Exploratory Analysis on the Impact of Morbidities on Colorectal Cancer Screening Uptake
|The safety and scientific validity of this study is the responsibility of the study sponsor and investigators. Listing a study does not mean it has been evaluated by the U.S. Federal Government. Read our disclaimer for details.|
|ClinicalTrials.gov Identifier: NCT02503969|
Recruitment Status : Completed
First Posted : July 21, 2015
Last Update Posted : December 2, 2015
|First Submitted Date||July 16, 2015|
|First Posted Date||July 21, 2015|
|Last Update Posted Date||December 2, 2015|
|Study Start Date||September 2015|
|Actual Primary Completion Date||December 2015 (Final data collection date for primary outcome measure)|
|Current Primary Outcome Measures
||Individual attended colorectal cancer screening [ Time Frame: The duration for which the individual was eligible for screening (between 1 and 10 years) ]
If the individual attended any colorectal cancer screen (irrespective of the number of invitations) - analysed as a binary yes/no indicator per individual using logistic regression.
|Original Primary Outcome Measures||Same as current|
|Current Secondary Outcome Measures
||Individual accepted colorectal cancer screening after refusing the initial invitation for colorectal cancer screening [ Time Frame: The duration for which the individual was eligible for screening (between 1 and 10 years) ]
This may not be assessed, depending on the available evidence. If analysed, will be as a binary yes/no indicator per individual using logistic regression.
|Original Secondary Outcome Measures||Same as current|
|Current Other Pre-specified Outcome Measures||Not Provided|
|Original Other Pre-specified Outcome Measures||Not Provided|
|Brief Title||Exploratory Analysis on the Impact of Morbidities on Colorectal Cancer Screening Uptake|
|Official Title||Exploratory Analysis on the Impact of Morbidities on Colorectal Cancer Screening Uptake|
The primary aim of this project is to examine the association between having a long-term condition (morbidity) and screening uptake for colorectal cancer. Whilst this project will consider all morbidity and co-morbidities, there will be a particular focus on common mental health disorders, such as depression and anxiety.
The secondary aim of this project is to examine other factors that may influence uptake rate. Information on a wide array of potential factors is available for this project. These include demographics (age, gender, ethnicity), socio-economic status (deprivation, education status) and lifestyle (smoking status, drinking patterns, degree of exercise). In addition, any potential moderating effect of these factors on the association between morbidity and screening uptake shall be explored.
In summary, the following shall be explored:
Linkage of datasets and ethical issues.
The South Yorkshire Cohort (SYC) data (including patient identifiers) are held by the Clinical Trials Research Unit (CTRU) at the University of Sheffield, whilst Bowel Cancer Screening Programme (BCSP) data are held by the NHS Cancer Screening Programmes (part of Public Health England).
Linkage will be based on the subset of respondents to the SYC who have given their consent for the SYC researchers to look at their NHS health records, and are eligible to be invited for colorectal cancer (CRC) screening. The proposed method for linkage will be as follows:
In this way, the only patient-identifiable data that the BCSP would receive is data that they already hold (NHS number), whilst the CTRU would not receive any additional patient-identifiable data.
It should also be stressed that the research team undertaking this project would not have access to patient-identifiable data at any stage of the project.
The following data shall be used within this project:
• Exposure variables. The exposure variables are the presence of long-standing conditions. These are collected within the SYC as self-reported long-standing conditions. Twelve named conditions are recorded, along with an "other" condition, which includes free-text to allow the respondent to state the condition. These conditions (and their prevalence amongst a preliminary sample of people aged 60 to 74 in the SYC) are: Depression (8%), Anxiety (10%), Fatigue (19%), Pain (28%), Insomnia (8%), Diabetes (10%), Breathing problems (13%), High blood pressure (31%), Heart disease (10%), Osteoarthritis (16%), Stroke (3%), Cancer (5%) and Other (29%).
All of the long-standing conditions shall be considered, with the exception of cancer (as this will include CRC, and patients with this are not eligible for screening) and free-text descriptions (these shall be analysed as "other").
The statistical analysis shall include the following sections:
• Initial exploratory analysis. This initial analysis shall provide an overview of the available data, and will highlight any issues that may need to be addressed within the statistical modelling.
A descriptive analysis shall compare the characteristics of people who attend CRC screening with those who do not attend. This comparison will include the exposure variables, confounding variables and additional descriptive variables (as detailed in the previous section). Comparisons will be tested for statistical significance, with the caveat that as no specific differences were hypothesised a priori, resulting p-values should be interpreted with caution. T-tests will be used to compare continuous variables, Fisher's exact test will be used to compare binary variables, and the Kruskal-Wallis test will be used to compare ordinal variables. Any p-values less than 5% will be taken to indicate a statistically significant association.
