Use of research questionnaires in the NHS Bowel Cancer Screening Programme in England: impact on screening uptake

Abstract

Background

The NHS Bowel Cancer Screening Programme in England offers biennial screening to those aged 60-74 using a faecal occult blood test (FOBt) sent by post. Data from this national clinical programme can also be used for research. Awareness of the impact of such studies on screening participation is important.

Aims

To investigate the effect on screening uptake of adding a research questionnaire to the postal screening invitation.

Methods

People invited for screening in 2008–10 in two areas of England were randomized to receive or not to receive an additional research study questionnaire, consent form and study information, either with their test kit or 2-3 days later. Uptake of screening was examined in relation to study mailings.

Results

Among 11,579 people invited for screening by the Midlands and North West Bowel Cancer Screening Hub, screening uptake was significantly lower in those who received study documents with their FOBt kit than in those who did not (48.6% vs 53.5% respectively: p < 0.001). The reduction in uptake was similar in men and women, and was greater in people living in more deprived areas. Among a further 36,195 people invited for screening by the Midlands and North West and Southern Hubs, sending study documents by separate mailing 2–3 days after the FOBt kit did not affect screening uptake (uptake with and without additional study mailing: Midlands and North West, 56.7% and 56.2% respectively, p = 0.6; Southern, 52.0% and 51.4% respectively, p = 0.5).

Conclusions

Researchers planning studies that include contact with potential participants within the NHS Bowel Cancer and similar screening programmes should be aware of the potential impact on uptake.

Introduction

The NHS Bowel Cancer Screening Programme uses biennial guaiac faecal occult blood testing (FOBt) to screen both men and women for colorectal cancer. The programme in England began inviting people aged 60–69 to take part in screening in July 2006, and achieved national roll-out in January 2010. Extension of the age range to include men and women up to the age of 75 began in September 2008 and is expected to be completed nationally during 2013. The programme currently sends some 3.8 million invitations a year, and over two million participants complete the FOBt.

The screening programme is an attractive setting for research on bowel cancer screening and epidemiology, as the programme is unique in having a single national database that holds comprehensive call/recall, laboratory and clinical investigation records. A large number of projects are planned or in progress, and it is important to ensure that this research is conducted in a way that is not detrimental to screening uptake. There is limited evidence on the impact of research studies on uptake from other UK screening programmes, and past experience may not be relevant to the Bowel Cancer Screening Programme, which is the first screening programme in the UK to include men, and the first to use a test kit sent by post, completed at home and returned without the direct involvement of a primary healthcare provider.

As part of a planned research study, we investigated the impact on uptake of FOBt of sending a research questionnaire and related study documents, either with the test kit or separately 2–3 days later.

Methods

The NHS Bowel Cancer Screening Programme in England

Five regional Hubs are responsible for providing all services relating to the FOBt (invitations, kit mailing and return, kit processing, sending FOBt results to participants and arranging a clinical appointment for those with a positive test result). Follow-up services for participants with abnormal FOBt results are provided by clinical staff at 59 local screening centres. The initial contact between the Hub and potential participants is an invitation letter and information booklet. After receiving the invitation letter, potential participants can contact the Hub to opt-out of screening; on average fewer than 1% do so. About two weeks after sending the invitation packs, the Hubs send test kits to potential participants, with written instructions describing how to complete the kit. Participants then mail the completed kits back to the Hub for processing. Test results (and, where appropriate, information on subsequent repeat screening or diagnostic tests) are returned to participants by letter.

Study participants

The study was conducted in two of the five regional Hubs - Midlands & North West and Southern - over several months in 2008, 2009 and 2010. All men and women sent an invitation to take part in the NHS Bowel Cancer Screening Programme during the study period and who did not contact the programme to opt out of receiving an FOBt kit were eligible for inclusion (this included people invited for a first or a subsequent screen).

Study documents

The study pack included a four-page study questionnaire with demographic, medical and lifestyle questions, and was designed (and described to potential participants as being) for use in a study investigating the characteristics of people with a false-positive FOBt result. In addition to the request for questionnaire responses, potential participants were asked to give permission for follow-up through medical records. Each questionnaire was accompanied by a study information sheet and two consent forms (one for the participant to keep and one to be filled in and returned with the questionnaire), and a postage-paid return envelope. Study documents explained that the research was being carried out by the Director of the NHS Cancer Screening Programmes, with researchers from the University of Oxford, and was funded by Cancer Research UK. The questionnaires that were sent 2–3 days after the kit were sent with an additional covering letter about the study from the Hub director.

