Comparison of hospital length of stay between two insurers for patients with pediatric asthma

1
??????5

• Comparison of hospital length of stay between two
  insurers for patients with pediatric asthma

2
Introduction

• This case study investigates the relative
  importance of several factors in predicting the
  length of time young patients with asthma stay in
  the hospital.
• It is believed that some insurers have been more
  successful than others at minimizing hospital
  lengths of stay (LOS). To test this, a sample of
  hospital medical records was drawn for each of
  several illnesses from metropolitan hospitals
  operating in one state. The data for this case
  study consists of information abstracted from the
  medical records of asthma patients between the
  ages of 2 and 18 years old.

3
Sampling scheme

• The sample of medical records was drawn in two
  stages.
• At the first stage, 29 metropolitan hospitals
  were sampled with probabilities proportional to
  an estimate of the number of asthma admissions it
  had during a single year.
• At the second stage, 393 asthma cases insured by
  Insurer A were randomly selected and 396 asthma
  cases insured by Insurer B were randomly selected
  from the 29 hospitals.

4
Recorded variables

• Information was abstracted from each patients
  medical record. Aside from the main variables
  of interest, insurer and LOS, the additional
  information falls into four categories.
• A. Patient Severity Variables describe the
  severity of the patients condition at admission
  to the hospital.
• B. Demographic Variables describe the patients
  age, sex, race.
• C. Hospital level variables describe aspects of
  the characteristics of the hospital in which the
  patient was treated.
• D. treatment Variables describe aspects of the
  patients treatment during hospitalization.

5
Variables

• The Pediatric Asthma data found in the file
  Case05.txt can be read by the SAS input statement
  file
• INPUT LOS 1-2 HOSPITAL 4-6 PATIENT 8-12 INSURER
  14-15 SUMCOMRB 17-18 CMPL11 21 CMPL12 23 SEVMOD
  25 SEVSEV 27 HISTF01 29-30 HISTF02 32 HISTF03
  34-35 AGE 37-38 FEMALE 40 RACE 42 SUMSERV 44-46
  BEDSIZE 48 OWNER 50 TEACHCAT 52 ANYCOMP 54
  HIGHPOC 56 MEDPOC 58 LOWPOC 60 DIAGTSTS 62-63
  _at_65 FTETOBED 7.4 _at_73 PCTINS1 7.4
• A, B, C, D

6
Questions of interest

• Are there differences in LOS between the
  insurers?
• Do the differences hold up once you have
  accounted for differences in hospital and patient
  characteristics?
• What are the differences in patient severity
  characteristics, demographics, hospital
  characteristics, and treatment variables between
  the two insurers?

7
Analysis

• Frequency of the Variables FREQ
• Descriptive of Variables MEANS
• Correlation b/w LOS and Variables CORR
• Contingency table FREQ TABLES
• Regression REG