Abstract
A recent study suggested a decline in the incidence of hip fracture among U.S. women between 2002–2012, followed by a leveling in the incidence rate from 2013–2015. Newly admitted nursing home residents are particularly vulnerable to hip fracture, and it is unclear whether that trend is observed in this high risk group. The purpose of our study was to describe trends in hip fracture rates and post-fracture mortality among 2.6 million newly admitted U.S. nursing home residents from 2007 to 2015, and to examine whether these trends could be explained by differences in resident characteristics. Medicare claims data were linked with the Minimum Data Set (MDS), a clinical assessment performed quarterly on all nursing home residents. In each year (2007–2015), we identified newly admitted long-stay (i.e., 100 days in the same facility) NH residents. Hip fracture was defined using Part A diagnostic codes. Follow-up time was calculated from the index date until the first event of hospitalized hip fracture, Medicare disenrollment, death, or until one year. Poisson regression was used to adjust rates of hip fracture for age and sex. The number of newly admitted nursing home residents ranged from 324,508 in 2007 to 257,350 in 2015. Although mean age remained similar (83 years), residents were more functionally dependent over time. There was a small absolute decrease in the incidence rate of hip fracturebetween 2007 (3.32/100 person-years) and 2013 (2.92/100 person-years), with an increase again in 2015 (3.03/100 person-years). Adjusting for patient characteristics partially attenuated these trends. One-year mortality was high following fracture in all years (42.6% in 2007, 42.1% in 2014). In summary, we observed a recent slight rise in the incidence rates of hip fracture among nursing home residents that was at least partially explained by differences in resident characteristics over time.
Introduction:
Each year in the United States, nearly 300,000 persons experience a hip fracture.(1) Rates of hip fracture are particularly high in nursing home residents, given the high burden of cognitive impairment, co-morbidities, and polypharmacy. Newly admitted nursing home residents are the most vulnerable to hip fracture, with previously reported rates of 2.7/100 person-years (PY).(2) Thirty-six percent of nursing home residents with hip fracture will die within 6-months, and another 17.3% of ambulatory residents will become completely disabled.(3) Among survivors, infections and pressure ulcers are common,(4, 5) leading to functional decline,(6, 7) and a diminished quality of life.(7)
Given the high morbidity, mortality, and financial expense associated with these fractures, hip fractures are a major public health concern. It is important then to characterize temporal trends of hip fractures to inform interventions and national policies aimed at ameliorating these fractures. A recent population-based study suggested a decline in the incidence of hip fracture among U.S. women between 2002 and 2012, followed by a plateau in the incidence rate from 2013–2015.(8) The plateau corresponded to changes in osteoporosis screening and treatment rates in the community, and it is likely that the temporal changes in management contributed to these trends. However, this study lacked information on clinical characteristics related to fracture, such as cognitive and functional status, and it is possible that temporal differences in the populations’ clinical characteristics and vulnerability to fracture could importantly contribute to these trends. It is also not clear whether these trends apply to nursing home residents, who have appreciably different, and higher, risks for fracture as compared with community-dwellers.
Using a large sample of nearly 3 million newly admitted nursing home residents with clinical characteristics from the Minimum Data Set (MDS) linked with administrative claims data, we investigated trends in hip fracture rates annually from 2007 to 2015, and conditional on finding trends over time, we examined whether the trends were attenuated by differences in patient characteristics over time. Further, we examined trends in 1-year mortality following a hip fracture in the same cohort of newly admitted nursing home residents.
Methods:
Study Design and Data Source:
This was a historiccohort study using ten consecutive years (2007–2016) of national Medicare claims data (Part A) linked with nursing home MDS assessments.(9) The study was approved by the Hebrew SeniorLife Institutional Review Board.
Study Population:
For our source population, a 100% sample of Medicare Part A claims was linked to MDS assessments for all nursing home residents 65 years or older who were enrolled in fee-for-service Medicare.(10) We excluded residents enrolled in a Medicare Advantage plan, (approximately 20% of nursing home residents) for which claims data on hip fractures is unavailable. This is routinely done in Medicare claims studies in order to avoid misclassification.
For each year of the study (2007–2015), we created a cohort of long-stay nursing home residents who fulfilled the following criteria: 1) admission to a skilled nursing facility during the year of observation; and 2) spent no more than 10 consecutive days outside the skilled nursing facility during the first 100 days. The first day of qualification as a long-stay resident (e.g., 100 days) was designated as the index date. We focused on newly admitted nursing home residents in a given calendar year because of the known high rates of fracture in this group, and to create a cohort in each calendar year with similar opportunity for fracture. Figure 1 illustrates the timeline of enrollment and follow-up for a representative study patient.
Figure 1.
Timeline of a representative patient in a study of hip fracture rates in nursing home residents (2007–2015)
Hip fracture:
The primary outcome was the incidence rate of hip fracture during each year of observation, starting follow-up at the index date. Hip fractures were ascertained using Medicare Part A claims data and were defined as a hospitalization with an International Classification of Diseases, 9th edition (ICD-9) diagnosis code of 820.xx or 733.14 in any position, with or without an accompanying procedural code during the period spanning 2006 to September 2015. The estimated positive predictive value using this definition ranged has been reported to be 89.9% when the fracture code was in a secondary position and 96.7% when the hip fracture diagnosis code was in the primary position.(11) Beginning in October 2015, hip fractures were identified in claims data using equivalent ICD-10 codes (S72.x, S79.0, M80.05, and M84.45). As a secondary definition of hip fracture used in a sensitivity analysis, we considered both diagnostic codes for hip and femoral shaft fracture. Only the first encounter for a hip fracture was included in our analysis.
