ColoRectal Cancer Predicted Risk Online (CRC-PRO) Calculator Using Data from the Multi-Ethnic Cohort Study

Brian J. Wells; Michael W. Kattan; Gregory S. Cooper; Leila Jackson; Siran Koroukian

doi:10.3122/jabfm.2014.01.130040

Article Figures & Data

Figures

Tables

Download figure
Open in new tab
Figure 1A.
Nomogram for predicting colorectal cancer risk in men. Instructions: Draw a perpendicular line from the patient's age to the “points” axis and record the value. Repeat this process for the remaining variables and tally. The 10-year risk of colorectal cancer (CRC) is identified where a line drawn straight down from the “total points” axis intersects the “10-year risk of CRC (%).” Please note that the “years of education” variable has a U-shaped relationship with the 10-year risk of CRC. That is, the lowest risk of CRC occurs at 8 years and increases as you move along the top of the axis from left to right until reaching the highest risk at 14 years, and then it decreases along the bottom of the axis as you move to the left from 14 to 16 years.
Download figure
Open in new tab
Figure 1B.
Nomogram for predicting colorectal cancer risk in women. NSAIDS, nonsteroidal anti-inflammatory drugs.
Download figure
Open in new tab
Figure 2.
Calibration curve for the prediction in men (A) and women (B). The calibration curves were created by plotting the mean predicted risk of colorectal cancer in each quintile of risk on the x-axis against the corresponding Kaplan-Meier (K-M) estimated incidence in the same quintile. Error bars reflect the 95% confidence interval around the K-M estimate. The risks are displayed as probabilities. Perfect calibration would fall directly on the 45-degree line.

Tables

Figures

View popup

Table 1. Differences between Variable Definitions in the Freedman (aka National Cancer Institute) Calculator and the Multiethnic Cohort Study

Variable	National Cancer Institute Definition	Multiethnic Cohort Study
Estrogen	Asks about estrogen use in the past 2 years. Also asks about menopausal status and allows different estrogen effects depending on menopausal status	Does not specifically ask about estrogen use in the past 2 years
Vegetable intake	“In the past 30 days, about how many servings per week of vegetables or leafy green salads did you eat?” (cups)	Vegetables quantified according to grams per day based on detailed food frequency questionnaire
Family history of colorectal cancer	Number of first-degree relatives with a history of cancer of the colon or rectum	Family history of colon cancer in first-degree relative (yes or no)
Activity	Defined according to the number of hours of activity in the past week, but also requires detailed information about how many months the participant was active in the past year	Only asked about current activity level
Nonsteroidal anti-inflammatory drugs and aspirin	Regular use defined as 3 times per week	Regular use was defined as twice per week

View popup

Table 2A. Descriptive Statistics for Men by Colorectal Cancer Outcome in the Multiethnic Cohort Study (n = 80,062)

Characteristics	Developed Colorectal Cancer		Patients with Missing Data
Characteristics	No (n = 78,576)	Yes (n = 1,486)	Patients with Missing Data
Mean age, years (SD)^*	59.8 (8.9)	64.2 (7.8)	0 (0.0)
Race/ethnicity^*			0 (0.0)
Black	11,202 (14.3)	231 (15.5)
Hawaiian	5,565 (7.1)	105 (7.1)
Japanese	22,247 (28.3)	558 (37.6)
Latino	19,879 (25.3)	301 (20.3)
White	19,683 (25)	291 (19.6)
Mean pack-years smoking (SD)^*	13.9 (16.6)	17 (18.4)	3,088 (3.9)
Alcohol, mean drinks/day (SD)^*^{^†}	1.0 (2.3)	1.3 (2.6)	0 (0.0)
Mean years of education (SD)^*	13.2 (3.4)	13 (3.1)	884 (1.1)
Family history of colon cancer^*	6,030 (7.7)	156 (10.5)	11,015 (13.8)
Mean body mass index (SD)	26.6 (4.1)	26.6 (4.2)	753 (0.9)
Regular use of aspirin^{^‡}			7,439 (9.3)
No	46,001 (58.5)	901 (60.6)
Yes, not currently	13,795 (17.6)	255 (17.2)
Yes, currently	18,780 (23.9)	330 (22.2)
Regular use of multivitamins^*^{^§}	37,395 (47.6)	627 (42.2)	4,189 (5.2)
Diabetes^*	9,749 (12.4)	214 (14.4)	0 (0.0)
Mean hours of moderate activity per day (SD)^*	1.3 (1.5)	1.2 (1.4)	3,590 (4.5)
History of cancer	7,038 (9.0)	130 (8.7)	1 (<0.1)
Preference for well-done meat	34,122 (43.4)	625 (42.1)	2,327 (2.9)
Mean intake of red meat per day, oz (SD)^*	2.6 (2.1)	2.5 (1.9)	0 (0.0)
Regular use of NSAIDs^*^{^†}			10,214 (12.8)
No	57,852 (73.6)	1,146 (77.1)
Yes, not currently	12,924 (16.4)	207 (13.9)
Yes, currently	7,800 (9.9)	133 (9)

