Skip to main content

Table 2 Variables included in risk models and TRIPOD classification of studies examining model performance

From: Risk prediction models for colorectal cancer in people with symptoms: a systematic review

Author, year

TRIPOD

levela

Demographic variables

Personal and Family Medical History

Symptoms

Signs

Investigations

Age

Sex

Smoking

Alcohol

Other

Family history of GI cancer

Other

Rectal bleeding

Change in bowel habit

Diarrhoea

Constipation

Abdominal pain

Weight loss

Loss of appetite

Mucous

Other

Abnor mal rectal examination

Other

Haemoglobinb

MCV

FOBT

Other

Models predicting gastrointestinal cancers and neoplasms

Adelstein, 2010 [32]

1b

    

Colonoscopy in last 10 years; history of diverticular disease, NSAID use, or aspirin use.

   

  

Anaemiab.

      

Adelstein, 2011 [31]

1a

 

Education level.

 

Colonoscopy in last 10 years; history of colorectal polyps, IBS, NSAID use or aspirin use.

      

Anaemiab; fatigue.

      

Fijten, 1995 [21]

1a, 4

      

d

             

Hamilton, 2005 [22]

1a

       

 

 

   

Abdominal tenderness

 

Blood glucose

Hamilton, 2009 [23]

1a

       

     

  

Hippisley-Cox, 2012 (Male) [25]

2a, 4

N/A

 

 

 

  

    

   

Hippisley-Cox, 2012 (Female) [25]

2a, 4

N/A

   

 

   

    

   

Hurst, 2007 [28]

1a

        

        

sMMP-9

Lam, 2002 [20]

1a

      

e

      

       

Mahadavan, 2011 [29]

1a

     

           

eDNA; CEA

Marshall, 2011 (BB equation) [24]

3

       

f

f

f

   

Abdominal mass

 

Marshall, 2011 (CAPER score) [24]

4

         

   

Abdominal mass

  

 

Selvachandran, 2002 (WNS) [30] c

4, 4, 4, 4

    

‘Family history’, ‘relevant medical history’.

   

 

Tenesmus; urgency; incomplete emptying; perianal symptoms; ‘abdominal symptoms’; tiredness.

      

Models predicting cancers of multiple organ systems alongside colorectal cancer

Hippisley-Cox, 2013 (Male) [27]

2a

N/A

BMI; Townsend deprivation score.

History of chronic pancreatitis, type 2 diabetes, or COPD; family history of prostate cancer.

 

 

Abdominal distension; heartburn; indigestion; dysphagia, haematemesis; haematuria; haemoptysis; neck lump; Night sweats; testicular lump; testicular pain; first occurrence of a venous thromboembolism; bruising; cough; impotence; nocturia; urinary frequency; urinary retention.

  

   

Hippisley-Cox, 2013 (Female) [26]

2a

N/A

BMI; Townsend deprivation score.

History of chronic pancreatitis, type 2 diabetes, COPD, or endometrial hyperplasia/polyps; family history of breast cancer or ovarian cancer.

 

 

Abdominal distension; heartburn; indigestion; dysphagia; haematemesis; rectal bleeding; haematuria; haemoptysis; neck lump; weight loss; night sweats; breast lump; breast pain; nipple discharge or breast skin changes; inter-menstrual bleeding; post-menopausal bleeding; post-coital bleeding; first occurrence of a venous thromboembolism; bruising; cough.

  

   
  1. aTypes of prediction model studies for each model defined according to the TRIPOD [17] guidelines. 1a – Development only; 1b – Development and validation using resampling; 2a – Random split-sample development and validation; 2b – Nonrandom split-sample development and validation; 3 – Development and validation using separate data; 4 – Validation study
  2. bIf anaemia was defined by a haemoglobin value, it was considered an investigation. Self-report of anaemia in the absence of a blood test was considered a symptom
  3. cSelvachandran 2002 [30] describes a copyrighted model, the Weighted Numerical Score (WNS) that is incompletely reported
  4. dspecifically blood mixed with/on stool
  5. especifically rectal bleeding independent of straining or defaecation, blood clots, or dark red blood
  6. fin addition to documentation of these symptoms, prescriptions used as a proxy. Laxative prescriptions taken to indicate constipation, anti-diarrhoeal prescriptions diarrhoea, and antispasmodic prescriptions abdominal pain