We identified history of hip fracture in the 180 days before index date using Medicare claims. While patients with a history of fracture were not excluded, we required > 100 days between a fracture that occurred before the index date and a fracture claim during follow-up in order to consider a new, incident hip fracture
Mortality:
Date of death was obtained from the Master Beneficiary Summary File from the Centers for Medicare & Medicaid Services.
Covariates:
We included the following covariates in our study: age, sex, race/ethnicity, functional status, cognitive status, ability to transfer, and history of hip fracture. We obtained demographic information on age, sex, and self-reported race from the Medicare Enrollment file. Residents were grouped into the following age categories, measured at the index date: 65–69 years, 70–74 years, 75–79 years, 80–84 years, 85–89 years, or 90 years or greater. Race/ethnicity was categorized as White, Black, Hispanic, Asian, or other.(12)
Information on clinical characteristics (functional status, cognitive status, and ability to transfer) was extracted from the MDS assessment most proximal to and before the index date. The MDS is a standardized instrument that contains comprehensive clinical information, including measures of cognitive and physical function, health conditions, falls, and psychosocial well-being.(13) Based on information from the medical record and interviews with nursing home residents, facility staff, and direct caregivers, assessments are conducted upon admission and quarterly thereafter. The MDS Version 2.0 was utilized from 2006, until the federally-mandated transition to Version 3.0 in October 2010.
Physical functioning in both MDS versions was evaluated with the 28-point MDS Activities of Daily Living (ADL) Long Form Scale (14) which measures resident self-performance in each of seven ADL domains: 1) locomotion in the facility, 2) dressing, 3) personal hygiene, 4) toilet use, 5) transfers, 6) bed mobility, and 7) eating. Higher scores indicate greater levels of functional dependence, and the scale ranges from 0 to 28. Transfer ability, defined as the resident’s level of dependence on staff necessary to move safely between surfaces (e.g. bed to chair) was categorized as either independent, requires supervision, requires limited staff assistance, or requires full assistance/cannot transfer independently.
From 2006 to 2010, cognitive function was ascertained with the Cognitive Performance Scale (CPS), a validated subscale in the MDS Version 2.0. (15) However, when the MDS was revised in 2010, the CPS was replaced by a more direct cognitive instrument, the Brief Interview of Mental Status (BIMS). The Cognitive Function Scale (CFS) is a subscale of the revised MDS assessment including the BIMS that permits harmonization of cognitive measures across the 2.0 and 3.0 versions.(16) Cognitive status was categorized as either intact or mild (CPS score 0–2), moderate (CPS score 3–4), or severely impaired (CPS score 5–6). In a validation study, CFS scores were highly correlated with the CPS categories (r=0.72, P <.001).
Statistical Analysis:
Residents were followed from the index date until the time of the first event of incident hip fracture, death, Medicare disenrollment, or until a period of one year from the index date. Follow-up was truncated at 365 days to ensure that every resident was followed for up to a year or censored before a year of follow up if they died, were transferred or sustained a hip fracture. The crude incidence rate (IR) of hip fracture was calculated as the total number of hip fractures during each year, divided by 100 PY of follow-up time. Crude IRs were stratified by age categories and sex.
We fitted Poisson regression with robust standard errors (17) to estimate adjusted rates of hip fractures per 100 PY. Regression estimates were standardized to account for the distribution of the covariates.We first controlled for demographics (age, sex, race), and then we added clinical characteristics that have been identified as important predictors of hip fracture risk in nursing homes (e.g. severity of cognitive impairment, level of dependence with transfers). In order to describe whether trends were attenuated by differences in clinical characteristics over time, we compared the crude and adjusted IRs. Lastly we evaluated trends in the incidence of hip fractures over time using the fully-adjusted Poisson model to estimate incidence rate ratios and 95% confidence intervals (CI) for each year (2008–2015), with 2007 as the reference. We further conducted a test of trend using a linear spline model to test for differences in the rate of fracture between the years 2007–2013 and 2013–2015. Models were repeated using the secondary definition of hip fracture. Non-overlapping 95% CI or p-values ≤ 0.05 were considered statistically significant.
For post-fracture mortality, we followed the long-stay residents for up to 1 year from the date of fracture and recorded death from any cause. Mortality was stratified by sex. All analyses were conducted using SAS version 9.4 (SAS Institute; Cary, NC) and Stata version 16.0.
