Skip to main content

Table 4 Comparison of results from original analysis with results from sensitivity analyses to assess the possibility of bias due to missing outcome data at follow up using two different methods

From: Suicidality in primary care patients who present with sadness and anhedonia: a prospective European study

 

Onset of depression (in non-depressed people at baseline)

Persistent depression (in depressed people at baseline)

Suicidal ideation

Suicide plans

Suicide attempts

Suicidal ideation

Suicide plans

Suicide attempts

OR (95 % CI) p value

OR (95 % CI) p value

OR (95 % CI) p value

OR (95 % CI) p value

OR (95 % CI) p value

OR (95 % CI) p value

Basic model

Adjusted for age, sex, lifetime depression and country

1.17 (0.58 to 2.37) 0.6561

3.05 (1.50 to 6.24) 0.0022

9.08 (2.57 to 32.03) 0.0006

2.93 (1.70 to 5.07) 0.0001

3.70 (2.08 to 6.57) < 0.0001

3.33 (1.47 to 7.54) 0.0040

Sensitivity analyses

Basic model further adjusted for important predictors of missing outcome data at follow up(educational level and alcohol dependency)

1.13 (0.56 to 2.30) 0.7278

2.93 (1.43 to 6.01) 0.0033

8.53 (2.41 to 30.15) 0.0009

2.95 (1.70 to 5.11) 0.0001

3.60 (2.02 to 6.41) < 0.0001

3.15 (1.38 to 7.16) 0.0063

Basic model weighted by inverse of probability of outcome data at follow up

1.10 (0.61 to 1.99) 0.7533

2.69 (1.40 to 5.16) 0.0029

7.50 (1.90 to 29.65) 0.0041

3.44 (1.74 to 6.79) 0.0004

3.87 (2.34 to 6.39) < 0.0001

3.39 (1.29 to 8.94) 0.0134

  1. Notes: All statistical models adjusted for age, sex, lifetime depression and country. Further adjusted models adjusted additionally for each specified confounder in turn. All analyses accounted for correlations between patients within the same health centre through inclusion in statistical models of a random effect of centre and for correlations between repeated measures within patients through inclusion of a random effect of individual