4 Análisis cuantitativo y cualitativo de los instrumentos aplicados
4.1 La observación directa
4.1.1 Análisis de la observación directa
The first study in this thesis (Chapter 4) used the think-aloud method to establish how people interpret the CRC screening offer when they read the existing information booklet, ‘Bo el Cancer Screening: The Facts’. Eighteen participants were recruited from a number of sources. Despite attempts to recruit people who lacked health literacy skills, the resulting sample was highly educated and they disproportionately contributed to the findings.
Participants made on average 15 reading mistakes during the task, with terminology such as colonoscopy, colorectal and adenoma being particularly troublesome. The value of detailed biological processes such as the function of the colon and the adenocarcinoma sequence were questioned by participants. In addition, the range of numerical information throughout the booklet led to confusion and calls for it to be simplified. Participants commented that the booklet should be shorter for fears that a lengthy complex document may inhibit individuals from processing the most relevant text.
The think-aloud study provided detailed commentary on areas of the booklet that were considered unnecessary, confusing and poorly designed. However, because of the underrepresentation of low health literacy groups, the study did not provide relevant data to the extent that I thought it might at the start of the study. Although I was able to consult other relevant literature and speak to Specialist Screening Practitioners (SSPs) who work in the programme, by using data from Study 1 I may have failed to adequately address comprehension barriers that are experienced by those with the lowest levels of health literacy.
Nonetheless, using the resources available to me, I developed a gist-based information leaflet that was guided by the FTT model. This was an iterative process that resulted in many changes that were informed by best practice guidelines for designing simplified health information. One issue that I faced during this process was the scarcity of research on how FTT should be conceptualised when designing CRC screening information. FTT has roots in child eye-witness testimony and basic cognitive psychology, and has only recently been applied to the field of medicine. Although the decisions I made during the design phase were informed by FTT research, I was forced to consider other factors such as the issue of informed decision-making and the requirements of the NHS BCSP. The gist leaflet may therefore not have been a true representation of FTT, but instead should be considered to be my best attempt to accommodate these often competing influences.
Study 2 used a performance-based approach to evaluate the gist leaflet’s readability and comprehensibility. The method is based on principles of engineering (i.e. designing, testing and modifying prototypes), and has been applied to the evaluation of medication labels. In rounds of approximately 8-10 participants, the volunteers read the gist leaflet and answered a series of simple true or false statements about CRC and CRC screening. Using a pre-defined threshold (80% of participants had to answer each item correctly), the leaflet went through three rounds of testing before it was deemed comprehensible to the public and fit-for-purpose. Changes were made to the content, design and layout of the information in response to incorrect statements and qualitative data that were also collected. These changes were also informed by expert groups, best practice guidelines and FTT.
As with the think-aloud study, study 2 did not have a sufficient number of low literacy participants. Although this was improved in the later rounds of testing, the overall sample did not reflect the purposive nature of the recruitment methods used. The user-testing method was suited to people with low levels of basic skills, although the true/false responses may have led to artificially high levels of comprehension. The semi-structured interview that was performed after the user-testing was also more suited to educated participants. This is because they were more able and perhaps more willing to articulate issues they had with the gist leaflet, as well as offer solutions. Using mixed-methods was important as it allowed the perspective of both high and low literacy groups to be ascertained. However, participants with high health literacy are likely to have disproportionately influenced the study findings.
To investigate the communicative effectiveness of the gist leaflet and provide a more thorough test of the comprehensibility stage, study 3 was developed. This used a multicentre parallel randomised controlled trial recruiting from deprived General Practices in the north of England. In line with the framework, screening intention was the primary outcome and perceived readability and usefulness of the information, gist knowledge, perceived risk of CRC and worry about CRC were secondary outcomes. The extent to which the intervention addressed communication inequalities was investigated by monitoring the effect of the intervention for low and high numeracy groups.
The study groups were composed of an intervention group who were given the gist leaflet plus ‘The Facts’ booklet, and a control group ho ere given ‘The Facts’ booklet only. This study was therefore an important part of the evaluation as it was the first to investigate whether there was any added benefit of providing the gist leaflet to the established programme booklet. The decision to use the gist leaflet as a supplement was made early on in the development of the gist leaflet to accommodate concerns from the NHS BCSP committee that informed decision- making would be harmed as a result of providing gist-based information.
Although this decision was in keeping with the policy context of the screening programme, it led to a significant difference in the likelihood of participants reading the information materials they were allocated to. When offered the choice, participants ere more likely to choose to read the gist leaflet than ‘The Facts’ booklet. This preference was particularly apparent among the low numeracy group.
Although these differences were small, it might be safe to assume that they may be more pronounced in a sample who are not invested in completing a questionnaire. Further consideration of hether the gist leaflet should supplement ‘The Facts’ booklet, or be delivered as a standalone leaflet is clearly necessary.
The readability and usefulness scores provided by participants indicated that both information materials were considered acceptable. However, in support of the superior Flesch-Kincaid readability scores, participants in the intervention group were more likely to report that the gist leaflet was readable. These effects were stronger for low numeracy groups providing evidence that it was more accessible for people with poor basic skills. Although the leaflets were considered equally useful, there was a tendency for low numeracy respondents to prefer the gist leaflet.
The intervention group were also more likely to have adequate knowledge. There were however no significant differences between the groups in intention or perceived risk. The gist leaflet therefore did not affect the socio-cognitive antecedents of behaviour in the way that was hypothesised at the beginning of the study. Collectively, the gist leaflet may marginally improve comprehension of the screening offer, but it is unlikely to affect inequalities in uptake because there was no effect on intention. It may also inadvertently decrease the likelihood that people will read the information they are provided with, thereby affecting informed decision- making.
The results of study 3 provided weak support for further evaluation without amendments to the gist leaflet, or consideration of the context in which it was delivered (i.e. as a supplement). However, as this project was part of a larger NIHR programme grant, there was pressure to deliver the planned programme of work. Also, the hubs and Real Digital International had undertaken a good deal of preparation for the national trial. Alterations to these plans would not have been possible without adding to the workload.
Despite these reservations, there was also some justification for continuing with the original plans. Study 3 was limited by a highly motivated and relatively educated sample who reported screening intention rather than objective uptake data. The decision was therefore made to continue with the original protocol (study 4); a national cluster randomised controlled trial recording screening uptake data. This
study design enabled me to ascertain the effect of the intervention in 99% of the eligible sample, and was therefore a more thorough test of the leaflet’s communicative effectiveness. As with study 3, the control group were given the standard information and the intervention group were given standard information plus the gist-based leaflet.
The primary aim of this study was to assess to extent to which the leaflet reduced the SES gradient in CRC screening uptake. Secondary outcomes included overall uptake and uptake among population sub-groups (e.g. gender, age and screening history). As the design process was largely undertaken with individuals who had not been invited to CRC screening, sub-sample analyses were performed among first time invitees.
Disappointingly, data from the whole sample showed there were no significant differences between the intervention and control group on any of the outcome measures. In sub-sample analyses that only included individuals that were being invited for the first time, there was also no significant effect on the SES gradient. In the sub-sample there was a small significant difference in screening uptake between the study groups. However, the extent to which this was clinically meaningful is unclear. The intervention was also more effective among men and older individuals invited for the first time. Overall, the gist leaflet did not meet the requirements for all three stages of the Garner framework because the message did not result in behavioural differences among the predicted groups.