Randomized Controlled Trials - Essay Example

?Question Given the research question and the information given about the lung cancer diagnosis intervention, it seems sensible to suggest that a cluster randomized controlled trial (RCT) design would be useful. In the cluster method, a group is the unit of randomization (James et al, 2004). This style is necessary when an intervention can only be administered to people as part of a group, such as a public education promotion as in the scenario described. In this case, a National Awareness and Early Detection Initiative has been developed with the purpose of leading individuals to earlier diagnosis of lung cancer, and this would benefit from being delivered to the sample as it would be in reality. For example, an awareness campaign would be public, and found in normal, everyday situations for maximum exposure targeting those with an early cough, particularly smokers. The huge strength of using this type of RCT design in this scenario is that it makes sense to deliver the public awareness campaign in this way, because the information would usually be delivered to a group (James et al, 2004). It would be extremely difficult to measure the efficacy of a campaign using billboard advertisements without using the campaign on a group of people, rather than individuals. Showing an advertisement to an individual may give some idea about the potential success of the campaign, but it would not be how it was delivered after the trial and it would be impossible to judge how it was received in this capacity. Additionally, with the cluster design, there is always a debate about how best to analyse the data. It is possible to analyse it using the targeted groups or clusters, or to take all the data and analyse each member of the population as an individual (Hussey & Hughs, 2007). Analysing the data from the groups accounts for the fact that individuals within a group are usually more similar to each other than to members outside the group (James et al, 2004), but it is worth noting that groups can be of many different sizes and this method usually returns less data than measuring as individuals giving less statistical power (Prescott et al, 1999). Analysing the participants individually gives much more data, but there may be some bias in the results due to the fact that the respondents in one town or one workplace are generally more similar to each other than the general population. A step-wedge design may be useful in this scenario. In a step-wedge design, the intervention is rolled out to different individuals or different clusters over a course of time (Hussey & Hughs, 2007). In this case, any awareness of early detection of lung cancer will do more harm than good, so it would be useful and perhaps more ethical to utilize a trial design in which everyone can benefit. A step-wedge design can also help to provide information about the timescale of exposure to a public awareness campaign. This may be useful for deciding how long to present the campaign to the population if the RCT returns positive results and thus may present money-saving options. Another situation where a step-wedge design is commonly used is where it may be financially or logistically impossible to deliver the intervention to all individuals at once (Prescott et al, 1999). In this case, it may be difficult to find billboard space to present advertisements which are all available at the same time, so this may overcome some of the logistical problems of the intervention. However, due to this style of design the trial is likely to take longer than in a standard trial (Hussey & Hughs, 2007), which may mean that it costs more money and the data will not be available for some time after starting the trial. Another problem with this type of design is that it may be impossible to prevent ‘contamination’ between the individuals within different time-scales (Hussey & Hughs, 2007). In this case, if the advertisements were placed in one local area, travel means that it would be impossible to say whether some individuals had seen the advertisement before the time allotted in the randomization process. This can give the results a bias that may misinform the analysis of the results. A pragmatic RCT design may also be useful here. It was mentioned in the information that the campaign would be primarily aimed at smokers. In reality, it is difficult to distinguish between non-smokers and smokers so easily, as some may smoke on certain occasions and some may smoke more than others. Using a pragmatic design also allows the inclusion of non-smokers in the trial, which is useful as lung cancer does present in non-smokers (Peto et al, 2000). This type of design has a huge amount of flexibility and reflects the heterogeneity of patients in practise, therefore it can be said to be more realistic than other trials focussing merely on one intervention in one very specific case (Cardini et al, 2006). It should be noted that pragmatic RCTs are generally considered to measure effectiveness rather than efficacy (Laurence et al, 2008), meaning that they may not provide