Skip to main content

Table 1 Coefficients of two-part model for alcohol consumption prediction

From: Modelling the consequences of a reduction in alcohol consumption among patients with alcohol dependence based on real-life observational data

  

Logistic regression model for probability of drinking

Negative binomial regression model for amount of alcohol consumed on drinking days

Parameter

Level

Estimate

Standard error

P-value

Odds ratio

Estimate

Standard error

P-value

Odds ratio

Intercept

 

−3.5123a

0.6325

<.0001

-

3.6665b

0.2766

<.0001

-

Age (years, ref = 0)

 

0.006217

0.01127

0.5813

1.0062

−0.01163

0.004807

0.0156

0.9884

Sex (ref = male)

Female

0.2055

0.2796

0.4624

1.2281

0.01766

0.1221

0.885

1.0178

Day (ref = Saturday)

Sunday

−0.05629

0.08452

0.5054

0.9453

−0.02609

0.01826

0.153

0.9742

Monday

−0.4179

0.0839

<.0001

0.6584

−0.05914

0.01834

0.0013

0.9426

Tuesday

−0.2794

0.08356

0.0008

0.7562

−0.0489

0.01838

0.0078

0.9523

Wednesday

−0.2098

0.08446

0.013

0.8107

−0.03008

0.01845

0.103

0.9704

Thursday

−0.05853

0.08508

0.4915

0.9431

−0.02463

0.01856

0.1846

0.9757

Friday

−0.03683

0.08563

0.6671

0.9638

0.004004

0.01865

0.83

1.004

Log (1 + consumption) day −1

 

0.5019

0.01368

<.0001

1.6519

0.07999

0.004127

<.0001

1.0833

Log (1 + consumption) day −2

 

0.361

0.01396

<.0001

1.4348

0.05808

0.00408

<.0001

1.0598

Log (1 + consumption) day −7

 

0.3777

0.01312

<.0001

1.4589

0.05105

0.003695

<.0001

1.0524

Treatment during follow-up (ref = No treatment)

Psychological

−0.1548

0.324

0.6328

0.8566

0.1968

0.1405

0.1615

1.2175

Psychological and pharmaceutical

−0.07805

0.58

0.893

0.9249

0.09311

0.2426

0.7012

1.0976

Time index

 

0.000461

0.000198

0.02

1.0005

0.02279

0.005067

<.0001

1.0231

Mean of baseline consumption per patient

 

0.000659

0.001405

0.6389

1.0007

0.01156

0.006173

0.061

1.0116

Standard deviation of baseline consumption per patient

 

0.00031

0.003629

0.9319

1.0003

0.0253

0.0153

0.0982

1.0256

Depressed (ref = No)

Yes

−0.1817

0.2674

0.4968

0.8339

0.1539

0.1148

0.1801

1.1664

Day 1

 

−3.9724

0.5882

<.0001

0.0188

−0.09164

0.1547

0.5537

0.9124

Day 2

 

−2.5372

0.5816

<.0001

0.0791

−0.1644

0.1721

0.3393

0.8484

Day 3

 

−0.475

0.386

0.2185

0.6219

−0.2108

0.1105

0.0565

0.8099

Day 4

 

−0.2679

0.4006

0.5036

0.765

−0.1374

0.1088

0.2067

0.8716

Day 5

 

−0.6315

0.4465

0.1573

0.5318

−0.3298

0.1207

0.0063

0.7191

Day 6

 

−1.0807

0.5291

0.0411

0.3394

−0.2246

0.1489

0.1314

0.7988

Day 7

 

−0.4915

0.5832

0.3994

0.6117

−0.2601

0.1488

0.0805

0.771

  1. aThe intercept estimate of the logistic regression is the coefficient from which the probability of drinking can be derived for the reference patient-day. On the reference patient-day, the probability of drinking is p = exp(3.5123/(1 + exp(3.5123)) = 0.0290
  2. bThe intercept estimate of the negative binomial model is the coefficient from which the mean amount of alcohol consumed can be derived for the reference patient-day. On the reference patient-day, the mean quantity of alcohol consumed, in grams, is c = exp(3.6665) = 39.11
Back to article page

BMC Public Health

ISSN: 1471-2458

Contact us