The study information sheet and consent forms were produced in accordance with guidance from the National Research Ethics Service [http://www.nres.nhs.uk/applications/guidance/consent-guidance-and-forms/].. Copies of the questionnaires and study documents are available on the study website (www.ceu.ox.ac.uk/bsfs).

Randomization and mailing

Part 1: Research documents included in screening kit mailing

Between May and August 2008, 11,579 participants invited for screening by the Midlands & North West Bowel Cancer Screening Hub were selected to receive (5,857 people) or not to receive (5,722 people) a study questionnaire pack (questionnaire, patient information sheet, consent forms and reply-paid envelope) included in the same envelope as their screening test kit. For logistical reasons, individual randomization was not feasible and so participants were pseudo-randomized, with every 25^th FOBt kit recipient chosen to receive the additional study pack, and every 26^th recipient chosen as a control. Invitations for screening within each region are automatically generated from NHS records by the Bowel Cancer Screening System based on date of birth, and sorted for mailing by NHS number, a unique identifier which is not based on any of the personal characteristics of the person to whom it is allocated. The combination of the two 'pseudo-random' elements of invitation generation, by birth date and NHS number, was thought to be sufficient to produce pseudo-randomization of the questionnaires by allocating them evenly through the potential study population.

Part 2: Research documents mailed separately 2–3 days after kit mailing

A total of 36,195 participants invited for screening by the Midlands and North West and Southern Bowel Cancer Screening Hubs were selected to receive or not to receive the study pack, mailed separately from the Hub 2–3 days after their screening kit mailing. To minimize the workload for Hub staff, pseudo-randomization was achieved by allocating recipients in mailing batches, the mailing lists being pseudo-randomly generated as described above. In September/October 2009, at the Midlands and North West screening hub, two daily batches of kit recipients, in all 6170 people, were later sent the additional study mailing, and four batches (13,163 people) were selected as controls. In May 2010, at the Southern screening hub, one day’s mailing batch of 5,784 people were sent the additional study mailing, and two batches mailed on the same day of the week, a week earlier and a week later, were chosen as controls (11,078 people).

Data sources and Statistical methods

Screening uptake was defined for this study as receipt of a completed FOBt kit by the Hub within 12 weeks of the initial invitation to take part (ie.10 weeks after the kit was mailed). Uptake was compared between people sent and not sent a questionnaire and study information, using a two sample test for proportions (z-test) with a 5% significance level. Uptake was also compared by age, sex, and level of deprivation. Screening uptake, age, and sex were ascertained from screening programme records (the Bowel Cancer Screening System [BCSS] database). Participants’ postcodes were used to assign a measure of area-level socioeconomic deprivation (Index of Multiple Deprivation 2007 ¹ ) based on quintiles of the distribution in the national population.

Ethics and governance: The study has ethical approval from Cambridge 1 Research Ethics committee, study reference 07/Q0104/4; and is approved by the NHS Bowel Cancer Screening Programme Research Committee [http://www.cancerscreening.nhs.uk/bowel/research-committee/]. The University of Oxford acts as research Sponsor and the study is listed on the NIHR Clinical Research Network Portfolio : Study ID 10711.

Results

The characteristics of people invited for screening, and randomized to receive or not to receive additional research study documents, are shown in Table 1. For each of the comparison groups, in both Hubs, the research study document recipients and their controls had very similar distributions by age, sex, and deprivation quintile, as would be expected with adequate pseudo-randomization. Differences in age distribution between the hubs, and between study parts for the Midlands and North West hub, reflect differences in the stage of the 2-year bowel cancer screening invitation cycle. There are clear differences between the two hubs in the proportions of the invited population in high and low deprivation categories, with more people in higher socioeconomic groups in the Southern hub area.

Table 1.

Characteristics of study participants, including those sent and not sent the research study questionnaire pack.

