Skip to main content

Table 3 Logistic regression analysis of postoperative outcomes of surgery

From: Chest closure without drainage after open patent ductus arteriosus ligation in Ugandan children: A non blinded randomized controlled trial

Outcome

All study (N = 62)

Drain (n = 31)

No drain (n = 31)

Unadjusted

Adjusted for significant variables at unadjusted analysis

OR

95 % CI

αP value

aOR

95 % CI

αP value

Primary endpoints

aPleural separation (48 h)

   ≤ 30 mm

61 (98.4)

31 (100)

30 (96.7)

1.000

   > 30 mm

1 (1.6)

0 (0.0)

1 (3.3)

1.0

     

aSurgical wound infection

  No

57 (91.9)

26 (83.8)

31 (100)

0.053

  Yes

5 (8.1)

5 (16.2)

0 (0)

      

 Oxygen amount (L/min)

   ≤ 1

43 (69.4)

18 (58.1)

25 (80.7)

0.33

0.11–1.04

0.059

0.25

0.06–1.04

0.057

   > 1

19 (30.6)

13 (41.9)

6 (19.3)

1.0

     

 Combined primary endpoint

  Favorable

45 (72.5)

18 (58.1)

27 (87.1)

0.21

0.06–0.73

0.015

0.07

0.00–2.50

0.144

  Unfavorable

17 (27.5)

13 (41.9)

4 (12.9)

1.0

     

Secondary endpoint

 ICU stay (hours)

   ≤ 24

56 (90.3)

27 (87.1)

29 (93.5)

0.47

0.08–2.75

0.399

   > 24

6 (9.7)

4 (12.9)

2 (6.5)

1.0

     
  1. aZero values in cells did not permit generation of Odds Ratios. Fisher’s exact test was employed for hypothesis testing, since expected values <5, by Cochran’s rules;(#)- used in computing the “revised” combined endpoint (unfavorable outcomes were pleural separation > 30 mm, surgical wound infection & oxygen consumption >1 L/min; (α) – P-value cut-off for one interim analysis (after 50 % recruitment) and one final analysis (after 100 % recruitment) is 0.029 using the Fixed Nominal Rule (see Table 2 above)