Data are n (%) unless otherwise indicated. The Multiethnic Cohort study enrolled an ethnically diverse mix of residents from Hawaii and California between 1993 and 1996.
↵* Statistically significant in univariate analysis (P < .05).
↵† One alcoholic drink is defined as 1 oz of alcohol, which is approximately equivalent to one 12 oz beer, one 4-oz glass of wine, or 1 shot of liquor.
↵‡ At least twice a week for 1 month or longer.
↵§ At least once per week for the past year.
NSAIDS, nonsteroidal anti-inflammatory drugs; SD, standard deviation.

View popup

Table 2B. Descriptive Statistics for Women by Colorectal Cancer Outcome (n = 100,568)

Characteristics	Developed Colorectal Cancer		Patients with Missing Data
Characteristics	No (n = 99,292)	Yes (n = 1,276)	Patients with Missing Data
Mean age, years (SD)^*	59.5 (8.8)	64 (7.9)	0 (0.0)
Race/ethnicity^*			0 (0.0)
Black	19,694 (19.8)	347 (27.2)
Hawaiian	7,464 (7.5)	77 (6)
Japanese	26,500 (26.7)	413 (32.4)
Latino	21,730 (21.9)	194 (15.2)
White	23,904 (24.1)	245 (19.2)
Mean pack-years smoking (SD)^*	6.6 (12.1)	7.4 (12.8)	3,739 (3.7)
Alcohol intake (mean drinks/day [SD])^*^{^†}	0.3 (1.1)	0.4 (1.6)	0 (0.0)
Mean years of education (SD)	13.0 (3.3)	12.8 (2.9)	1,255 (1.2)
Family history of colon cancer^*	9,165 (9.2)	174 (13.6)	12,707 (12.6)
Mean body mass index (SD)	26.4 (5.5)	26.6 (5.7)	2,346 (2.3)
Regular use of aspirin^{^‡}			4,971 (4.9)
No	61,593 (62)	806 (63.2)
Yes, not currently	19,193 (19.3)	234 (18.3)
Yes, currently	18,506 (18.6)	236 (18.5)
Regular use of multivitamins^*^{^§}	53,596 (54)	625 (49)	2,649 (2.6)
Diabetes^*	10,910 (11)	188 (14.7)	0 (0.0)
Mean hours of moderate activity/day (SD)	1.1 (1.2)	1 (1.2)	2,420 (2.4)
History of cancer	11,576 (11.7)	151 (11.8)	0 (0.0)
Preference for well-done meat^*	53,082 (53.5)	739 (57.9)	1,287 (1.3)
Mean intake of red meat per day, oz (SD)^*	1.7 (1.6)	1.6 (1.4)	0 (0.0)
Regular use of NSAIDs^*^{^‡}			6,488 (6.2)
No	62,132 (62.6)	849 (66.5)
Yes, not currently	21,284 (21.4)	266 (20.8)
Yes, currently	15,876 (16)	161 (12.6)
Estrogen use^*^{^‖}			3,277 (3.1)
No	53,754 (54.1)	709 (55.6)
Yes, not currently	17,752 (17.9)	278 (21.8)
Yes, currently	27,786 (28)	289 (22.6)

Data are n (%) unless otherwise indicated. The Multiethnic Cohort study enrolled an ethnically diverse mix of residents from Hawaii and California between 1993 and 1996.
↵* Statistically significant in univariate analysis (P < .05).
↵† One alcoholic drink is defined as 1 oz of alcohol, which is approximately equivalent to one 12 oz beer, one 4 oz glass of wine, or one shot of liquor.
↵‡ At least twice a week for 1 month or longer.
↵§ At least once per week for the past year.
↵‖ Estrogen use was defined as female hormones administered by pill, injection, or patch for menopause or other reasons.
NSAID, nonsteroidal anti-inflammatory drug; SD, standard deviation.

