Skip to main content

Table 2 Generalized linear model using a Poisson link function to predict the recovery time from severe OHSS

From: Predictive factors for recovery time in patients suffering from severe OHSS

Parameter

Pregnant patients (n = 80)

Non-pregnant patients (n = 121)

 

ß

Standard error (ß)

95% CI

p

ß

Standard error (ß)

95% CI

p

   

Lower

Upper

   

Lower

Upper

 

(Intercept)

1.061

0.5915

-0.099

2.220

0.073

2.425

0.5608

1.326

3.542

<0.0001

Use of cabergoline

-0.182

0.3329

-0.834

0.471

0.585

-0.013

0.0918

-0.193

0.167

0.890

Late onset OHSS

0.282

0.2364

-0.181

0.746

0.233

-0.009

0.2914

-0.580

0.562

0.976

Polycystic ovary syndrome

0.334

0.1335

0.073

0.596

0.012

-0.025

0.1528

-0.274

0.325

0.868

Use of hCG for ovulation induction

0.222

0.1389

0.002

0.495

0.048

0.015

0.0941

-0.169

0.200

0.872

Number of retrieved oocytes (n)

0.015

0.0089

-0.002

0.033

0.084

0.004

0.0072

-0.010

0.018

0.603

Year of COH/OHSS treatment: 1996-2007

0.528

0.3666

-0.190

1.247

0.149

-0.071

0.0984

-0.261

0.128

0.482

Use of long protocol

0.053

0.1263

-0.194

0.301

0.673

-0.047

0.1014

-0.246

0.151

0.641

Age (years)

0.013

0.0110

-0.008

0.035

0.230

-0.003

0.0088

-0.020

0.014

0.740

Body mass index (kg/m 2 )

0.008

0.0086

-0.009

0.025

0.360

-0.006

0.0102

-0.026

0.014

0.585

Multiple pregnancy (n)

0.195

0.1077

-0.016

0.406

0.060

-

-

-

-

-

Time between oocyte retrieval and hospitalization for OHSS (days)

0.009

0.0180

-0.027

0.044

0.636

-0.020

0.0212

-0.061

0.022

0.353

  1. Coefficient estimates β, standard error se(β), 95%-confidence interval (CI) = β ± zα/2 se(β), and corresponding p-value are summarized in the table. The categorical factors are compared to the missing category, i.e., no use of cabergoline, early onset OHSS, no polycystic ovary syndrome, use of other substances than hCG analogues for ovulation induction, year of treatment 2008-2010, controlled ovarian hyperstimulation with other protocols than long protocol, and no multiple pregnancy. In the analysis of non-pregnant patients, the parameter “year of treatment” had to be excluded for this analysis, since it was redundant to the parameter “use of long protocol”. Statistically significant p-values are printed in bold numbers.