Results:
The cohort consisted of 2,610,193 newly-admitted nursing home residents who met all study criteria between January 1, 2007 and December 31, 2015. The absolute number of qualifying newly admitted long-stay residents declined over time, from 324,508 in 2007 to 257,350 in 2015. The population was primarily female (68.8% in 2007, 65.1% in 2015) and the median age was 84 years (range 65–115 years). Most residents had at least moderate cognitive and functional impairment. (Table 1) Notably, the proportion of residents who were Black or Hispanic, increased over time. Also, the proportion of newly admitted residents with severe functional impairment increased over time, as evidenced by the increase in the mean Activities of Daily Living score and the proportion of residents requiring extensive or total assistance with transfers. The proportion of newly-admitted residents with a recent hip fracture declined somewhat overtime (8.2% in 2008, 7.7% in 2015).
Table 1.
Characteristics of 2,610,193 long-stay nursing home residents (2007–2015)*
| 2007 (n=324,508) | 2008 (n=315,270) | 2009 (n=299,825) | 2010 (n=292,778) | 2011 (n=299,964) | 2012 (n=282,570) | 2013 (n=274,501) | 2014 (n=263,427) | 2015 (n=257,350) | |
|---|---|---|---|---|---|---|---|---|---|
| Age, mean ± SD | 83.5 (7.8) | 83.5 (7.9) | 83.5 (7.9) | 83.4 (8.0) | 83.4 (8.1) | 83.3 (8.2) | 83.2 (8.3) | 83.1 (8.4) | 83.1 (8.5) |
| Age Category, % | |||||||||
| 65–69 | 6.3 | 6.6 | 6.8 | 7.3 | 7.4 | 7.9 | 8.2 | 8.5 | 8.9 |
| 70–74 | 9.5 | 9.6 | 9.8 | 9.9 | 10.1 | 10.5 | 10.8 | 11.3 | 11.6 |
| 75–79 | 15.5 | 15.0 | 14.9 | 14.7 | 14.6 | 14.5 | 14.6 | 14.7 | 14.6 |
| 80–84 | 23.1 | 22.8 | 22.3 | 21.8 | 21.3 | 20.8 | 20.2 | 19.6 | 19.2 |
| 85–89 | 24.7 | 24.8 | 24.8 | 24.7 | 24.4 | 23.9 | 23.4 | 22.9 | 22.5 |
| 90+ | 21.0 | 21.3 | 21.4 | 21.5 | 22.1 | 22.4 | 22.8 | 23.0 | 23.2 |
| Female, % | 68.8 | 68.5 | 67.8 | 67.4 | 67.2 | 66.7 | 66.0 | 65.3 | 65.1 |
| Race, % | |||||||||
| White | 83.8 | 83.7 | 83.3 | 82.9 | 82.4 | 81.9 | 81.9 | 81.8 | 81.8 |
| Black | 10.3 | 10.2 | 10.4 | 10.6 | 10.7 | 10.8 | 10.7 | 10.7 | 10.8 |
| Hispanic | 3.9 | 4.0 | 4.1 | 4.3 | 4.5 | 4.8 | 4.8 | 4.8 | 4.8 |
| Asian | 1.2 | 1.3 | 1.4 | 1.4 | 1.5 | 1.6 | 1.7 | 1.6 | 1.6 |
| Other | 0.9 | 0.9 | 0.8 | 0.8 | 0.9 | 1.0 | 1.0 | 1.1 | 1.1 |
| Severity of Cognitive Impairment, % | |||||||||
| Intact or Mild | 48.1 | 48.6 | 47.5 | 48.1 | 46.0 | 47.6 | 48.6 | 49.6 | 50.3 |
| Moderate | 42.6 | 42.3 | 43.3 | 43.1 | 42.8 | 41.4 | 40.9 | 40.4 | 40.1 |
| Severe | 9.3 | 9.1 | 9.2 | 8.8 | 11.3 | 10.9 | 10.4 | 10.2 | 9.6 |
| Daily Function MDS ADL Scale, mean ± SD | 14.7 (7.4) | 14.9 (7.3) | 15.3 (7.1) | 15.6 (6.8) | 15.7 (6.7) | 16 (6.5) | 16.1 (6.3) | 16.1(6.2) | 16.2 (6.0) |
| Transfer Assistance, % | |||||||||
| Independent | 17.0 | 15.8 | 14.2 | 12.3 | 11.7 | 10.3 | 9.4 | 8.6 | 7.8 |
| Supervision | 8.8 | 8.5 | 8.3 | 7.9 | 8.9 | 9.0 | 9.4 | 10.0 | 10.5 |
| Limited Assistance | 24.5 | 23.8 | 23.2 | 22.9 | 20.6 | 19.0 | 18.6 | 17.8 | 17.8 |
| Extensive Assistance | 36.6 | 39.0 | 41.6 | 44.8 | 47.4 | 51.1 | 52.6 | 53.9 | 55.2 |
| Total Dependence | 12.6 | 12.3 | 12.2 | 11.7 | 10.6 | 9.7 | 9.0 | 8.6 | 8.0 |
| Dementia or Alzheimer’s Disease, % | 53.0 | 52.9 | 53.8 | 53.0 | 57.5 | 56.1 | 55.5 | 55.0 | 54.3 |
| Residents with prior hip fracture, N (%) | † | 25,717 (8.2) | 24,165 (8.1) | 23,503 (8.0) | 22,422 (7.5) | 21,259 (7.5) | 20,653 (7.5) | 20,248 (7.7) | 19,756 (7.7) |
| One year mortality‡, N (%) | 27.7 | 26.9 | 26.4 | 27.2 | 26.4 | 27.3 | 26.3 | 27.5 | 26.8 |
Data from Minimum Data Set (MDS) assessments for qualifying newly admitted long-stay nursing home residents
Unable to calculate given incomplete data
One year mortality defined as the proportion of residents who died within one year of long-stay qualification
CPS=Cognitive Performance Scale; ADL=Activities of Daily Living; Katz ADL Scale – Higher scores indicate greater levels of functional
Incidence of hip fracture
In total, there were 66,781 hip fractures during 2,158,300 PY of follow-up (crude IR 3.1/100 PY). From 2007 to 2009, the unadjusted IRs for hip fracture were relatively stable, then declined with each subsequent year of follow-up, until 2013, when the incidence of fracture increased through 2015. (Table 2) Women and residents age 85 years and older had the highest hip fracture IRs over time, although the trends over time mirrored the overall cohort. (Table 2)
Table 2.