the desired result. In the case of the early detection intervention, this may still be relevant as all that is needed is for people to have an awareness about the early signs of lung cancer. Additionally, to be pragmatic means that it reflects reality. This type of campaign would, in reality, have no exclusion criteria because billboard advertisements would be visible to everyone in the local area and thus the intervention would be useful for a community rather than an individual. In this way, the trial can already be said to be pragmatic in design because there is no way of ensuring that exclusion criteria are strictly met in this type of trial. Additionally, it has been suggested that this type of trial is most useful for ‘interventions that inform policy decisions’ (Laurence et al, 2008, p50), which is relevant in the case of a public health awareness campaign such as this. Question 2 Aim The aim is to conduct a randomised controlled trial to assess how the pricing of school meal options influences the choice of secondary school students between healthy and unhealthy options. Research Question Does providing lower priced healthy school meals act as an incentive to secondary school students to purchase these meals over higher priced unhealthy options? Participant Recruitment Schools will be asked whether they wish to participate in the scheme. To recruit these schools, information will be circulated to schools throughout the country, providing information about the study as well as background information about the effects of unhealthy eating. To participate, the school has to sign an agreement and will receive conformation about which group they are in. As each school responds, they will be assigned as to whether they offer similarly priced healthy vs. non-healthy options or cheaper healthy options in the order that they are recruited, alternately. N=1 will offer healthy options cheaply, n=2 will not, and so on. To make it generalizable to the UK as a whole, the scheme will be offered to schools nationwide. This will give the study a cluster design, and each school will be statistically analysed as a cluster. It is felt that a maximum of 100 schools should be allowed to participate to reduce the costs involved with subsidising the meals whilst still providing enough data from which to draw a conclusion. Inclusion Criteria State secondary schools in the UK Schools which provide meals to their students at variable price Exclusion Criteria Schools participating in any other study Schools that already offer low-priced healthy options Schools that are running a healthy eating awareness campaign of any sort Schools which do not have the facilities to provide healthy and non-healthy options to students, such as small schools with no on-site catering facilities Trial Design This is a prospective explanatory cluster randomised controlled trial where schools will be randomised to either a scheme providing healthy school meals at a lower price will encourage young people to choose these options. It is thought that financial factors are important in obesity and unhealthy eating (Shepherd et al, 2006), and provision of cheaper healthy options should help to combat this income inequality. To help the schools offer the healthy options more cheaply, a subsidy will be offered by the trial co-ordinators to ensure that they are able to continuously offer these options at a cheaper rate throughout the 10-week trial. Over the course of 10 weeks, schools will offer healthy and non-healthy options at either a subsidised or a similar price. Each week, a total of how many healthy and non-healthy options were sold in each type of participating school. These will then be logged at each point to see whether healthy options become more appealing over time. At the end of the period, a total will be calculated to give a full value on how many healthy options have been sold where they are cheaper and where they are similarly priced. The food options will be classified into categories, to avoid confounding variables like drinks being more commonly bought than full meals, for example. This information will also be incorporated into the statistical data to ensure that the results are fair. When all this information is collected, a significance level of 0.05 will be used to analyse whether the results are statistically significant and may be beneficial to schools across the country. To prevent bias, schools that already promote healthy eating in some way will not be included in the trial. These students may already have some inclination towards choosing healthy options that will not be influenced by the price. As the aim of the study is to analyse the effect of price on student choice, this would act like a confounding variable. Similarly, schools that have already used this attempt or already offer lower-priced healthy options are excluded because this will not give a true representation of how the scheme would work in schools where this option is not currently used. Additionally, the students themselves are not to know that they are participating in the trial as this