Hip Fractures in VA/Medicare-Eligible Veterans: Mortality and Costs

1
Hip Fractures in VA/Medicare-Eligible Veterans
Mortality and Costs

• Elizabeth Bass, PhD,1 Dustin D. French, PhD,1
  Douglas D. Bradham, DrPH,2 Laurence Z.
  Rubenstein, MD, MPH3
• 1VISN-8 Patient Safety Center of Inquiry, James A
  Haley VAMC
• 2VA Cooperative Studies Program Coordinating
  Center at Perry Point, MD and University of
  Maryland School of Medicine
• 3UCLA David Geffen School of Medicine and VA
  Greater Los Angeles Healthcare System Geriatric
  Research, Education and Clinical Center
• HERC Cyber Seminar
• July 25, 2007

2
Outline

• Brief overview of previous literature
• Data sources
• Model, including comorbidity adjuster
• Mortality outcomes
• Cost outcomes
• Implications

3
Background

• Why hip fractures are a real problem in the
  elderly
• Usually includes inpatient stay and rehab
• Leads to other clinical issues
• High rates of mortality
• Previous research
• Used small samples
• Used mostly female samples
• Whats going on in the VA?

4
Goals

• establish risk-adjusted mortality rates for
  elderly veterans who sustained a hip fracture
  over a 12 month time period
• confirm gender difference
• estimate costs to Medicare

5
Data sources

• Medicare (Standard Analytical Files and
  Denominator File) for veterans supplied by the VA
  Information Resource Center (VIReC) 4.7 million
  individuals
• VHA-eligible
• enrolled in the VHA
• use VHA care
• receive compensation from the Department of
  Veterans Affairs
• Covers enrollment phase and follow-up period for
  fracture patients in 1999-2002 (2003) to address
  right-censoring (VIREC now has data through 2004)

6
Methods

• Retrospective, incident-hip fracture cohort
  analysis of veterans aged 65
• Patients selected by a first-ever admitting
  diagnosis of hip fracture (ICD-9-CM codes
  820-820.9 or 905.3) to a Medicare facility from
  1999-2002
• Mortality viewed at several time points up to 1
  year
• Costs defined as Medicare payments
• Selection of comorbidity adjuster

7
Comorbidity adjuster

• What are the options?
• Why an Elixhauser?

8
Model Methods365-day Mortality

• Model
• Mortality f(age, gender, comorbidity adjuster)
• Statistical Analysis Coxs proportional hazard
• Duration(t)Death age, gender, comorbidities

9
Model Methods365-day Costs

• Model
• Total Costs f(age, gender, inpatient length of
  stay, death within one year, comorbidity
  adjuster)
• Statistical Analysis OLS regression (no
  transformation after testing several functional
  forms)
• Total Medicare payments age, gender, length of
  stay, death, comorbidities
• beginning with first admission date

10
Results

• Demographics (n43,165)
• 87 male
• 94 Caucasian
• 80 mean age
• 7 days median length of stay
• 49 discharged to a skilled nursing facility
  (SNF)

11
Results mortality

• Unadjusted one year mortality rates
• (30 days 9.7, 90 days 17.5, 180 days
  24, 365 days 32.2)
• were approximately 10 higher than the adjusted
  rates (30 days 8.9, 90 days 15.6, 180 days
  21.8, one year 29.9).

12
Results mortality
13
Results mortality

• Big differences by gender the mortality odds for
  women 12 months after hip fracture were 18,
  compared to 32 for men.
• In other words, men were about twice as likely to
  die within one year of the hip fracture compared
  to women.

14
Results mortality
15
Results mortality

• The comorbidity adjustment coefficients suggest
  that
• metastasic cancer increased the risk of death by
  almost four times (hazard ratio 3.57)
• congestive heart failure increased risk by 63
• renal failure increased risk by 95
• lymphoma increased risk by 63
• weight loss increased risk by 90
• Contrary to expectation, hypertension and alcohol
  abuse were negatively correlated with one-year
  mortality.

16
Results Medicare payments

• Medicare spent nearly 3 billion for patients in
  this cohort from 1999-2003
• 70 of total annual Medicare payments for all
  services occurred within the first 30 days
  following hospital admission
• Hospital and physician reimbursements were
  approximately 3/4 of payments

17
Results Medicare payments

• Service use
• Inpatient 100
• Physician 99.1
• Outpatient care 83.6
• Skilled nursing facilities 64.1
• Durable Medical Equipment 57.2
• Home Health Agency 45.8
• Hospice 2.7
• With the exception of DME and hospice (70.1 for
  DME, 8.4 for hospice), these percentages showed
  only a modest increase at the end of 365 days.

18
Results Medicare payments

• Average Medicare Payments within 365 Days for Hip
  Fracture Patients 1999-2003 (N43,104)
• Provider Type Mean () 99 Confidence Interval
  ()
• Inpatient 26,884 26,545-27,223
• Carrier 24,401 23,895-24,906
• SNF 12,208 12,058-12,358
• Hospice 7,073 6,605-7,541
• Home Health Agency 5,249 5,146-5,352
• DME 4,447 4,217-4,676
• Outpatient 3,200
  3,085-3,315
• Total All Services 69,389 68,539-70,239
• SOURCE Medicare SAFs 1999-2003.

19
Results Medicare payments

• Cost function (OLS regression)
• Variable Coefficient estimate ()
• Age (years) -575
• Gender (1female) -3,557
• 1 year mortality (1died) 3,270
• Inpatient LOS (days) 1,944
• SOURCE Medicare SAFs 1999-2003

20
Results Medicare payments

• Variable Coefficient estimate ()
• Renal failure 52,043
• Lymphoma 47,185
• Metastastic cancer 35,359
• Diabetes w/complications 34,332
• Alcohol abuse -15,193
• More common comorbidities (COPD, CHF
  arrhythmias added between 6,200-10,000)
• SOURCE Medicare SAFs 1999-2003.

21
Limitations

• Did not control for provider characteristics
• Risk adjuster for broad patient population
• Varying levels of comorbidities
• Unable to fully explain gender difference in
  mortality and costs, alcohol effect and why older
  patients cost less

22
Concluding Remarks

• Economic implications
• High loss of life (downward spiral)
• Resource use intense
• What to do? Clinical intervention strategies such
  as
• gait and balance testing
• osteoporosis diagnosis
• medication review
• use of hip protectors

23
Tips for VA-Medicare projects

• Apply for Medicare data from VIReC as soon as
  possible
• Contact ResDAC frequently
• Have a good data manager
• Include researchers with varied backgrounds

24
Contact Information

• Elizabeth Bass
• VA Patient Safety Center of Inquiry
• Tampa, FL
• Elizabeth.Bass_at_va.gov
• 813-558-3908