Skip to main content

Table 4 Generalized linear model (GLM)-Poisson regression model

From: Medical expenses of urban Chinese patients with stomach cancer during 2002–2011: a hospital-based multicenter retrospective study

Parameter

Univariate analysis

Multivariate analysis

Relative risk (RR)

95% CI

RR

95% CI

Intercept

3036.012**

3033.281–3038.745

Age at diagnosis (45~ 54 VS < 45)

1.010**

1.010–1.010

1.029**

1.028–1.029

Age at diagnosis (55~ 64 VS < 45)

0.984**

0.984–0.985

1.006**

1.006–1.006

Age at diagnosis (65~ VS < 45)

1.042**

1.042–1.042

1.113**

1.112–1.113

Region (East VS West)

1.470**

1.470–1.471

1.071**

1.071–1.072

Region (Central VS West)

1.465**

1.464–1.465

1.116**

1.115–1.116

Hospital level (3A VS Non-3A)

1.588**

1.587–1.589

1.383**

1.382–1.383

Hospital type (Specialized VS General)

1.105**

1.105–1.106

1.013**

1.013–1.013

Drug proportion (%)

1.009**

1.009–1.009

1.009**

1.009–1.009

Clinical stage (II VS I)a

0.995**

0.995–0.995

0.952 **

0.952–0.952

Clinical stage (III VS I)b

1.123**

1.123–1.124

1.018**

1.018–1.018

Clinical stage (IV VS I)

1.087**

1.087–1.087

1.079**

1.079–1.079

Type of therapy (Surgery VS Palliative care)

2.323**

2.322–2.324

2.330**

2.328–2.331

Type of therapy (Surgery & Chemotherapy VS Palliative care)

3.223**

3.222–3.225

2.129**

2.128–2.130

Type of therapy (Surgery & Radiotherapy VS Palliative care)

2.990**

2.986–2.995

2.014**

2.011–2.017

Type of therapy (Chemotherapy VS Palliative care)

1.909**

1.908–1.910

1.436**

1.435–1.436

Type of therapy (Radiotherapy VS Palliative care)c

1.103**

1.102–1.104

1.166**

1.165–1.167

Type of therapy (Radiotherapy & Chemotherapy VS Palliative care)

2.758**

2.756–2.760

1.962**

1.961–1.963

Type of therapy (Others VS Palliative care)

0.856**

0.855–0.857

0.880**

0.880–0.882

Number of episodes per patient

1.201**

1.201–1.201

1.101**

1.101–1.101

Year (2003 VS 2002)

1.146**

1.146–1.147

1.107**

1.106–1.108

Year (2004 VS 2002)

1.328**

1.327–1.328

1.198**

1.197–1.199

Year (2005 VS 2002)

1.481**

1.480–1.482

1.316**

1.315–1.316

Year (2006 VS 2002)

1.391**

1.391–1.392

1.230**

1.229–1.230

Year (2007 VS 2002)

1.535**

1.534–1.535

1.273**

1.273–1.274

Year (2008 VS 2002)

1.614**

1.614–1.615

1.266**

1.266–1.267

Year (2009 VS 2002)

1.674**

1.673–1.675

1.415**

1.414–1.415

Year (2010 VS 2002)

1.792**

1.791–1.793

1.499**

1.499–1.500

Year (2011 VS 2002)

2.049**

2.048–2.050

1.579 **

1.578–1.580

Number of inpatient days per patient

1.008**

1.008–1.008

1.006 **

1.006–1.006

Accompanying diseases (Yes VS No)

1.220**

1.220–1.221

1.188**

1.188–1.188

  1. Note: a **P < 0.01
  2. b We include link = log. When we write our model out, log (μ) = β0 + β1x1 + ... + βpxp, where μ is the count we are modeling, and log ( ) defines the link function (i.e., how we transform μ to write it as a linear combination of the predictor variables)
  3. c Goodness of fit for the multivariate GLM model: deviance = 0.3437 (p ≈ 1), Pearson Χ2 = 1.255(p ≈ 1), so the goodness of fit for the model is very good