Determinants of non-participation in a mass screening program for colorectal cancer in Finland

Abstract

Background: For an effective colorectal cancer (CRC) screening program, high participation rate is essential. However, non-participation in CRC screening program has increased in Finland.

Material and methods: The study was based on a population-based nationwide cohort of persons invited for CRC screening in 2004–2011. Information on the first round of the CRC screening participation and related background factors was obtained from the Finnish Cancer Registry, and information about health behavior factors from the Health Behavior Survey (HBS) in 1978–1999. Non-participation in CRC screening was analyzed with Poisson regression as incidence rate ratios (IRR) with 95% confidence intervals (95% CI).

Results: Of all persons invited for CRC screening (79 871 men and 80 891 women) 35% of men and 21% of women refused. Of those invited for screening, 2456 men (3.1%) and 2507 women (3.1%) were also invited to the HBS. Persons, who declined HBS, were also more likely to refuse CRC screening (men IRR 1.40, 95% CI 1.26–1.56, women 1.75, 1.52–2.02) compared to HBS participants. Never married persons had about a 75% higher risk for refusing than married ones. The youngest age group (60 years) was more likely to refuse screening than the older age groups (62 or >64 years). Smoking was associated with non-participation in screening (current smokers, men: IRR 1.32, 95% CI 1.05–1.67, women: 2.10, 1.61–2.73).

Conclusions: Participation in CRC screening was affected by gender, age, and marital status. Persons, who refused the HBS, were also more likely to refuse CRC screening. Smoking was a risk factor for non-participation in CRC screening.

Colorectal cancer (CRC) screening has been shown to decrease CRC mortality by 16% in randomized trials [1]. CRC usually develops from precancerous polyps in the colon or rectum, which can be found with screening, and they can be removed before they turn into cancer [2].

Previous studies on determinants of participation in fecal occult blood test (FOBT) based screening have indicated that female gender [3], high education [4,5], high social status [6], and marriage [5,7] increase the probability of CRC screening participation.

Age [8], some diseases, such as ulcerative colitis, Crohn’s disease, and few inherited syndromes are known to be risk factors for CRC [9]. Personal lifestyle factors, such as low fiber and diet high in fat content, a sedentary lifestyle, obesity, smoking and heavy use of alcohol, are known to increase the risk for CRC, whereas a protective role of dietary fiber, physical activity and the use of non-steroidal anti-inflammatory drugs have also been reported [10–12].

The aim of this study was to evaluate determinants of non-participation in the CRC screening program in Finland. Additionally, we aimed to identify CRC risk factors that are related to low participation and could potentially bias the observed effects of screening.

Material and methods

Information on the CRC screening program was obtained from the Finnish Cancer Registry maintained by the Cancer Society of Finland. Data included the first ever invitation year, inviting municipality, marital status and age at invitation. Altogether 160 762 persons aged 60–64 years were invited for the first time for CRC screening in 2004–2011.

Additional data on health behavior determinants were obtained from the nationwide Finnish Health Behavior and Health among the Finnish Adult Population surveys (HBS) [13] from the years 1978 to 1999 (excluding the year 1985) organized by the National Public Health Institute, currently known as the National Institute for Health and Welfare (THL). These cross-sectional annual surveys were based on postal questionnaires mailed yearly to individual random samples of approximately 5000 residents of Finland (male 50%, female 50%) aged 15–64 years.

Nine factors [alcohol use, smoking status, body mass index (BMI), weekly intake of vegetables and berries/fruits, use of fat (on bread/in cooking), and physical activity (leisure and commuting)] were identified from the surveys to describe health behavior. As some of the survey questions changed over time, data on all of the factors were not available for the whole study period. Data on alcohol consumption was available in 1982–1999, smoking in 1983–1999, vegetable and berry use in 1979–1999, and leisure time physical activity in 1978–1993. Therefore, an ‘unknown’ category was included in all health behavior factors. It refers to both unknown and unavailable data.

Analyses on the effects of demographic factors on screening non-participation were conducted for all persons invited to the CRC screening (n = 160 762). Of these, 35% of men and 21% of women refused the invitation. Of the CRC screening invited persons, 2456 men (3.1%) and 2507 women (3.1%) were also invited to the HBS. HBS information on health behavior factors was finally available for 2.4% of CRC screening invited men and 2.6% of the women. Analyses of effects of health behavior factors on CRC screening non-participation were conducted for this group.

Non-participation in the first ever CRC screening was used as the outcome. Analyses were conducted in StataMP 12 [14] with the Poisson regression [incidence rate ratio (IRR) with 95% confidence intervals (CI)]. The Poisson regression was used instead of the logistic regression as non-participation was not a rare event.

In the analyses age at the time of screening invitation, the region, year of invitation and marital status were included in the baseline models as potential confounders for CRC screening non-participation.

