Exploratory Analysis of 2012 National Emergency Medical Services Information Systems (NEMSIS) Data to Derive a Drug/Alcohol Scoring Index (DOSI) to Predict Prehospital Survival Open Access

Ziolkowski, Lilia Maria Xepoleas (2017)

Permanent URL: https://etd.library.emory.edu/concern/etds/q811kk441?locale=en
Published

Abstract

Emergency Medical Services (EMS) are a vital and necessary resource for communities and are relied upon for a multitude of services. As drug overdoses and poisonings continue to trend upward, the impact on EMS agencies increases. There are a limited number of data repositories dedicated to aggregate surveillance EMS data at local, state, and national levels.

2012 NEMSIS data was used to create a screening tool to predict the survivability of a drug/alcohol poisoning or overdose (DAPO) event when EMS services are requested. The screening tool, Drug Overdose Scoring Index (DOSI) was derived from candidate variables. The DOSI screening tool was tested and validated with a randomly selected set of cases from the 2012 NEMSIS case records.

DOSI variables were selected if P<0.05 in final regression model; 911 call, EMS level, gender, and EMS Time@scene. AUC's of 0.794 (P<0.001, 95% CI: 0.773, 0.816) and 0.802 (P<0.001, 95% CI: 0.773, 0.816) were reported, with good discriminative ability by ROC analysis. The DOSI threshold score = 157 with showed 82% sensitivity 68% specificity.

General dataset characteristics indicated: overall mortality was < 1% for all EMS responses; mortality from drug/alcohol related events was 3.25% (N=1,092,509). Whites were less likely to survive. 20-29 year olds (N=222,490) had highest number of DPI(+) cases. Only 21.8% of 911 calls reported drug/alcohol use. Residence was the most common incident location at 53% of EMS response calls. Nearly 81% of DPI(+) patients were treated and transported.

NEMSIS reporting is voluntary; generalizations beyond the dataset population cannot be made, limiting utility. 911 calls were underreported since patient health status may be unknown or withheld; specific drugs and amounts taken were not available, so associations or influences could not be assessed. At local, regional, or state levels, DAPO data may be useful in matching EMS resources to community needs. As one of the largest surveillance repositories for pre-hospital events, many other areas of public health in addition to the overdose epidemic can be explored.

Table of Contents

TABLE OF CONTENTS
INTRODUCTION 1
OVERVIEW OF EMERGENCY MEDICAL SERVICES (EMS) 1
EMS RESPONSE AND DRUG OVERDOSE 3

OVERDOSE EVENTS 4

ALCOHOL POISONING 6

POLYPHARMACY 7

MEDICAL AMNESTY AND GOOD SAMARITAN LAWS 8

NATIONAL EMERGENCY MEDICAL INFORMATION SYSTEMS (NEMSIS) DATA 8
PROBLEM STATEMENT 11
PURPOSE STATEMENT 13
METHOD 14
NEMSIS DATA & POPULATION CHARACTERISTICS 14

DRUG OVERDOSE SCORING INDEX (DOSI) 15

RECEIVER OPERATOR CHARACTERISTIC (ROC) CURVE 15
RESULTS 17
DPI(+) DEMOGRAPHIC DATASET CHARACTERISTICS 17

PEARSON'S CHI-SQUARE TEST OF ASSOCIATION 18

LOGISTIC REGRESSION 19

ROC CURVE DERIVATION AND VALIDATION 19
DISCUSSION 21
INTRODUCTION 21
OVERVIEW OF DPI(+) CHARACTERISTICS 23
EMS OUTCOME MEASURES 24
DOSI SCREENING TOOL 26
ROC CURVE 26
STUDY DESIGN LIMITATIONS 27
NEMSIS DATA COLLECTION REVISION 31
OVERDOSE SURVEILLANCE THROUGH NEMSIS 31
CONCLUSION 34
REFERENCES 38
APPENDIX 42
DOCUMENT 1A & 1B. NEMSIS DATA ELEMENTS AND DATA DICTIONARY
TABLE 1. COMMON OVERDOSE DRUGS AND EMS TREATMENT
TABLE 2. 2012 NEMSIS DATASET CHARACTERISTICS
TABLE 3. CHI SQUARE ANALYSIS OF CANDIDATE RISK FACTORS
TABLE 4. DRUG OVERDOSE SCORING INDEX (DOSI)
TABLE 5. CALCULATED DOSI SCORES FOR DPI(+) CASES
FIGURES 1A & 1B: HISTOGRAM OF DOSI CASE FREQUENCIES
FIGURE 2A & 2B: ROC CURVES FOR PREDICTIVE & VALIDATION MODEL

About this Master's Thesis

Rights statement
  • Permission granted by the author to include this thesis or dissertation in this repository. All rights reserved by the author. Please contact the author for information regarding the reproduction and use of this thesis or dissertation.
School
Department
Degree
Submission
Language
  • English
Research field
Keyword
Committee Chair / Thesis Advisor
Committee Members
Partnering Agencies
Last modified

Primary PDF

Supplemental Files