Incidence rates of hip fracture and one-year mortality following a hip fracture among newly admitted nursing home residents (2007–2015)
| Year | Person years follow up (100 PY) | Hip Fracture, total [n] | IR Unadjusted (95% CI) | IR Adjusted* (95% CI) | 1-year mortality† (95% CI) | |
|---|---|---|---|---|---|---|
| 2007 | Total | 2,702 | 8,967 | 3.32 (3.25, 3.39) | 3.11 (3.05, 3.18) | 42.6 (41.6, 43.7) |
| Women | 1,891 | 6,600 | 3.49 (3.41, 3.58) | 3.28 (3.20, 3.36) | 39.1 (37.9, 40.2) | |
| Men | 811 | 2,367 | 2.92 (2.80, 3.04) | 2.79 (2.68, 2.91) | 52.6 (50.5, 54.6) | |
| 2008 | Total | 2,634 | 8,630 | 3.28 (3.21, 3.35) | 3.12 (3.05, 3.18) | 41.1 (40.1, 42.2) |
| Women | 1,834 | 6,459 | 3.52 (3.44, 3.61) | 3.36 (3.27, 3.44) | 37.6 (36.4, 38.7) | |
| Men | 800 | 2,171 | 2.71 (2.60, 2.83) | 2.63 (2.52, 2.74) | 51.8 (49.7, 53.9) | |
| 2009 | Total | 2,519 | 8,068 | 3.20 (3.13, 3.27) | 3.09 (3.03, 3.16) | 40.9 (39.8, 41.9) |
| Women | 1,735 | 5,899 | 3.40 (3.32, 3.49) | 3.29 (3.20, 3.37) | 36.8 (35.6, 38.0) | |
| Men | 785 | 2,169 | 2.76 (2.65, 2.88) | 2.70 (2.59, 2.82) | 51.9 (49.8, 54.0) | |
| 2010 | Total | 2,411 | 7,505 | 3.06 (2.99, 3.13) | 3.03 (2.96, 3.09) | 41.5 (40.4, 42.6) |
| Women | 1,652 | 5,488 | 3.27 (3.18, 3.36) | 3.23 (3.14, 3.32) | 37.9 (36.6, 39.2) | |
| Men | 759 | 2,017 | 2.62 (2.50, 2.73) | 2.61 (2.50, 2.73) | 51.3 (49.1, 53.5) | |
| 2011 | Total | 2,470 | 7,374 | 2.92 (2.85, 2.99) | 2.90 (2.83, 2.97) | 41.5 (40.3, 42.6) |
| Women | 1,687 | 5,421 | 3.14 (3.06, 3.23) | 3.13 (3.04, 3.21) | 38.0 (36.7, 39.3) | |
| Men | 783 | 1,953 | 2.44 (2.33, 2.55) | 2.44 (2.33, 2.55) | 51.0 (48.7, 53.2) | |
| 2012 | Total | 2,313 | 6,833 | 2.89 (2.82, 2.96) | 2.94 (2.87, 3.01) | 40.9 (39.7, 42.1) |
| Women | 1,571 | 4,986 | 3.10 (3.02, 3.19) | 3.17 (3.08, 3.26) | 37.0 (35.7, 38.3) | |
| Men | 741 | 1,847 | 2.43 (2.32, 2.54) | 2.49 (2.37, 2.61) | 51.4 (49.2, 53.7) | |
| 2013 | Total | 2,263 | 6,513 | 2.82 (2.75, 2.89) | 2.91 (2.84, 2.98) | 41.3 (40.1, 42.5) |
| Women | 1,521 | 4,740 | 3.05 (2.97, 3.14) | 3.16 (3.07, 3.25) | 37.8 (36.5, 39.2) | |
| Men | 743 | 1,773 | 2.33 (2.22, 2.44) | 2.40 (2.29, 2.52) | 50.7 (48.3, 53.0) | |
| 2014 | Total | 2,158 | 6,366 | 2.89 (2.82, 2.97) | 3.02 (2.94, 3.09) | 42.1(40.9, 43.3) |
| Women | 1,434 | 4,555 | 3.12 (3.03, 3.21) | 3.25 (3.15, 3.34) | 38.6 (37.2, 40.0) | |
| Men | 724 | 1,811 | 2.45 (2.34, 2.57) | 2.55 (2.43, 2.67) | 50.9 (48.6, 53.2) | |
| 2015 | Total | 2,113 | 6,525 | 3.03 (2.96, 3.10) | 3.19 (3.11, 3.27) | |
| Women | 1,400 | 4,634 | 3.25 (3.16, 3.34) | 3.42 (3.32, 3.52) | ||
| Men | 713 | 1,891 | 2.60 (2.48, 2.72) | 2.73 (2.60, 2.85) | ‡ |
Adjusted for age, sex, race, cognitive function, dependence in transfers
One year mortality is defined as the proportion of residents who died within one year following the hip fracture and is not cause-specific
Unable to calculate given incomplete data
In the adjusted models, rates of hip fracture were reduced, relative to crude IR estimates; this effect was greatest in the fully adjusted model. (Figure 2) In the fully-adjusted model using 2007 as the reference, hip fracture rates did not change significantly in 2008 (IRR2008 1.00, (95% CI, 0.97, 1.03)) or 2009 (IRR2009 1.00, (95% CI, 0.97, 1.03)). (Table 3) Following this period, there was a modest, but statistically significant decline in fractures over the next five years (2010 – 2014). Subsequently, a small increase in hip fractures was observed; the 2015 hip fracture IRR was similar to 2007 results (IRR 1.00, (95% CI, 0.97, 1.04)). Using a linear test of trend, both the decline in hip fracture rates from 2007–2013 and the increase in fracture rates from 2013–2015 were statistically significant (p<0.001). The absolute difference in fracture rates between 2007–2013 and 2013–2015 were quite small.