The time between the HBS and CRC screening varied considerably. Therefore, a separate analysis including a weighting function on lag time was conducted: persons with a short lag were given more weight than those with a long lag [from one for those with the shortest lag towards null for those with the longest lag, range 2–33 years, weight =1/(1 + lag in 10 years)].

Differences in characteristics of non-participation in CRC screening were further assessed between HBS participants and HBS non-participants. There were 149 persons with two HBS invitations and two persons with three HBS invitations. These were handled as separate subjects in the analyses. The compilation of register data was conducted with permission from THL including ethical consideration.

Results

The non-participation rate of CRC screening increased in time from the year 2004 (reference 17.0%) to the year 2011 (40.6%; men IRR 2.00, 95% CI 1.82–2.21, women IRR 2.32, 95% CI 2.06–2.62) (Tables 1 and 2). Never married (men 1.74, 1.69–1.80, women 1.75, 1.67–1.84), had a clearly increased risk for refusing the screening in relation to married in both sexes. The youngest age group (60 years, reference) was slightly more likely to refuse screening than older age groups at invitation (62 years 0.93, 0.86–0.92 or ≥64 years 0.90, 0.86–0.92).

Table 1. Characteristics of persons invited for the first time to colorectal cancer screening (n = 160 762) by gender and participation, 2004–2011.

	Men (n = 79 871)				Women (n = 80 891)
	participation				participation
	No		Yes		No		Yes
Characteristics	n	%	n	%	n	%	n	%
Total	28 231	100	51 640	100	16 828	100	64 063	100
Year of screening invitation
2004	473	1.7	1670	3.2	300	1.8	2096	3.3
2005	2838	10.1	8808	17.1	1611	9.6	10 305	16.1
2006	3697	13.1	8764	17.0	1987	11.8	10 449	16.3
2007	3232	11.5	7528	14.6	1893	11.3	8941	14.0
2008	4536	16.1	7380	14.3	2725	16.2	9590	15.0
2009	4444	15.7	7656	14.8	2651	15.8	9575	15.0
2010	4600	16.3	5371	10.4	2879	17.1	7057	11.0
2011	4411	15.6	4463	8.6	2782	16.5	6050	9.4
University hospital district
Uusimaa	5161	18.3	9167	17.8	3376	20.0	11 442	17.9
Southwest	5572	19.7	8965	17.4	3679	21.9	11 850	18.5
Pirkanmaa	9166	32.5	18 523	35.9	5320	31.6	23 462	36.6
Northern Savo	3942	14.0	7059	13.7	2065	12.3	8247	12.9
Northern Ostrobothnia	4390	15.6	7926	15.4	2388	14.2	9062	14.2
Screening region
Urban	15 472	54.8	28 214	54.6	10 262	61.0	37 125	58.0
Semiurban	5373	19.0	10 193	19.7	3048	18.1	12 139	19.0
Rural	7386	26.2	13 233	25.6	3518	20.9	14 799	23.1
Age at invitation
60	19 880	70.4	32 058	62.1	11 582	68.8	40 172	62.7
62	4377	15.5	9918	19.2	2587	15.4	12 154	19.0
≥64	3974	14.1	9664	18.7	2659	15.8	11 737	18.3
Marital status
Married	15 419	54.6	37 893	73.4	8887	52.8	41 651	65.0
Never married	5676	20.1	5134	9.9	2185	13.0	4699	7.3
Divorced	6313	22.4	7402	14.3	4092	24.3	11 906	18.6
Widowed	781	2.8	1200	2.3	1629	9.7	5781	9.0
Unknown	42	0.2	11	0.0	35	0.2	26	0.0

Table 2. The risk for non-participation to colorectal cancer (CRC) screening of the first time invited persons (n = 160 762) according to screening background factors, by gender, 2004–2011.

	Men (n = 79 871)		Women (n = 80 891)
	IRR	95% CI	IRR	95% CI
Year of CRC invitation
2004	1.00	Ref.	1.00	Ref.
2005	1.08	0.98–1.19	1.06	0.94–1.20
2006	1.29	1.17–1.42	1.23	1.09–1.40
2007	1.27	1.16–1.41	1.33	1.17–1.50
2008	1.59	1.44–1.75	1.65	1.46–1.86
2009	1.51	1.38–1.67	1.60	1.42–1.81
2010	1.89	1.71–2.08	2.15	1.91–2.43
2011	2.00	1.82–2.21	2.32	2.06–2.62
University hospital district
Uusimaa	1.00	Ref.	1.00	Ref.
Southwest	1.03	1.00–1.07	1.00	0.95–1.05
Pirkanmaa	0.94	0.91–0.97	0.86	0.82–0.90
Northern Savo	0.97	0.93–1.01	0.93	0.88–0.99
Northern Ostrobothnia	0.98	0.94–1.02	0.97	0.92–1.02
Screening region
Urban	1.00	Ref.	1.00	Ref.
Semiurban	1.01	0.97–1.04	0.99	0.95–1.04
Rural	1.03	1.00–1.06	0.98	0.94–1.02
Age at invitation
60	1.00	Ref.	1.00	Ref.
62	0.93	0.90–0.96	0.95	0.91–0.99
≥64	0.90	0.86–0.92	0.99	0.95–1.03
Marital status
Married	1.00	Ref.	1.00	Ref.
Never married	1.74	1.69–1.80	1.75	1.67–1.84
Divorced	1.56	1.51–1.60	1.41	1.36–1.47
Widowed	1.43	1.33–1.53	1.31	1.24–1.38
Unknown	2.31	1.70–3.12	2.55	1.83–3.56