In addition to the descriptive analysis, the functional form of the association between any continuous variables and the outcome shall be visually assessed using smoothing methods. If a non-linear functional form is indicated then the use of non-linear functions (fractional polynomials, natural splines) shall be considered.
The primary interest is which morbidities affect uptake rate, with particular interest in mental morbidities (of which depression and anxiety are measured in the SYC). Because of this, only interactions with these two mental morbidities shall be considered. To examine the association between the mental morbidities and uptake, and to see how the other variables influence this association, a series of models shall be presented:
The purpose of displaying a series of models will be to show the un-adjusted association between morbidities and screening uptake, and the highlight the degree to which these associations are mediated by patient characteristics. A distinction is made between 'intrinsic' characteristics (age, gender and ethnicity), which are (generally) beyond a person's control to change, and the remaining characteristics, over-which a person has more control
Power analyses were conducted using G*Power 3.1.9. The required sample size to detect a significant effect of a pre-specified variable was calculated. For this analyses an alpha level of 5% and a two-tailed test were used. There were a number of additional factors that needed to be estimated or chosen:
Table 1: Sample size required as a function of power, correlation, and odds ratio.
Power = 80% Power = 95% Odds Ratio R2 = 0.2 R2 = 0.4 R2 = 0.2 R2 = 0.4 1.2 16,066 21,422 26,605 35,473 1.3 7,780 10,373 12,876 17,168 1.4 4,751 6,335 7,857 10,476 1.5 3,292 4,389 5,437 7,249 1.6 2,466 3,289 4,069 5,425 1.8 1,603 2,137 2,637 3,516 2.0 1,174 1,566 1,925 2,567 2.2 926 1,234 1,513 2,017 2.4 767 1,022 1,248 1,665 2.6 658 877 1,067 1,422 2.8 579 772 935 1,247 3.0 519 693 837 1,116
The sample size available is expected to be approximately 7,500 indicating that any odds ratios of 1.4 or greater will be detected with 80% power, whilst any odds ratios of 1.5 or greater will be detected with 95% power.
An example of a change in uptake rates relating to an odds ratio of 1.5 would be a decrease in uptake from 60% to 57%. Further examples are presented in Table 2.
Table 2: Decreases in uptake rate corresponding to an odds ratio of 1.5 Uptake rate in group 1: 90% 80% 70% 60% Uptake rate in group 2: 81% 72% 64% 57%
An alternative method for estimating samples sizes was also tested (Campbell, Julious, and Altman 1996). This uses published tables which estimate the sample size (per group) required to identify a pre-specified difference in two proportions at at a 5% significance level with 80% power. The group sizes are then adjusted to take into account any differences in group size. For example, if the group with depression represents 8% of the total population, then to detect a difference in uptake of 65% amongst people without depression and 60% amongst people with depression, an overall sample size of 425 is required, of which 34 would need to have depression and 391 would not have depression. The overall sample size required for a range of differences in uptake rates are presented in Table 3 (the uptake rates are all multiples of 5% as these values are used in the published tables).
Table 3: Sample sizes required to detect pre-specified differences in uptake. Uptake in group A Uptake in group B Total sample size 60% 55% 1 188 65% 60% 425 70% 65% 213 75% 70% 138
* Assuming that the comparison is between people with depression (prevalence 8%, corresponding to group B) and without depression.
|Study Design||Observational Model: Case Control
Time Perspective: Cross-Sectional
|Target Follow-Up Duration||Not Provided|
|Sampling Method||Non-Probability Sample|
|Study Population||This study shall be restricted to participants in the South Yorkshire Cohort who have ticked both boxes on the Consent Form for their data to be used (the boxes being labelled "May we look at your health records?" and "May we use the information you provide to look at the benefit of health treatments?"), and are eligible for screening in the NHS Bowel Cancer Screening Programme (ages 60 to 69).|
|Intervention||Other: Exposure to having a long-term condition.
Long-term conditions are self-reported: twelve named conditions are recorded, along with an "other" condition - these shall all be considered with the exception of the named condition "cancer".
* Includes publications given by the data provider as well as publications identified by ClinicalTrials.gov Identifier (NCT Number) in Medline.
|Original Estimated Enrollment
|Actual Study Completion Date||December 2015|
|Actual Primary Completion Date||December 2015 (Final data collection date for primary outcome measure)|
|Ages||60 Years to 69 Years (Adult, Older Adult)|
|Accepts Healthy Volunteers||Yes|
|Contacts||Contact information is only displayed when the study is recruiting subjects|
|Listed Location Countries||United Kingdom|
|Removed Location Countries|
|Other Study ID Numbers||15/YH/0028|
|Has Data Monitoring Committee||No|
|U.S. FDA-regulated Product||Not Provided|
|IPD Sharing Statement||Not Provided|
|Responsible Party||Ben Kearns, University of Sheffield|
|Study Sponsor||University of Sheffield|
|PRS Account||University of Sheffield|
|Verification Date||December 2015|