Figure 2.
Trends in annual incidence of hip fracture among long-stay nursing home residents (2007–2015)
Table 3.
Incidence rate ratios of hip fracture among newly admitted nursing home residents (2007–2015)
| Year | IRR Unadjusted (95% CI) | IRR Adjusted* (95% CI) |
|---|---|---|
| 2007 | REF | REF |
| 2008 | 0.99 (0.96, 1.02) | 1.00 (0.97, 1.03) |
| 2009 | 0.97 (0.95, 1.00) | 1.00 (0.97, 1.03) |
| 2010 | 0.93 (0.90, 0.96) | 0.97 (0.94,1.00) |
| 2011 | 0.89 (0.86, 0.92) | 0.93 (0.91, 0.96) |
| 2012 | 0.88 (0.85, 0.90) | 0.94 (0.91, 0.97) |
| 2013 | 0.86 (0.83, 0.89) | 0.93 (0.90, 0.96) |
| 2014 | 0.87 (0.85, 0.90) | 0.95 (0.92, 0.98) |
| 2015 | 0.93 (0.90, 0.96) | 1.00 (0.97, 1.04) |
Adjusted for age, sex, race, cognitive function, dependence in transfers
Using the alternative definition of hip fracture (inclusion of femoral shaft fractures), we identified approximately 10% additional fractures in each year. Observed trends using the alternative definition of hip fracture were similar (Results not shown).
Mortality
The proportion of residents in the hip fracture cohort who died from any cause in the 365 days following the event was considerable, and largely flat from 2007 to 2014. (Table 2) Substantially more men than women died following a hip fracture during each year of observation.
Discussion
These findings provide important context for evaluating rates of hip fracture among U.S. nursing home residents, the population sector at greatest risk for hip fracture occurrence and post-fracture mortality. In a large national sample of newly admitted long-stay nursing home residents, we found high rates of hip fracture throughout the nine years of observation. Inspection of the unadjusted hip fracture IRs reveals a somewhat U-shaped curve, with the nadir occurring around 2013; however, controlling for differences in relevant patient characteristics somewhat attenuated this pattern. The absolute difference in fracture rates between 2007–2013 and 2013–2015 were quite small.
Our results confirm and extend previous studies of temporal trends in hip fracture. Several recent international cohort studies have described a similar decline in hip fracture rates through 2012.(18–20) A U.S. study conducted in a 20% sample of female Medicare beneficiaries described a decline in the incidence rate of hip fracture among United States women between 2002 and 2012, followed by a plateau in the incidence rate from 2013–2015.(8) This leveling in the rates of hip fracture parallels a marked decline in screening and treatment of osteoporosis in the community setting, although largely observed as an ecologic association. We also observed a modest decline in hip fracture rates through 2013, followed by a modest upwards trend in hip fracture rates in more recent years. Rates of screening and treatment for osteoporosis in the nursing home are much lower than in the community setting:(21) for example, one U.S. study found only 36% of high-risk nursing home residents received any intervention to prevent fracture (mostly calcium and vitamin D), with variation in treatment across facilities.(22) Given the low overall prevalence of screening and treatment in this population, we would not expect temporal changes in treatment to have a large impact on our results.