CRC: colorectal cancer; IRR: incidence rate ratio; 95% CI: 95% confidence interval; ref: reference.

Frequencies of non-participation in CRC screening by health behavior determinants are shown in Table 3. Among HBS participants, smoking was associated with non-participation in CRC screening, as current smokers had a higher risk for non-participation than non-smokers (men IRR 1.32, 95% CI 1.05 − 1.67, women IRR 2.10, 1.61 − 2.73) (Table 4). Inclusion of a weighting function on lag time did not affect the results (results not shown). Subgroup analyses on those invited for CRC screening showed that in relation to the group not invited for the HBS, the risk for refusing CRC screening was clearly increased among HBS non-participants (men IRR 1.40, 95% CI 1.26–1.56, women IRR 1.75, 95% CI 1.52–2.02), and slightly decreased among those who participated HBS (men IRR 0.83, 95% CI 0.77–0.91, women IRR 0.86, 95% CI 0.78–0.96) (results not shown in tables).

Table 3. Health behavior factors of persons who were invited to colorectal cancer screening (2004–2011) and who also participated in Health Behavior Survey (1978–1999) by gender and colorectal screening participation.

	Men			Women
	Invited	Non-participants		Invited	Non-participants
		(29.3%)			(17.5%)
	n	n	%	n	n	%
Total	1885	553	100	2076	364	100
Alcohol consume during last week
No	366	108	29.5	717	128	17.9
Yes	1004	288	28.7	904	148	16.4
Unknown	515	157	30.5	455	88	19.3
Tobacco smoking
Never/former	605	158	26.1	1243	181	14.6
Current	485	274	56.5	383	104	27.2
Unknown	795	221	27.8	450	79	17.0
Body mass index
<18.5	7	3	42.9	33	5	15.2
18.5–25.0	873	272	31.1	1275	220	17.3
25.1–29.9	811	232	28.6	545	91	16.7
≥30	179	44	24.6	185	39	21.1
Unknown	15	2	13.3	38	9	23.7
Vegetable intake
No/<3 days a week	713	246	34.5	526	105	20.0
3–7 days a week	1028	266	25.9	1392	226	16.2
Unknown	144	41	28.5	158	33	20.9
Berry/fruit intake
No/<3 days a week	764	261	34.2	433	84	19.4
3–7 days a week	882	221	25.1	1383	234	16.9
Unknown	239	71	29.7	260	46	17.7
Fat use in cooking
Butter	549	169	30.8	514	95	18.5
No spread/low fat	1141	327	28.7	1328	232	17.5
Unknown	195	57	29.2	234	37	15.8
Fat on bread
Butter	668	209	31.3	603	114	18.9
No spread/low fat	1203	341	28.3	1460	246	16.8
Unknown	14	3	21.4	13	4	30.8
Physical activity, leisure
Inactive	314	114	36.3	265	56	21.1
<2 times a week	525	142	27.0	556	85	15.3
≥2 times a week	594	176	29.6	674	132	19.6
Unknown	452	121	26.8	581	191	32.9
Physical activity, commuting
By car	887	249	28.1	486	64	13.2
<30 min a day	477	135	28.3	754	133	17.6
≥30 min a day	138	44	31.9	293	58	19.8
Not working	346	155	44.8	508	100	19.7
Unknown	37	10	27.0	35	9	25.7

Table 4. The risk for non-participation to colorectal cancer (CRC) screening (2001–2011) according to health behavior factors (1978–1999), by gender.

