Skip to main content

Table 2 Abridged a Model Output from Multiple Linear Regression Models for resource intensive healthcare outcomes

From: Can diverse population characteristics be leveraged in a machine learning pipeline to predict resource intensive healthcare utilization among hospital service areas?

Variable

Coefficientc

Standard Error

Z Statistic

P Value

Percent Changec

Emergency Room Visitsb

% Employees whose commute time to work is between 30–59 min

-0.001

0.000

-4.340

0.000

-0.10%

% Children without a usual place of health care

-0.151

0.038

-3.918

0.000

-14.0%

% Employees whose commute method to work is walking

-0.004

0.001

-3.736

0.000

-0.40%

% of school aged enrolled in private grades 1–4

-0.004

0.001

-3.732

0.000

-0.40%

% Adults never visited doctor

0.659

0.192

3.441

0.001

93.3%

Inpatient Daysb

% Employed within health care or social assistance jobs

0.009

0.002

5.639

0.000

0.90%

% Children with food allergies

-0.320

0.063

-5.083

0.000

-27.4%

% Children whose last dentist visit was more than 5 years ago

0.103

0.027

3.819

0.000

10.8%

% Children whose last health care professional visit was 6 months ago or less

-0.204

0.054

-3.792

0.000

-18.5%

% of population not paying cash for rent

-0.006

0.002

-3.621

0.000

-0.60%

Hospital Expendituresb

% Employees whose commute time to work is less than 15 min

0.003

0.000

8.630

0.000

0.30%

% Employed within health care or social assistance jobs

0.007

0.001

6.537

0.000

0.70%

Expenditures on men’s nightwear ($/capita)

-0.611

0.101

-6.021

0.000

-45.7%

% Male population 15 + who never married

0.000

0.000

-5.204

0.000

0.0%

% Employed within agriculture, forestry, fishing, or hunting jobs

0.004

0.001

5.091

0.000

0.40%

  1. aModel output provided for top 5 variables based on absolute value of T statistic
  2. bExpressed as annual 2017 log per capita values
  3. cPercent change in non-log transformed outcome, the sign of associated coefficient indicates direction of change