Instead, we noted considerable differences in resident characteristics across study years, that when adjusted for, partially attenuated any trends. For example, a greater proportion of residents were male with dependence in ADLs and transfers in 2013–2015 as compared with 2007. In contrast with community based studies, nursing home residents with ADL or transfer dependence are at a much lower risk of hip fracture, likely because they do not have the opportunity to fall.(23) Over the same time period, a greater proportion of newly admitted residents were Black or Hispanic. These differences in race and functional characteristics appear to explain part of the modest decrease in the rate of fracture we observed from 2007 to 2013. Future studies evaluating fracture trends should consider whether there may be significant clinical changes, such as change in function or polypharmacy, in the study population that would affect fracture risk.
It is disappointing that rates of hip fracture in the nursing home have remained persistently high for more than a decade. Available strategies exist to prevent falls and associated injuries, albeit, the success of falls prevention programs in the nursing home has been less than in the community setting.(24) In general, nursing home residents are older and sicker, with more cognitive and functional impairment than community-dwellers. Rates of hip fracture in our study of newly admitted nursing home residents were nearly 10-fold greater than in the previously reported Medicare sample of women during the same time period.[8] One possible explanation of these high rates is the underutilization of medications to treat osteoporosis. There is limited direct evidence that these drugs prevent fracture in a frail, nursing home population. We recently conducted a large, observational study to compare the incidence of hip and non-vertebral fracture in nursing home residents newly prescribed a bisphosphonate versus calcitonin (n=10,418).(25) Over a mean follow-up of 2.5 years, new bisphosphonate users had a 17% less relative risk than calcitonin users to experience hip fracture (5.1% versus 5.8%; HR, 0.83; 95% CI, 0.71 to 0.98). This relative difference was smaller than the effect size observed in bisphosphonate trials of younger women with postmenopausal osteoporosis; however, the absolute risk reduction over this short time interval was quite similar.(26) More data are needed to inform judicious use of these drugs in this setting. A second possibility for the persistently high rates of hip fracture is the lack of effective fall prevention strategies in this setting. Aside from vitamin D supplementation, there is little direct evidence that single interventions prevent falls in this setting.(24) Development of effective strategies to prevent falls in the nursing home is urgently needed. Because of the substantial difference in how falls were recorded in the MDS version 2.0 vs. 3.0, we were unable to assess trends in falls over the duration of the study. A third possible reason for the persistently high fracture rates in our study is a lack of adequate staff support. We observed a marked decline in the number of newly admitted residents between 2007 and 2015. Even if staffing levels were stable during this time period, it may not have been sufficient given the marked increase in the functional dependence of the residents that were observed. Increased patient to staff ratio has been associated with greater fall rates in the nursing home. (27)
In our analysis, we observed little change in mortality following a hip fracture between 2007 and 2015. Similarly a study of centenarians with hip fracture found no difference in mortality between 1990 and 2012.(28) A more recent community-based study from Kaiser Permanente (mean age 81 years) reported that one year mortality following hip fracture was 21%.(29) Mortality was higher in our study of newly admitted nursing home residents with hip fracture: 41–42%. Effective strategies to reduce mortality following a hip fracture include avoidance of surgical delay (30) and possibly geriatric co-management.(31) It remains unclear whether these strategies are also effective in reducing mortality in a more frail, nursing home population.
The strengths of our study include a large cohort that includes every U.S. nursing home resident in Medicare fee-for-service. The Centers for Disease Control and Prevention reported that the number of persons aged 65 years or greater residing in a nursing home decreased from 1.32 million to 1.16 million between 2004 and 2014.(32) We observed a similar decline in the numbers of newly admitted nursing home residents over this same time period, and so we expect that our study population is representative and unbiased. Further, we had repeated measures of function and cognition that were used to adjust rates of hip fracture, which is not typically found in population based-studies. We elected to adjust for these functional domains, rather than co-morbidities, given the strong association between functional characteristics and fracture in this population.(23)
This study also has limitations. First, there were differences between the way several characteristics were measured between MDS versions 2.0 through 2010 and version 3.0 in later years. We only considered measures that were minimally changed or have been previously validated and comparable across versions, but it is possible that some of the differences in patient characteristics seen after 2010 reflect differences in the way the characteristics were measured. Notably, dependence in transfers was already increasing from 2007–2010, before the MDS version change occurred, and so it is likely that the observed trends in patient characteristics were not due to changes in the assessment tool. Second, ICD classification used to ascertain hip fractures changed in the final 15 months of the study. Although the ICD-10 diagnostic codes are greatly expanded, we used similar codes and definition as with ICD-9. In a sensitivity analysis using a more broad definition of hip fracture that included femur fractures, the results were similar. Third, we did not have complete information on Medicare Part D and prescription medications for osteoporosis and other conditions during all years of observation. There has been considerable effort to decrease antipsychotic use, a risk factor for fracture, in nursing home facilities during the study period.(33) Adjusting for antipsychotics and other medications associated with falls and fracture may have further attenuated the slight downward trend in fractures observed in the early study years. Fourth, we excluded the minority of nursing home residents enrolled in Medicare Advantage because we did not have claims information on them. Nursing home patients enrolled in Medicare Advantage are younger and less cognitively impaired as compared with patients in fee-for-service Medicare(34), and it is unclear if our results generalize to the Medicare Advantage population. Fifth, we elected to follow residents for a maximum of one year. This ensures that the adjusted rates reflect differences in outcomes in the calendar year, rather than differences over time. An alternative approach would have been a model allowing longer follow-up time with repeated information on resident characteristics or a model that assigned person-time and fractures to the actual calendar year of occurrence. It is unclear that either alternative model would be an improvement. Finally, we restricted our analysis to residents who survived 100 days in a nursing home facility, in order to ensure that we were examining patient characteristics and fractures in a vulnerable, long-stay population and excluding most patients who were likely to return to the community. We have previously demonstrated that rates of hip fracture are somewhat higher in the first 100 days after admission (2.7 hip fractures/100 person-years).(2) Because we were consistent in our inclusion criteria across the study period, this is unlikely to explain the trends we observed.
