paracentesis and mortality in u.s. hospitals josé l. gonzález, md wednesday, june 25 th,...

24
Paracentesis and Mortality in U.S. Hospitals José L. González, MD Wednesday, June 25 th , 2014 Journal Club

Upload: grant-kelly-boone

Post on 29-Dec-2015

217 views

Category:

Documents


0 download

TRANSCRIPT

Paracentesis and Mortality in U.S. HospitalsJosé L. González, MD

Wednesday, June 25th, 2014Journal Club

Intro: Retrospective Observational Design

Does paracentesis decrease in-hospital mortality?

Reasons for this Study:

ASLD recommends

Quality indicator

Data linking paracentesis and outcomes is lacking

Epidemiology

Paracentesis is performed about 60% of the time

Occurs in 25% of patients w/ clinically significant ascites

SBP is fatal in 30% of patients

Methods:

Data Source: 2009 Nationwide Inpatient Sample (NIS)

Data stratified by: size ownership teaching status location

Sample:

>18 years of age

Excluded transfers from OSH

ICD-9 Codes: Ascites SBP HES (if ascites is a secondary dx)

*All of the above pts had to have a 2º dx of cirrhosis

+/- Paracentesis procedure code

Variables considered:

Early vs Delayed :: <1 day vs >1day

Age

Sex

Race / Ethnicity

Weekday vs weekend

Insurance provider

Income

Comorbidities

Hospital Factors Considered:

Size

Ownership

Private

U.S. region

Teaching status

Rural vs. urban

Outcomes:

1º In-hospital mortality

2º Hospital length of stay Hospital charges

Statistics:

Categorical variables: Pearson X2

Continuous variables: Student t test

Re-examination of stats after excluding those who died on the day of admission

Results

40 million DCs in 2009

17,741 met inclusion criteria

10,743 paracentesis were performed (61%)

Diagnosis N paracentesis performed

HES 10,500

56%

Ascites 2,977

SBP 4,233 77%

Results

Results:Para or no para

Increased likelihood to have had paracentesis Slightly younger Higher median income Dx of Sepsis & ARF Less likely to be in the South Teaching or urban hospital

56.4% in the South & 64.1% in the NE

Results:Para or no para

No difference: Sex, race, admitting circumstance, primary payer, # of

comorbidities, hospital size or ownership

Para independently associated w/ Self-pay ARF Teaching status of hospital

Less likely to be done on the weekends

Results:Primary Outcome

Those who received a para had a lower in-hosp mortality than those who did not (6.5% vs 8.5%, P = .03)

In-hosp mortality was lower in the Midwest

Those who died: Had more comorbidities More likely to have had sepsis More likely to have had RF

Results:Primary Outcome

Dx of HES or ascites:

(6.8% vs 9.1% adjusted OR) 0.54: 95% CI, 0.38-0.76

Dx of SBP

(5.8% vs 4.7% adjusted OR) 0.91: 95% CI, 0.38-2.19

Results:Primary Outcome

Delayed para <1 day vs >1 day

More likely to be Female be Admitted on weekend have Medicare Have more comorbidities To have ARF To be in a private, nonprofit hosp

And less likely to be in a teaching hospital

5.7% vs 8.1% p = 0.49, but not stat sig (0.78-2.02 CI)

Results:Secondary Outcome

Hospital Length of Stay and Hospital Charges

Para = 6.6 days, $44,586

No para = 5.3days, $ 31,746

Conclusions

Pts w/ cirrhosis and ascites, only 61% undergo para

Paracentesis in these patients is associated w/ improved mortality

Paracentesis in all pts studied is associated w/ increased LOS and hospital charges

Discussion

Only 61% of patients admitted for ascites or HES had a paracentesis

1996 survey data: IM graduating residents are comfortable w/ the procedure

Weekend admissions are associated w/ decrease para

Detail in NIS info doesn’t tell us why, potential reasons

Low index of suspicion for SBP Tx empirically

Discussion

Mechanism for beneficial effect? Probably due to increased detection and tx of SBP

Para 6.8%

HES or ascites

No Para 9.1%

Para 5.8%

SBP

No para 4.7%

DiscussionSecondary Outcomes

Unit of obs = each admission, so readmission can’t be assessed

LOS and $ were increased in paracentesis group Undiagnosed SBP cases may have been DCd b4

recognition? How much did increased mortality contribute to

decreased LOS/$?

Study Limitations

Administrative data reliant on coding Canadian study, > 80% sensitivity for patacentesis

Data don’t distinguish between diagnostic and therapeutic paras

Subclinical ascites?

Did severity of illness influence decision to perform paras?

Increased likelihood in sepsis and ARF Other studies show that worse liver dz is ass. w/

recommended ascites care

Association but not causality

Sources

Orman E, Hayashi P, Bataller R et al. Paracentesis and Mortality in U.S. Hospitals. Clinical Gastroenterology and Hepatology 2014; 12:496-503.

Runyon, Bruce. Management of Adult Patients with Ascites Due to Cirrhosis: Update 2012. AASLD Practice Guideline, 2012.