	Men (n = 1885)		Women (n = 2076)
	IRR	95% CI	IRR	95% CI
Alcohol consume during last week
No	1.00	Ref.	1.00	Ref.
Yes	1.00	0.79 − 1.26	0.88	0.69 − 1.13
Unknown	1.12	0.84 − 1.48	1.11	0.80 − 1.56
Tobacco smoking
Never/former	1.00	Ref.	1.00	Ref.
Current	1.32	1.05 − 1.67	2.10	1.61 − 2.73
Unknown	1.04	0.85–1.28	1.25	0.95 − 1.63
Body mass index
<18.5	1.35	0.43 − 4.24	0.74	0.30 − 1.81
18.5–25.0	1.00	Ref.	1.00	Ref.
25.1–29.9	0.95	0.79 − 1.13	0.98	0.76 − 1.26
≥30	0.78	0.57 − 1.09	1.22	0.86 − 1.73
Unknown	0.43	0.11 − 1.76	1.42	0.72 − 2.82
Vegetable intake
No/<3 days a week	1.00	Ref.	1.00	Ref.
3–7 days a week	0.87	0.72 − 1.05	0.86	0.67 − 1.11
Unknown	0.79	0.47 − 1.33	1.29	0.68 − 2.43
Berry/fruit intake
No/<3 days a week	1.00	Ref.	1.00	Ref.
3–7 days a week	0.84	0.69 − 1.02	1.02	0.78 − 1.34
Unknown	1.02	0.66 − 1.57	0.75	0.42 − 1.32
Fat use in cooking
Butter	1.00	Ref.	1.00	Ref.
No/low fat	0.99	0.79 − 1.25	0.95	0.72 − 1.25
Unknown	0.87	0.59 − 1.31	0.93	0.58 − 1.48
Fat on bread
Butter	1.00	Ref.	1.00	Ref.
No/low fat	1.01	0.81 − 1.27	0.97	0.74 − 1.29
Unknown	0.77	0.24 − 2.43	1.60	0.57 − 4.53
Physical activity, leisure
Inactive	1.00	Ref.	1.00	Ref.
Once a week/2–3 times a month	0.81	0.63 − 1.04	0.76	0.54 − 1.07
>2 times a week	0.92	0.72 − 1.17	1.00	0.73 − 1.38
Unknown	0.86	0.65 − 1.14	0.80	0.56 − 1.16
Physical activity, commuting
By car	1.00	Ref.	1.00	Ref.
≤30 minutes a day	0.96	0.77 − 1.19	1.26	0.93 − 1.71
>30 minutes a day	1.13	0.82 − 1.56	1.50	1.05 − 2.15
Not working	1.06	0.83–1.34	1.41	1.02 − 1.95
Unknown	1.10	0.57 − 2.09	1.76	0.86 − 3.62

CRC: colorectal cancer; IRR: incidence rate ratio; 95% CI: 95% confidence interval; ref: reference.

Discussion

This study showed that the male gender, never having married and young age at screening invitation was related to an increased risk for non-participation to CRC screening.

As has been reported before [15], older persons are more likely to participate in CRC screening than younger, and men are more likely to refuse CRC screening than women. Our results are thus in line with most previous studies on participation in CRC screening [3].

HBS was used for information on health behavior determinants in the study. Information on these factors were available for less than 3% of CRC screening invited persons leaving only 3961 individuals in the study material for analyses of health behavior factors. We found that persons who did not participate in the HBS were also less likely to participate later in CRC screening. As our data on lifestyle factors only cover HBS participants who are probably more health conscious than the entire target population our results cannot be fully generalized to the entire target population.

Of the health behavior factors available, only smoking increased the risk for non-participation in CRC screening significantly in our study. Current smokers were less likely to participate than non-smokers in both men and women.

Information on health behavior determinants were obtained from the postal national HBS based on self-reporting. People are likely to underestimate unhealthy health behavior, such as cigarette smoking [16], or alcohol abuse as self-estimated [17]. It has been shown that the main reasons for non-response in self-reported surveys are probably the predisposing sociodemographic and behavioral factors [18,19]: non-respondents are more likely to represent the low socioeconomic status and have more unhealthy behaviors than respondents.

The time between the HBS and the CRC screening invitation was more than 10 years for 84% of the study population, and health behavior-related factors may have changed in these individuals during this period. However, our results did not change when lag time weighting was added in the analyses.

Our data on CRC screening is based on the national registration of all invitations and attendance is reported from the screening center electronically. Thus, recall bias is not an issue in this data on screening uptake. In some of the previous studies on health behavior determinants and CRC screening participation, information on screening have been based on self-reported telephone [12,20], or postal surveys [21]. Similarly, our data on age, marital status, and gender are register based and covering on the national level, including no potential for bias. However, the data on health behavior is available only from persons willing to answer survey questions in 1978–1999.

Conclusions

Participation in CRC screening is affected by age and marital status. Health conscious persons are more likely to participate in CRC screening and health behavior surveys than others, and smoking seems to be a risk factor for non-participation in CRC screening among these persons.

Disclosure statement

The authors report no conflicts of interest. The authors alone are responsible for the content and writing of the paper.