may influence their behaviour (Cardini et al, 2006). To prevent this, the knowledge of the trial and the assignment group should be kept to a minimum number of staff and should not be mentioned within earshot of the students. However, there is always the chance that the students may notice the price change and therefore this should be kept in mind when analysing the results. The intervention here is the cheaper healthy meals. This is in order to combat obesity in the young and promote healthy eating in schools, which will have a beneficial health effect. The use of the control is to provide a point for comparison between the schools to see whether providing these options cheaply is having an effect. The trial will involve pricing options such as chips, burgers and other typical ‘fast food’ alternatives at a higher price than salads and full balanced meals as assessed by a dietician. Additionally, fizzy, sweetened drinks will be priced higher than juices and bottled water, as these have shown to have a negative effect on weight and health (James et al, 2004). To prevent bias, there should be no advertising of certain food products during the campaign to minimise the impact that this has on the results of the study. This means that posters may have to be removed for the course of the study and vending machines altered. Catering staff should be instructed not to advise students on the taste or benefits of one type of meal over another, again to prevent bias between the options on any other factor apart from pricing. Question 3 The purpose of this RCT is to understand the difference between the outcomes of an NHS provided weight loss programme to a commercial one, according to the research question. One problem with this is that the question suggests that it is targeting commercial weight loss programmes in general, whereas in reality it is only examining one (Best Be Slim) which means that it is not answering the more general research question and the results cannot be generalised to commercial programmes in general, as they all have something different to offer the obese individual. Another major problem with the design of the study is the length. The research question refers to ‘long-term’ weight loss, but the study design only takes measurements from the individuals after 10 weeks of being on the program and at three months after the programme has ended. As obesity is such a problematic and long-term problem, it may be appropriate to take a further measurement after one year. Additionally, Best Be Slim usually offers 6 month and 12 month memberships, meaning that this programme is designed to work over a longer period, meaning that it may not be representative of how useful it would be pragmatically. The main measure against bias seems to be exclusion of those who attend a Best Be Slim session whilst being put into the NHS group. This is a good idea, as it would prevent mixing of the two groups. However, there is no detail as to how this will be enforced and therefore it seems to be relying on patient response, which may not be accurate for a number of reasons (Cardini et al, 2006). Additionally, the recruitment process is based on those who respond and show interest in the trial, who may be a certain ‘type’ of person and who may be more (or less) motivated to lose weight than other individuals. Again, this could lead to a bias towards a certain type of individual and alter the results of the trial. Another problem is with the inclusion and exclusion criteria, which are very general for an explanatory RCT (Jewkes et al, 2006). There is no distinction made between the weights and BMIs of individuals who fit into the obese category. It is mentioned that weight loss will be measured as a percentage, but it still does not account for the fact that those who are seriously overweight usually find it easier to lose weight (Jefferey et al 2010) and therefore may have more success in the programme. Additionally, age plays a part in weight loss and again there is no distinction made between those participating, just that they must be aged 18 or over, which could bias or skew the results. The randomization process described is good because it creates groups of equal sizes. There is also a low probability of confounding, because there is no distinction between who goes in which group (nothing which appears to be related to the age or gender of the participant, for example). However, it has a selection bias in that the investigator could easily predict which group an individual would be assigned to next by seeing which group currently had the lowest number of participants. However, it should be noted that all randomization methods have drawbacks (Prescott et al, 1999), and therefore this seems appropriate for this RCT. There is also a problem with the statistical justification for the sample size. For example, it does not refer to how big the population of Ambridge actually is. It is described as a small town, and it seems unlikely that there will be 166 obese patients over the age of 18 who are willing to take part in a study in such a scenario. It