Study part 1: Midlands & N West Hub				Study part 2: Midlands & N West Hub				Study part 2: Southern Hub
No study Q		Study Q sent with screening kit		No study Q		Study Q sent after screening kit		No study Q		Study Q sent after screening kit
n	%	n	%	n	%	n	%	n	%	n	%
Total	5,722	5,857	13,163	6,170	11,078	5,784
Sex
Men	2,855	49.9	2,916	49.9	6,513	49.5	3,063	49.6	5,630	50.8	2,867	49.6
Women	2,866	50.1	2,934	50.2	6,650	50.5	3,107	50.4	5,448	49.2	2,917	50.4
Missing	1	<0.1	7	<0.1	0	0	0	0
Age, years
<60	16	0.3	16	0.3	211	1.6	121	2.0	0	0.0	0	0.0
60–64	3,106	54.3	3,201	54.7	6,676	50.7	2,975	48.2	6,185	55.8	3,181	55.0
65–69	2,443	42.7	2,490	42.5	5,431	41.3	2,489	40.3	4,805	43.4	2,574	44.5
70+	18	0.3	29	0.5	843	6.4	584	9.5	88	0.8	29	0.5
missing	139	2.4	121	2.1	2	<0.1	1	<0.1	0	0
Deprivation Quintile
Most deprived 1	1,427	24.9	1,490	25.4	2,836	21.5	1,336	21.7	596	5.4	331	5.7
2	992	17.3	1,065	18.2	2,322	17.6	1,074	17.4	1,495	13.5	708	12.2
3	1,127	19.7	1,117	19.1	2,668	20.3	1,259	20.4	2,406	21.7	1,238	21.4
4	1,124	19.6	1,170	20.0	2,911	22.1	1,346	21.8	2,917	26.3	1,535	26.5
Least deprived 5	1,018	17.8	949	16.2	2,333	17.7	1,113	18.0	3,661	33.0	1,970	34.1
missing	34	0.6	66	1.1	93	0.7	42	0.7	3	<0.1	2	<0.1

Q = study questionnaire pack.

In the first Midlands and North West groups (Table 2: part 1 of the study), overall uptake was significantly lower in people who had received the research study documents with their FOBt kit (48.6% uptake) than in those sent only the standard screening FOBt kit mailing (53.5% uptake; p for difference, <0.001). As shown in Table 2, with the standard (no study documents) mailing, uptake of the FOBt screening test was greater in women (57.5%) than in men (49.6%) and was strongly related to socioeconomic status, with uptake rising from 40.7% in people in the most deprived quintile to 62.7% in those in the least deprived quintile. Inclusion of the research study documents with the kit mailing was associated with significantly reduced uptake in both men and women, and for those aged 60–64 and aged 65–69. While uptake was lower than for the standard mailing in all socioeconomic groups, the reduction was greater with higher levels of deprivation (32.6% uptake with study documents vs 40.7% without in the most deprived group; 60.8% vs 62.7% in the least deprived group).

Table 2.

Study part 1: screening test uptake* for Midlands & N West hub, in those sent and not sent research study questionnaire pack with screening kit.

No Q sent with screening kit				Q sent with screening kit
Kits Sent	Returned	Not Returned	% Uptake	Kits Sent	Returned	Not Returned	% Uptake	Difference	P-Diff
Total	5,722	3,062	2,660	53.5%	5,857	2,846	3,011	48.6%	−4.9%	0.000
Men	2,855	1,415	1,440	49.6%	2,916	1,363	1,553	46.7%	−2.8%	0.032
Women	2,866	1,647	1,219	57.5%	2,934	1,483	1,451	50.5%	−6.9%	0.000
Age 60–64	3,106	1,700	1,406	54.7%	3,201	1,557	1,644	48.6%	−6.1%	0.000
Age 65–69	2,443	1,324	1,119	54.2%	2,490	1,259	1,231	50.6%	−3.6%	0.011
Deprivation Quintile
Most deprived 1	1,427	581	846	40.7%	1,490	486	1,004	32.6%	−8.1%	0.000
2	992	502	490	50.6%	1,065	479	586	45.0%	−5.6%	0.011
3	1,127	649	478	57.6%	1,117	601	516	53.8%	−3.8%	0.071
4	1,124	675	449	60.1%	1,170	668	502	57.1%	−3.0%	0.150
Least deprived 5	1,018	638	380	62.7%	949	577	372	60.8%	−1.9%	0.394
missing	34	17	17	50.0%	66	35	31	53.0%	3.0%	0.774

*Screening test uptake measured as number/percentage of faecal occult blood test kits returned completed by 12 weeks after invitation sent.