View popup

Table 3A. Variable Impact on the Apparent C-Statistic for the Colorectal Cancer Model in Men

Variable	C-Statistic	Change in C-Statistic	Variables (n)
No Model	0.5	—	0
Age	0.663280	0.163280	1
Race/ethnicity	0.672989	0.009710	2
Pack-years of smoking	0.678214	0.005224	3
Alcoholic drinks per day	0.681442	0.003229	4
Body mass index	0.684263	0.002821	5
Years of education	0.686596	0.002333	6
Regular use of aspirin	0.688405	0.001809	7
Family history of colon cancer	0.689931	0.001526	8
Regular use of multivitamins	0.691143	0.001212	9
Red meat intake (oz) per day	0.691765	0.000622	10
History of diabetes	0.692365	0.000600	11
Moderate physical activity per day (hours)	0.692879	0.000513	12
History of cancer^*	0.693176	0.000297	13
Regular use of NSAIDs^*	0.693397	0.000221	14
Preference for well-done meat^*	0.693538	0.000141	15 (Full)

The outcome was colorectal cancer.
↵* The variable was excluded from the final model because of the relatively small amount of additional prediction accuracy that it could have contributed to the model. The final model, which included 12 variables, had an apparent c-statistic of 0.6929, which was within 0.001 of the accuracy associated with the c-statistic of the full model, which was 0.6935.
NSAIDS, nonsteroidal anti-inflammatory drugs.

View popup

Table 3B. Variable Impact on the Apparent C-Statistic for the Colorectal Cancer Model in Women

Variable	C-Statistic	Change in C-Statistic	Variables (n)
No Model	0.5	—	0
Age	0.657655	0.157655	1
Race/ethnicity	0.667666	0.010010	2
Years of education	0.670909	0.003243	3
Use of estrogen	0.674048	0.003139	4
History of diabetes	0.676648	0.002600	5
Pack-years of smoking	0.679123	0.002476	6
Family history of colon cancer	0.680963	0.001839	7
Regular use of multivitamins	0.682497	0.001534	8
Body mass index	0.683796	0.001300	9
Regular use of NSAIDs	0.684893	0.001097	10
Alcoholic drinks per day	0.685869	0.000976	11
Preference for well-done meat^*	0.686268	0.000399	12
Moderate physical activity per day (hours)^*	0.686525	0.000257	13
Regular use of aspirin^*	0.686790	0.000266	14
Red meat intake per day (oz)^*	0.686869	0.000078	15
History of cancer^*	0.686865	-0.000003	16 (Full)

The outcome was colorectal cancer.
↵* The variable was excluded from the final model because of the relatively small amount of additional prediction accuracy that it could have contributed to the model. The final model, which included 11 variables, had an apparent c-statistic of 0.6859, which was within 0.001 of the accuracy associated with the c-statistic of the full model, which was 0.6869.
NSAIDS, nonsteroidal anti-inflammatory drugs.

View popup

Table 4A. Adjusted Multiple Regression Analysis for Colorectal Cancer in Men

Variable	Comparison Groups^*	Adjusted Hazard Ratio
Variable	Comparison Groups^*	Point Estimate	95% CI
Age (years)	67 vs 52	3.03	2.69–3.41
Diabetes	Yes vs no	1.12	0.96–1.30
Regular multivitamin use	Yes vs no	0.83	0.74–0.92
Family history of colon cancer	Yes vs no	1.27	1.08–1.51
Education (years)	16 vs 12	0.94	0.85–1.03
Race/ethnicity	Black vs white	1.18	0.99–1.42
	Hawaiian vs white	1.39	1.10–1.75
	Japanese vs white	1.52	1.31–1.77
	Latino vs white	1.03	0.86–1.23
Body mass index (kg/m²)	28.7 vs 23.8	1.12	1.04–1.21
Alcoholic drinks per day (n)	1.15 vs 0	1.26	1.13–1.41
Moderate activity per day (hours)	1.6 vs 0.4	0.92	0.83–1.02
Regular aspirin use	Not currently vs no	0.97	0.84–1.11
	Yes vs no	0.81	0.71–0.92
Pack-years smoking (n)	19.8 vs 0	1.06	0.93–1.21

↵* Continuous variables were modeled using restricted cubic splines with 3 knots. Comparison groups for the continuous variables are based on the 75th percentile versus the 25th percentile.
CI, confidence interval.

