Skip to main content

Table 3 First 25 Predictors of ≥3 emergency department visits in the past 12 months based on 2015–2018 data

From: Predictive model of multiple emergency department visits among adults: analysis of the data from the National Survey of Drug Use and Health (NSDUH)

Variable Name

Reference Cell

Odds Ratio in a stepwise regression

Order in Stepwise regression based on formal p-value of first category occurrence

Order in Random Forest Model Importance

Fair/Poor Health

Excellent/Very Good/Good Self-Reported Heath

2.63

1

1

Income <$25,000

Income = > $75,000

2.53

2

2

Female Gender

Male

1.62

9

3

Asthma (Current)

No Asthma

1.87

8

4

All/Most of the Time of the Time Feeling Restless or Fidgety in Past Year

Some/Little/None of the Time

1.47

16

5

18–20 Years Old

Age 50–64 Years Old

2.30

4

6

Black/African American Race

White Race

1.91

7

7

Heart Condition in Past Year

No Heart Condition

2.30

5

8

All/Most of the Time of the Time Feeling Nervous in Past Year

Some/Little/None of the Time

1.28

26

9

Education Level: 10th–12th grades completed

6th grade or less completed

1.43

17

10

Diabetes (Ever)

No

1.62

10

11

Nicotine Dependence (Past Month)

No

1.38

19

12

Marital Status: Divorced, Separated, or Widowed

Married

1.16

33

13

COPD/Chronic Bronchitis (Ever)

No

1.48

15

14

Tobacco Use (Past Year)

No

1.29

23

15

Major Depressive Episode (Past Year)

No

1.23

25

16

No High Blood Pressure or Not Taking Medication for High Blood Pressure

Ever had High Blood Pressure or Currently Taking Medication

1.29

17

17

Illicit Drug Use (Past Year)

No

1.33

21

18

Alcohol Use (Past Year)

No

0.81

29

19

Illicit Drug Use Disorder (Past Year)

No

20

Sexually Transmitted Disease (Past Year)

No

1.58

11

21

Pain Reliever Misuse (Past Year)

No

1.29

24

22

Had Insurance Coverage

No Coverage

1.19

32

23

Pain Reliever Use Disorder (Past Year)

No

1.49

13

24

Population Density: Segment not in a CBSA

Segment is in a CBSA

1.25

28

25

  1. Note: In random forest, variable importance is evaluated with respect to the variable as a whole, rather than a specific category. In fact, the variable categories are randomly “scrambled.” Thus, only one ranking value is issued for the variable. COPD Chronic obstructive pulmonary disease, CBSA Core Based Statistical Area
  2. — The variable/category is not present in the top category list or was not selected by the stepwise algorithm