Q = research study questionnaire pack.

P-diff = p value for difference between % uptake in those not sent and sent Q.

For age analyses, results for participants younger or older than the screening target age group of 60–69 years [see Table 1] are not shown, so numbers do not sum to total.

By contrast, screening uptake was not affected by separate mailing of the study documents 2-3 days after the standard FOBt kit mailing (Table 3). Overall uptake in the study document group and in the standard mailing group was 56.7% and 56.2% respectively for the Midlands and North West hub (p for difference, 0.6), and 52.0% and 51.4% respectively (p for difference 0.5) for the Southern hub. Results were similar for both areas for both men and women, by age, and, overall, by socioeconomic group.

Table 3.

Study part 2-screening test uptake* for Midlands and N West and for Southern hubs, in those sent and not sent a research study questionnaire pack shortly after their screening kit.

No Q sent after screening kit				Q sent after screening kit
Kits Sent	Returned	Not Returned	% Uptake	Kits Sent	Returned	Not Returned	% Uptake	Difference	P-Diff
Midlands & N West Hub
Total	13,163	7,400	5,763	56.2%	6,170	3,496	2,674	56.7%	0.4%	0.563
Men	6,513	3,491	3,022	53.6%	3,063	1,667	1,396	54.4%	0.8%	0.451
Women	6,650	3,909	2,741	58.8%	3,107	1,829	1,278	58.9%	0.1%	0.937
Age 60–64	6,676	3,726	2,950	55.8%	2,975	1,635	1,340	55.0%	−0.9%	0.436
Age 65–69	5,431	3,165	2,266	58.3%	2,489	1,407	1,082	56.5%	−1.7%	0.144
Deprivation Quintile
Most deprived 1	2,836	1184	1652	41.7%	1,336	626	710	46.9%	5.1%	0.002
2	2,322	1215	1107	52.3%	1,074	590	484	54.9%	2.6%	0.157
3	2,668	1,568	1,100	58.8%	1,259	706	553	56.1%	−2.7%	0.111
4	2,911	1,816	1,095	62.4%	1,346	837	509	62.2%	−0.2%	0.900
Least deprived 5	2,333	1,551	782	66.5%	1,113	706	407	63.4%	−3.0%	0.078
missing	93	66	27	71.0%	42	31	11	73.8%	2.8%	0.734
Southern Hub
Total	11,078	5,693	5,385	51.4%	5,784	3,005	2,779	52.0%	0.6%	0.487
Men	5,630	2,674	2,956	47.5%	2,867	1,375	1,492	48.0%	0.5%	0.686
Women	5,448	3,019	2,429	55.4%	2,917	1,630	1,287	55.9%	0.5%	0.684
Age 60–64	6,185	3,172	3,013	51.3%	3,181	1,655	1,526	52.0%	0.7%	0.496
Age 65–69	4,805	2,461	2,344	51.2%	2,574	1,336	1,238	51.9%	0.7%	0.574
Deprivation Quintile
Most deprived 1	596	211	385	35.4%	331	127	204	38.4%	3.0%	0.369
2	1,495	626	869	41.9%	708	327	381	46.2%	4.3%	0.056
3	2,406	1,198	1,208	49.8%	1,238	622	616	50.2%	0.5%	0.797
4	2,917	1,586	1,331	54.4%	1,535	811	724	52.8%	−1.5%	0.328
Least deprived 5	3,661	2,071	1,590	56.6%	1,970	1,117	853	56.7%	0.1%	0.925
missing	3	1	2	33.3%	2	1	1	50.0%	16.7%	0.709

*Screening test uptake measured as number/percentage of faecal occult blood test kits returned completed by 12 weeks after invitation sent.

Q = research study questionnaire pack.

P-diff = p value for difference between % uptake in those not sent and sent Q.

Results by age for participants younger or older than the screening target age group of 60-69 years [see Table 1] are not shown.

