

D. J.

Smith

Delzell

(2013). Favorable selection, risk adjustment, and the Medicare Advantage program. Health Services Research, 48, 1039-1056. doi:10.1111/1475-6773.12006.

28.

Morrisey

M. A.

Sloan

F. A.

Valvona

(1988). Medicare prospective payment and posthospital transfers to subacute care. Medical Care, 26, 685-698.

29.

National Osteoporosis Foundation. (2012). America’s bone health: The state of osteoporosis and low bone mass in our nation. Washington, DC: Author.

30.

Newhouse

J. P.

Byrne

D. J.

(1988). Did Medicare’s prospective payment system cause length of stay to fall? Journal of Health Economics, 7, 413-416.

31.

Nguyen-Oghalai

T. U.

Kuo

Y. F.

Zhang

D. D.

Graham

J. E.

Goodwin

J. S.

Ottenbacher

K. J.

(2008). Discharge setting for patients with hip fracture: Trends from 2001 to 2005. Journal of the American Geriatrics Society, 56, 1063-1068.

32.

Ray

N. F.

Chan

J. K.

Thamer

Melton

L. J.

III . (1997). Medical expenditures for the treatment of osteoporotic fractures in the United States in 1995: Report for the National Osteoporosis Foundation. Journal of Bone and Mineral Research, 12, 24-35.

33.

Thorpe

K. E.

Ogden

L. L.

Galactionova

(2010). Chronic conditions account for rise in Medicare spending from 1987 To 2006. Health Affairs, 29, 718-724.

34.

Wang

Bhattacharyya

(2011). Trends in incidence of subtrochanteric fragility fractures and bisphosphonate use among the US elderly, 1996-1997. Journal of Bone and Mineral Research, 26, 553-560.

35.

Yun

Kilgore

M. L.

Curtis

J. R.

Delzell

Gary

L. C.

Saag

K. G.

. . . Locher

J. L.

(2010). Identifying types of nursing facility stays using Medicare claims data: An algorithm and validation. Health Services and Outcomes Research Methodology, 10, 100-110.