seems fair to say that there may be more information needed to see whether this is a fair goal, but from the information presented it does seem unlikely that this goal will be met and that it is appropriate for the purpose of the study. The statistical significance used for the analysis is a good one because it is standard. SPSS is also an extremely commonly used measurement and analysis tool, so it has a lot of previous research which suggests that it is appropriate (O Connor, 2000). However, it does not give details on how the statistical analysis will be done so not much can be said about whether there is bias in the analysis or not. Another problem with the study and the analysis is that the participants are required to complete food intake diaries. This is a completely different type of data than the blood pressure and weight loss values. Although an overall calorie value can be used, it would also be helpful to see what type of food the participants are consuming to see whether their behaviours have changed or simply their portion size. In this case, it is inferred that either this information is going to be ignored in favour of purely statistical analysis, or that there will be two types of analysis performed (one on the numerical values, one on the types of food the participants are having). Additionally, one part of the data collection relies on the patients completing a postal return for a 3-month follow up. This is always problematic because it requires that the participants actually complete and post the form. Much of the sample size could be lost at this point. It may perhaps be best to offer an incentive that will only be given to the patients after they have completed the 3-month follow up return, or to require them to go for a professional check-up as they did at the baseline and 10-week points. Additionally, the postal return system means that blood pressure values cannot be taken, but if these participants return for a full evaluation, these can be included in the data report as they give important clues about the health of the individual. Overall, it seems that there is a lot of reliance on the participants, which has shown to be a problem in past RCTs (Prescott et al, 1999), in that they have to return the form at the end, and they are responsible for volunteering for the study, and they have to keep food-intake diaries. This is perhaps the main problem with the study as often mistakes can be made by the participants when they are not controlled for. The main change to the study would be to make it longer to make it more appropriate to the importance of long-term weight loss as specified in the research question. References Cardini, F., Wade, C., Regalia, A.L., Gui, S., Li, W., Raschetti, R., Kronenberg, F., 2006. Clinical research in traditional medicine: Priorities and methods. Complementary therapies in medicine 14, 282–287. Hussey, M.A., Hughes, J.P., 2007. Design and analysis of stepped wedge cluster randomized trials. Contemporary Clinical Trials 28, 182–191. James, J., Thomas, P., Cavan, D., Kerr, D., 2004. Preventing childhood obesity by reducing consumption of carbonated drinks: cluster randomised controlled trial. Bmj 328, 1237. Jeffery, R.W., Epstein, L.H., Wilson, G.T., Drewnowski, A., Stunkard, A.J., Wing, R.R., 2000. Long-term maintenance of weight loss: current status. Health psychology 19, 5. Jewkes, R., Nduna, M., Levin, J., Jama, N., Dunkle, K., Khuzwayo, N., Koss, M., Puren, A., Wood, K., Duvvury, N., 2006. A cluster randomized-controlled trial to determine the effectiveness of Stepping Stones in preventing HIV infections and promoting safer sexual behaviour amongst youth in the rural Eastern Cape, South Africa: trial design, methods and baseline findings. Tropical Medicine & International Health 11, 3–16. Laurence, C., Gialamas, A., Yelland, L., Bubner, T., Ryan, P., Willson, K., Glastonbury, B., Gill, J., Shephard, M., Beilby, J., Committee, members of the P.T.M., 2008. A pragmatic cluster randomised controlled trial to evaluate the safety, clinical effectiveness, cost effectiveness and satisfaction with point of care testing in a general practice setting – rationale, design and baseline characteristics. Trials 9, 50. O’connor, B.P., 2000. SPSS and SAS programs for determining the number of components using parallel analysis and Velicer’s MAP test. Behavior Research Methods 32, 396–402. Peto, R., Darby, S., Deo, H., Silcocks, P., Whitley, E., Doll, R., 2000. Smoking, smoking cessation, and lung cancer in the UK since 1950: combination of national statistics with two case-control studies. Bmj 321, 323–329. Prescott, R.J., Counsell, C.E., Gillespie, W.J., Grant, A.M., Russell, I.T., Kiauka, S., Colthart, I.R., Ross, S., Shepherd, S.M., Russell, D., 1999. Factors that limit the quality, number and progress of randomised controlled trials. Health Technol Assess 3, 1–143. Shepherd, J., Harden, A., Rees, R., Brunton, G., Garcia, J., Oliver, S., Oakley, A., 2006. Young people and healthy eating: a systematic review of research on barriers and facilitators. Health Education Research 21, 239–257. Read More