Discussion

We found that including a research questionnaire and study information with a mailed FOBt kit was associated with significantly reduced uptake of screening for bowel cancer in an organized screening programme in England. Sending the same study questionnaire and information by separate mailing a few days after sending the kit did not affect uptake. These results underline the importance of conducting pilot studies for planned research associated with cancer screening programmes, and highlight apparent differences in the impact of research on programmes screening for different cancers. Our findings may also be more broadly applicable to programmes screening for other diseases.

Studies similar to ours, in the UK and in Europe, have found no evidence for an effect of adjunct research studies on uptake of mammography screening for breast cancer. In a study conducted in the NHS Breast Screening Programme, 6,400 women invited for breast screening by the Oxfordshire and West of London breast screening units were randomized to receive or not receive a self-administered research questionnaire with their standard invitation to screening. There was no difference in screening attendance between the two groups. ² Two further studies investigating the effect on breast screening attendance of sending mailed invitations to take part in adjunct research projects around the time of breast screening invitation were conducted in Utrecht (the Netherlands) ³ and Tromsø (Norway). ⁴ Neither study found any effect of receiving adjunct research invitations on breast screening attendance. It was also found that conducting research interviews before inviting women to the NHS Breast Screening Programme in London (England) had no effect on subsequent screening uptake. ⁵

To our knowledge there has been one previous study of the effect on screening uptake of adjunct research in a bowel screening setting, conducted by O’Sullivan and colleagues within the setting of the pilot for the NHS Bowel Cancer Screening Programme in Coventry and North Warwickshire. ⁶ People sent an FOBt kit were selected to receive or not receive a self-administered questionnaire addressing their attitudes to participating in FOBt screening. The questionnaires were mailed two working days after the FOBt kits, and no significant difference in overall screening uptake between the two groups was seen. The results of the second part of our study are consistent with this finding.

We do not know why including a questionnaire within FOBt kit mailings reduced screening uptake, but sending questionnaires out separately 2–3 days later did not. The NHS Bowel Cancer Screening Programme, unlike those in the UK for other cancers, does not involve direct contact with a healthcare practitioner for the initial screening test. The FOBt kit mailings necessarily include information about the kit and how to perform it in written format. The questionnaire pack also contained large amounts of written information, and receiving both at the same time may have been overwhelming for some potential screening participants. This might be particularly likely for those with limited literacy skills and/or those already less likely to return a completed kit, and may explain the greater effect of questionnaire inclusion with the screening kit on uptake in more deprived socioeconomic areas. ⁷ There was no evidence for such an effect when the study documents were sent 2–3 days after the screening kit; indeed our results (Table 3) show a tendency for this kind of study mailing to be associated, if anything, with increased screening uptake in the more deprived socioeconomic groups.

Our study benefited from the availability of a well organized screening programme which routinely sends out large numbers of screening kits. The NHS Bowel Cancer Screening programme BCSS database allowed follow-up of patient behaviour from kit mailing to return, allowing accurate calculation of screening uptake with a very small percentage of missing information on age, sex and deprivation index. Individual randomization of potential participants to receive or not receive the study documents was not feasible. However screening invitation lists are organized both by birth date and by randomly-assigned NHS number (and, for those who have been screened previously, the date of completion of their previous screen), and there is no obvious reason why the pseudo-randomization used, based on these lists, should not result in an acceptably equal spread of patient characteristics across the questionnaire and control groups. Comparison of known patient characteristics (age, sex, quintile of deprivation index) between the questionnaire group and controls suggests that our randomization methods were adequate. The figures presented in this paper for screening uptake apply to the research setting, and should not be taken to represent accurately the operational characteristics of the NHS Bowel Cancer Screening Programme, either at the time of the study or subsequently. Programme definition of uptake is not identical to that used in this study.

Large-scale, well-organized screening (and other health) programmes with centralized electronic databases have huge potential as research resources. When research studies involve direct contact with programme participants it cannot be assumed that they will not have an impact on screening behaviour.

Declarations

The study was funded by Cancer Research UK. The funders did not influence study design, analysis, interpretation or the decision to publish. JW and JG had full access to study data and act as guarantors. The study was designed by JG, JP and VB, with analyses by JW. KS and MM contributed to database design and data entry, SS and SH to data collection; the manuscript was drafted by JW and JG with contributions from all authors.The authors have no competing interests to declare.