11 میں سے 1

LOGISTIC REGRESSION

11 میں سے 2

LOGISTIC REGRESSION

Set of independent variables

Categorical outcome measure, generally dichotomous

11 میں سے 3

LOGISTIC REGRESSION

  • Which variables affect the probability of a certain outcome?
    • Produces odds ratios that aid interpretation

​

11 میں سے 4

Methods

  • Discriminant analysis - least squares
  • Logistic regression - maximum-likelihood method
    • coefficients make observed results most likely
    • non-linear
    • iterative
    • data assume S-shaped curve

11 میں سے 5

ODDS

  • Based on Probabilities
  • Probability of occurrence/probability of nonoccurrence
    • Probability of developing lung cancer/probability of not developing lung cancer
    • Can calculate the odds of developing lung cancer for smokers and for nonsmokers

​

11 میں سے 6

ODDS RATIO

  • Ratio of one probability to the other
  • Ratio of odds of developing lung cancer for smokers vs the odds of developing lung cancer for nonsmokers
  • OR= 1: 0.40

11 میں سے 7

SPSS - Logistic Regression

  • ANALYZE
    • Regression
      • Binary Logistic

​

      • Options
        • Classification Plots
        • Hosmer-Lemeshow goodness of fit
        • Casewise listing of residuals
        • Iteration history
        • CI for exp B

11 میں سے 8

Example from real study

    • Logistic regression analysis
  • This section examines the effects of group variable together with other variables on PU incidence. It also explores the relationship between the recorded variables and PU incidence. It includes logistic regression procedures and the findings.

​

11 میں سے 9

  • It is a procedure that uses maximum likelihood estimation for analysing the relationship between multiple independent variables and a dependent variable that is categorically measured on a nominal scale as mentioned by Polit (1996).

​

11 میں سے 10

  • The logistic regression procedure included entering predictors in the analysis process. The binary logistic regression was chosen as it is suited models where the dependent variable is dichotomous. Nominal predictor variables (recorded variables) were coded (dummy coding) as dichotomous variables. The following variables were transformed and re-coded as different variables in dichotomous level (0, present and 1, not present) before entering in the logistic regression analysis:

​

  • 1) Diagnoses which transformed into medical, surgical, oncology, renal, neuro-surgery and rehabilitation variables.
  • 2) Protective mattresses which transformed into Standard, Therakair, Alternating, Genadyne, Atmosair, Clinitron, Gel, and Water variables.
  • 3) The Braden scale was recoded into reversed categories where the high scores coded as high risk and the lower scores represented low risk, while group, age, gender, CJ, turning schedule, skin barrier creams and vitamins variables were left in the same coding. PU incidence was the dependent variable. It was encoded into (0, not developed PU and 1, developed PU).

​

11 میں سے 11

OR

* P Value

Wald x²

B

Predictor Variable

1.132

0.003

8.920

0.124

Age

1.181

0.046

3.986

0.166

CJ scores

1.159

0.001

11.557

0.147

Reversed Braden scores

1.814

0.005

7.818

0.595

Standard mattress

0.550

0.003

9.036

- 0.597

Skin barrier creams

2.669

0.016

5.759

0.982

Neuro-surgical diagnosis

Logistic regression analysis: Predictors of PU development in the sample

N= 719

The analysis report included Wald x², B coefficient estimation associated with each predictor, P value, and Odds Ratio (OR) to provide estimated relative risk. Logistic regression analysis was performed at α = 0.05 level of significance.