In summary, rates of hip fracture in U.S. nursing homes remain persistently high over the past decade. Inspection of the unadjusted hip fracture IRs appears U-shaped with a nadir around 2013 and a slight rise in more recent years. A marked increase in the number of newly admitted residents who are male, Black or Hispanic, and dependent in transfers, at least partially explained the decline in hip fracture observed through 2013, as controlling for these characteristics mostly attenuated this downward trend. It is less clear what could be driving the more recent modest rise in hip fracture rates we observed. We suspect that this rise may partially be explained by substantial differences in the way hip fractures are captured via ICD-10 claims codes in the final years of the study, but also that temporal trends in prescribing medications associated with falls and fracture may contribute. Linked datasets including measures of function offer an important opportunity to study trends in fracture independent of patient characteristics. Our findings and those in community dwelling older adults underscore the magnitude of the hip fracture problem in the U.S. and should prompt widespread interventions to reduce the suffering associated with hip fractures in older adults.
Acknowledgements:
This work was supported by the National Institute of Health, National Institute on Aging (NIA), 1R01AG045441, 5P01AG027296-05, and R21AG061632.
Footnotes
These analyses were presented in part as abstracts on November 14, 2018 at the Gerontological Society of America Annual Meeting, Boston, Massachusetts and on September 30 2018 at the American Society for Bone and Mineral Research Annual meeting.
References:
- 1.Hall MJ, DeFrances CJ, Williams SN, Golosinskiy A, Schwartzman A. National Hospital Discharge Survey: 2007 summary. Natl Health Stat Report. 2010(29):1–20, 4. [PubMed] [Google Scholar]
- 2.Berry SD, Lee Y, Zullo AR, Kiel DP, Dosa D, Mor V. Incidence of Hip Fracture in U.S. Nursing Homes. J Gerontol A Biol Sci Med Sci. 2016;71(9):1230–4. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Neuman MD, Silber JH, Magaziner JS, Passarella MA, Mehta S, Werner RM. Survival and functional outcomes after hip fracture among nursing home residents. JAMA Intern Med. 2014;174(8):1273–80. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Berry SD, Samelson EJ, Bordes M, Broe K, Kiel DP. Survival of aged nursing home residents with hip fracture. J Gerontol A Biol Sci Med Sci. 2009;64(7):771–7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Baumgarten M, Margolis DJ, Orwig DL, Shardell MD, Hawkes WG, Langenberg P, et al. Pressure ulcers in elderly patients with hip fracture across the continuum of care. J Am Geriatr Soc. 2009;57(5):863–70. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Dharmarajan TS, Tankala H, Patel B, Sipalay M, Norkus EP. Outcome in ambulatory status immediately following hip fracture surgery in the acute setting: a comparison of nursing home residents and community older adults. J Am Med Dir Assoc. 2001;2(3):115–9. [PubMed] [Google Scholar]
- 7.Beaupre LA, Jones CA, Johnston DW, Wilson DM, Majumdar SR. Recovery of function following a hip fracture in geriatric ambulatory persons living in nursing homes: prospective cohort study. J Am Geriatr Soc. 2012;60(7):1268–73. [DOI] [PubMed] [Google Scholar]
- 8.Michael Lewiecki E, Wright NC, Curtis JR, Siris E, Gagel RF, Saag KG, et al. Hip fracture trends in the United States, 2002 to 2015. Osteoporos Int. 2018;29(3):717–22. [DOI] [PubMed] [Google Scholar]
- 9.Morris JN, Hawes C, Fries BE, Phillips CD, Mor V, Katz S, et al. Designing the national resident assessment instrument for nursing homes. Gerontologist. 1990;30(3):293–307. [DOI] [PubMed] [Google Scholar]
- 10.Intrator O, Hiris J, Berg K, Miller SC, Mor V. The residential history file: studying nursing home residents’ long-term care histories(*). Health Serv Res. 2011;46(1 Pt 1):120–37. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Wright NC, Daigle SG, Melton ME, Delzell ES, Balasubramanian A, Curtis JR. The Design and Validation of a New Algorithm to Identify Incident Fractures in Administrative Claims Data. J Bone Miner Res. 2019. [DOI] [PubMed] [Google Scholar]
- 12.McBean AM. Improving Medicare’s data on race and ethnicity. Medicare Brief. 2006(15):1–7. [PubMed] [Google Scholar]
- 13.Hawes C, Phillips CD, Mor V, Fries BE, Morris JN. MDS data should be used for research. The Gerontologist. 1992;32(4):563–4. [DOI] [PubMed] [Google Scholar]