View popup

Table 4B. Adjusted Multiple Regression Analysis for Colorectal Cancer in Women

Variable	Comparison Groups^*	Adjusted Hazard Ratio
Variable	Comparison Groups^*	Point Estimate	95% CI
Age (years)	67 vs 52	2.81	2.48–3.18
Diabetes	Yes vs no	1.26	1.08–1.48
Regular use of multivitamins	Yes vs no	0.85	0.76–0.95
Family history of colon cancer	Yes vs no	1.37	1.17–1.61
Education (years)	14 vs 12	1.01	0.97–1.06
Race/ethnicity	Black vs white	1.41	1.18–1.67
	Hawaiian vs white	1.10	0.85–1.44
	Japanese vs white	1.43	1.20–1.70
	Latino vs white	0.95	0.77–1.17
Body mass index (kg/m²)	29.3 vs 22.5	1.10	1.00–1.21
Alcoholic drinks per day (n)	0.1 vs 0	0.99	0.93–1.06
Regular use of NSAIDs	Not currently vs no	0.95	0.83–1.10
	Yes vs no	0.79	0.66–0.94
Pack-years of smoking (n)	8.9 vs 0	1.20	1.05–1.38
Estrogen use	Not currently vs no	0.96	0.83–1.10
	Yes vs no	0.78	0.68–0.90

↵* Continuous variables were modeled using restricted cubic splines with 3 knots. Comparison groups for the continuous variables are based on the 75th percentile versus the 25th percentile.
CI, confidence interval; NSAIDS, nonsteroidal anti-inflammatory drugs.

View popup

Table 5A. Coefficients for the Model Predicting Colorectal Cancer in Men

Variables	Coefficient	Standard Error
Age^*
Linear component	0.0917	0.0106
Nonlinear component	−0.0234	0.0106
Diabetes	0.1102	0.0758
Regular use of multivitamins	−0.1918	0.0532
Family history of colon cancer	0.2425	0.0850
Years of education^*
Linear component	0.0721	0.0181
Nonlinear component	−0.0734	0.0193
Race/ethnicity
Hawaiian	0.1609	0.1208
Japanese	0.2535	0.0820
Latino	−0.1366	0.0925
White	−0.1673	0.0921
Body mass index^*
Linear component	0.0180	0.0162
Nonlinear component	0.0090	0.0194
Alcoholic drinks per day^*
Linear component	0.2838	0.0752
Nonlinear component	−1.7375	0.5356
Hours of moderate activity per day^*
Linear component	−0.0907	0.0702
Nonlinear component	0.1103	0.1496
Regular aspirin use
Yes, not currently	−0.0323	0.0719
Yes	−0.2096	0.0655
Pack-years of smoking^*
Linear component	0.0002	0.0059
Nonlinear component	0.0177	0.0170

↵* Continuous variables were modeled using restricted cubic splines with 3 knots. Therefore, continuous variables contain one coefficient for the linear component and a nonlinear component.

View popup

Table 5B. Coefficients for the Model Predicting Colorectal Cancer in Women

Variable	Coefficient	Standard Error
Age^*
Linear component	0.0900	0.0112
Nonlinear component	−0.0276	0.0114
Diabetes	0.2333	0.0821
Regular use of multivitamins	−0.1665	0.0568
Family history of colon cancer	0.3159	0.0820
Years of education^*
Linear component	0.0744	0.0204
Nonlinear component	−0.0757	0.0212
Race/ethnicity
Hawaiian	−0.2410	0.1284
Japanese	0.0140	0.0844
Latino	−0.3967	0.0977
White	−0.3406	0.0892
Body mass index
Linear component	0.0075	0.0146
Nonlinear component	0.0121	0.0177
Alcoholic drinks per day
Linear component	−0.0886	0.3046
Nonlinear component	0.4092	0.7886
Regular use of NSAIDs
Yes, not currently	−0.0464	0.0731
Yes	−0.2370	0.0887
Pack-years of smoking^{^†}
Linear component	0.0627	0.0498
Nonlinear component 1	−1.8515	2.1240
Nonlinear component 2	2.2999	2.6795
Estrogen use
Yes, not currently	−0.0443	0.0719
Yes	−0.2450	0.0724

↵* Continuous variables were modeled using restricted cubic splines with three knots. Therefore, continuous variables contain one coefficient for the linear component (displayed first) and a non-linear component (displayed in the next row and denoted with a single quotation mark).
↵† Unable to obtain a natural spline using 3 knots and therefore restricted cubic splines with 4 knots were used for this variable.
NSAIDS, nonsteroidal anti-inflammatory drugs.