Patient Safety Events IoM estimated that between 44,000 and

advertisement
Patient Safety Events
„ IoM estimated that between 44,000 and
Adverse Patient Safety Events: Costs of
Readmissions and Patient Outcomes Following
Discharge
98,000 people die in hospitals each year as a
result of medical errors
„ Medical errors is the eighth leading cause of
Didem M. Bernard, Ph.D.
William E. Encinosa, Ph.D.
death in the U.S.
„ IoM recommended at least a 50% reduction in
medical errors over 5 years
Background
Research Questions
„ Medical injuries during hospitalizations lead to
Do adverse patient safety events during
hospitalizations have an effect on patient
outcomes following discharge?
„ Whether adverse patient safety events have a
Are there longlong-term costs associated with
potentially preventable adverse medical
events?
excess length of stay, charges and mortality
(Zhan and Miller, 2003)
significant impact on patient outcomes and
costs following discharge is less clear
Patient Safety Indicators
„ We use the AHRQ Patient Safety Indicators (PSI)
developed by the UCSF
- Stanford Evidence
- Based
Practice Center (EPC) and UC
- D
avis (2003).
„ EPC selected 14 indicators among 63 complication
indicators using
– Indicator validity results from the literature
– Empirical analyses of the precision and reliability
of the candidate PSIs
– Review of potential indicators by 11 clinical
panels
Patient Safety Indicators
1
Anesthesia Complications *
2
Accidental Puncture or Laceration During Procedure
3
Foreign Body Left In During Procedure*
4
Hemorrhage or Hematoma
5
Wound Dehiscence
6
Infection Due to Medical Care
7
Pulmonary Embolism and Deep Vein Thrombosis
* Labeled as medical errors
1
Patient Safety Indicators (cont.)
8
Iatrogenic Pneumothorax*
9
Acute Respiratory Failure
10 Sepsis
11 Physiologic and Metabolic Derangements
12 Transfusion Reaction
13 Hip Fracture*
14 Decubitus Ulcer*
Data
„ 2002 California Hospital Discharge Data
AHRQ Healthcare Cost and Utilization Project
(HCUP) State Inpatient Database (SID)
„ Hospital Inpatient Cost to Charge Ratios
„ Area Wage Index (CMS)
„ Number of Hospitals: 349
* Labeled as medical errors
Analytic Sample and Methods
„ Major Surgery Discharges (17<age<65)
From February 1, 2002 to November 30, 2002
„ Index admission: A major surgery admission with no
hospitalizations in a 30 day window prior to the
admission date (N=602,361)
Patient Outcomes
„ Probability of readmissions
„ Probability of death during index admission
„ Probability of death during any readmission within 30
days of initial discharge
„ Readmission: Any hospitalization within 30 days of
initial discharge
Costs of Adverse Patient Safety Events
„ Index admission costs
„ Costs of readmissions
„ Total costs for index admissions and readmissions
within 30 days following discharge
Basic Model: Hospital Costs
yi = β 0 + βPSIPSIi + β xXi + εi
i=person
y= cost of hospitalization*
PSI= (0,1) indicator equals 1 if patient has at least one
adverse patient safety event
x=control variables
* Index admission, readmission or total hospital costs
2
Basic Model: Patient Outcomes
yi = β 0 + βPSIPSIi + β xXi + εi
i=person
y=(0,1) indicator equals 1 if patient is readmitted within
30 days of discharge*
PSI= (0,1) indicator equals 1 if patient has at least one
adverse patient safety event
x=control variables
Control Variables
„ Patient: age, sex, race/ethnicity, transfer admission,
emergency admission, 30 chronic conditions, MDC,
zipcode level median household income
„ Hospital: teaching status, ownership type, bedsize
„ Market: Health Service Area dummies
* y=(0,1) indicator equals 1if patient dies during hospitalization
hospitalization
Patient Outcomes
Average Costs
PSI=0
PSI=1
Live discharge
Death during
index admission
84.55%
63.03%
0.86%
5.97%
PSI=0
$12,579
$17,927
$15,198
Readmission
Death during
readmission
14.14%
28.32%
PSI=1
$31,096
$33,445
$41,510
0.46%
2.68%
Difference
147%
N
583,836
18,889
Index
Hospitalization Readmission
87%
173%
Simulation Results: Outcomes
Estimation
- link
„ Costs: GLM (Generalized Liner Method) log
with Gamma distribution (Manning and Mullahy,
2001)
„ Patient Outcomes: Logit regressions
„ Simulation results:
A : Predicted costs/outcomes with PSI=1
B : Predicted costs/outcomes with PSI=0
(A – B) : Predicted costs/outcomes due to adverse
patient safety events
Total Costs
Death
during index
Death during
hospitalization readmission
Death within
30 days of
discharge
PSI=0
0.92%
3.4%
1.4%
PSI=1
1.96%
5.2%
2.8%
% Difference
1.04%*
1.8%*
1.4%*
* Significantly different from zero at the 99% level
3
Simulation Results: Outcomes (cont.)
Simulation Results: Costs
Probability of
readmission
Index
Hospitalization Readmission
Total Costs
PSI=0
15.1%
PSI=0
$12,801
$18,264
$15,618
PSI=1
20.1%
PSI=1
PSI COST
$21,190
$8,388*
(232)
$26,198
$7,933*
(529)
$26,607
$10,989*
(284)
% Difference
65.5%
43.4%
70.4%
Difference
5.0%*
* Significantly different from zero at the 99% level
* Significantly different from zero at the 99% level
Summary
Policy Implications
„ Excess costs and adverse outcomes of
adverse patient safety events are not limited
to the initial hospitalization
„ Adverse patient safety events during
hospitalizations lead to
– Higher probability of readmissions
– Higher probability of inin-hospital death
following discharge
– Higher inpatient costs following discharge
A reduction in adverse patient safety events
will not only improve quality of care but also
reduce longlong-term health care costs
4
Download