- 14.Carpenter GI, Hastie CL, Morris JN, Fries BE, Ankri J. Measuring change in activities of daily living in nursing home residents with moderate to severe cognitive impairment. BMC Geriatr. 2006;6:7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Hartmaier SL, Sloane PD, Guess HA, Koch GG, Mitchell CM, Phillips CD. Validation of the Minimum Data Set Cognitive Performance Scale: agreement with the Mini-Mental State Examination. J Gerontol A Biol Sci Med Sci. 1995;50(2):M128–33. [DOI] [PubMed] [Google Scholar]
- 16.Thomas KS, Dosa D, Wysocki A, Mor V. The Minimum Data Set 3.0 Cognitive Function Scale. Med Care. 2017;55(9):e68–e72. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Zou G A modified poisson regression approach to prospective studies with binary data. Am J Epidemiol. 2004;159(7):702–6. [DOI] [PubMed] [Google Scholar]
- 18.Chen FP, Shyu YC, Fu TS, Sun CC, Chao AS, Tsai TL, et al. Secular trends in incidence and recurrence rates of hip fracture: a nationwide population-based study. Osteoporos Int. 2017;28(3):811–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Orimo H, Yaegashi Y, Hosoi T, Fukushima Y, Onoda T, Hashimoto T, et al. Hip fracture incidence in Japan: Estimates of new patients in 2012 and 25-year trends. Osteoporos Int. 2016;27(5):1777–84. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Wu TY, Hu HY, Lin SY, Chie WC, Yang RS, Liaw CK. Trends in hip fracture rates in Taiwan: a nationwide study from 1996 to 2010. Osteoporos Int. 2017;28(2):653–65. [DOI] [PubMed] [Google Scholar]
- 21.Parikh S, Brookhart MA, Stedman M, Avorn J, Mogun H, Solomon DH. Correlations of nursing home characteristics with prescription of osteoporosis medications. Bone. 2011;48(5):1164–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Colon-Emeric C, Lyles KW, Levine DA, House P, Schenck A, Gorospe J, et al. Prevalence and predictors of osteoporosis treatment in nursing home residents with known osteoporosis or recent fracture. Osteoporos Int. 2007;18(4):553–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Berry SD, Zullo AR, Lee Y, Mor V, McConeghy KW, Banerjee G, et al. Fracture Risk Assessment in Long-term Care (FRAiL): Development and Validation of a Prediction Model. J Gerontol A Biol Sci Med Sci. 2018;73(6):763–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Cameron ID, Dyer SM, Panagoda CE, Murray GR, Hill KD, Cumming RG, et al. Interventions for preventing falls in older people in care facilities and hospitals. Cochrane Database Syst Rev. 2018;9:CD005465. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Zullo AR, Zhang T, Lee Y, McConeghy KW, Daiello LA, Kiel DP, et al. Effect of Bisphosphonates on Fracture Outcomes Among Frail Older Adults. J Am Geriatr Soc. 2019;67(4):768–76. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Black DM, Thompson DE, Bauer DC, Ensrud K, Musliner T, Hochberg MC, et al. Fracture risk reduction with alendronate in women with osteoporosis: the Fracture Intervention Trial. FIT Research Group. J Clin Endocrinol Metab. 2000;85(11):4118–24. [DOI] [PubMed] [Google Scholar]
- 27.Leland NE, Gozalo P, Teno J, Mor V. Falls in newly admitted nursing home residents: a national study. J Am Geriatr Soc. 2012;60(5):939–45. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Mosfeldt M, Madsen CM, Lauritzen JB, HL JO. Centenarian hip fracture patients: a nationwide population-based cohort study of 507 patients. Acta Orthop. 2019:1–10. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Okike K, Chan PH, Paxton EW. Effect of Surgeon and Hospital Volume on Morbidity and Mortality After Hip Fracture. J Bone Joint Surg Am. 2017;99(18):1547–53. [DOI] [PubMed] [Google Scholar]
- 30.Pincus D, Ravi B, Wasserstein D, Huang A, Paterson JM, Nathens AB, et al. Association Between Wait Time and 30-Day Mortality in Adults Undergoing Hip Fracture Surgery. JAMA. 2017;318(20):1994–2003. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Baroni M, Serra R, Boccardi V, Ercolani S, Zengarini E, Casucci P, et al. The orthogeriatric comanagement improves clinical outcomes of hip fracture in older adults. Osteoporos Int. 2019;30(4):907–16. [DOI] [PubMed] [Google Scholar]
- 32.Services USDoHaH. Health, United States, 2016: With Chartbook on Long-term Trends in Health. 2017. [PubMed]
- 33.Mitka M CMS seeks to reduce antipsychotic use in nursing home residents with dementia. JAMA. 2012;308(2):119, 21. [DOI] [PubMed] [Google Scholar]
- 34.Kumar A, Rahman M, Trivedi AN, Resnik L, Gozalo P, Mor V. Comparing post-acute rehabilitation use, length of stay, and outcomes experienced by Medicare fee-for-service and Medicare Advantage beneficiaries with hip fracture in the United States: A secondary analysis of administrative data. PLoS Med. 2018;15(6):e1002592. [DOI] [PMC free article] [PubMed] [Google Scholar]