Exploring the Reliability and Validity of the Methods: Randomization and Blinding - Essay Example

This essay will examine the methodology of above study exploring the reliability and validity of the methods. Two methodologies will be discussed; randomization, split into random assignment and allocation concealment. The second element to be evaluated is blinding. These are fundamental parts of the methodology the selection of which is very important. The use of appropriate methods for the type of study and what is being tested is good in order to maintain the right results (Akoben, 2005). Randomised Controlled Trials (RCT) is probably the most frequently used forms of studies in medical research. (Akoben, 2005) The study is a RCT evaluating the efficacy of a removal device (RD) with non-removal with total contact cast (TCC) in the healing of diabetic plantar foot ulcer ( Faglia et al, 2010). Forty five adult patients took part. The study was conducted at two centres specialising in diabetic foot management. As with all RCT there were precise inclusion and exclusion criteria. The outcome measure was the reduction in the size of the diabetic plantar foot ulcer. Randomised trials are widely considered to be the ‘gold standard’ for medical trials. The concept that participants are chosen randomly is considered to reduce the likelihood of bias producing the best results. (Akoben, 2005) This can be called into question, however, as they may not accurately represent what would happen in ‘real’ life. However, this particular study does not appear to suffer from this issue as it involves measurable treatments used in ‘real’ life settings, providing a fair test of the two treatments. The one aspect that cannot be truly measured in a trial is the difference in the results that would be achieved for each person using both methods. But it is impossible to apply similar tests on one person using different methods. (Akoben, 2005) Randomisation removes the option for participant preference and also any question that a particular treatment was applied to particular participants that would provide more favourable results thus creating bias (Popp & Deshaies, 2008). Randomisation is poorly reported in some studies, it is rarely stated or not stated at all how the sequence of numbers is generated (Bradfordhill, 2007). Randomisation is part of two processes. The first part is random assignment, the generation of unpredictable randomised sequences of numbers. The second part is allocation concealment. Although these processes are separate they are interrelated (Schulz and Grimes, 2002). Random assignment happens when patients are registering for the trial (Pocock, 1983). Different forms of random assignment can be chosen. Simple unrestricted randomisation is similar to the repeated action of tossing a coin. The advantage is that it is unpredictable but the treatment numbers can be unequal (Pocock, 1983). Then there is restricted randomisation which involves blocking of numbers to achieve lack of bias and ensures the groups are the same size (Pocock, 1983). Stratified randomisation is where investigators aim to avert imbalance in baseline characteristics by using pre randomised stratification. These are normally based around important patients prognostic factors such as the patient age or disease (Pocock, 1983). This is a good approach if the size of the group is small but it is not good for larger trials as the results are not feasible (Pocock, 1983). Random assignment in the design of this research was instrumental, in the choosing of the right specimen. This method lurking rules are that it works with two options that is either parts, be it heads or tails. In the research conducted by (faglia et al, 2010) he applied a principal of simple randomization by giving patients envelopes which contained code breaks of either non removable fiber glass or removable cast. Upon opening of the letters a patient would find only one option, and that would be their assigned group. The method was not well applied in this research, if I may mention though because all the rules of simple randomization were not followed. With simple randomization, it works best with huge populations, hence when the clinical research is big, this method generates equal numbers of personalities in groups. In small research specimens such as the one in this case, there were only 45 candidates, meaning one group was larger than the other, and for correct results, this is not the ideal situation. Though all the basics had been taken care of, even by generating random numbers for patients before printing the patients, the size was not right for this kind of method application. In this research (faglia et al 2010) we can conclude that random assignment failed since it failed to produce equal candidates in both groups. In the first group, that was assigned fiber glass the population was made up of 23 patients contrary to the removable cast group which had a population of 22 patients. Meaning that the results of the ultimate result had a significant margin of error, an ideal situation is where the population is equal and factors inherent in these populations are similar. The blocking process could have been used as this would facilitate the allocation between two centres (Pocock 1983). Because separate randomisation was performed at each centre, it would have been more feasible to activate the blocking method. However as it is not documented which process they followed regarding random assignment, they may have followed an inappropriate process such as randomised by date of birth or the patient’s hospital number (Schulz and Grimes 2002). Without random assignment carried out correctly variable bias can creep into the study (Pocock, 1983). If the researchers were using date of births or hospital number of a patient, it meant that random assignment was not effective. Random assignment is about chance, there is nothing that speaks of chance when choosing on a basis of birth dates. The researchers can be able to see whether a patient is old or young, they are able to pick a date and assign it a group. That is not chance, that is picking and assigning in full knowledge that a specific date was assigned a certain group. The process requires that there should not be a hint of relationship between randomness and anything else. The process should be based on chance purely. All the underlying factors and characteristics within the target group need to be the same such that the baselines are the same. In the group that the researchers (faglia et al 2010) chose, at least they ensured that the specimens had similar characteristics. On both groups the researchers choose individuals that were suffering from foot ulcers, all the patients required treatment. None of the patients were outside hospitals; all of them were from the hospitals. Though there is one factor that couldn’t be determined and that is whether the patients were at the same level of treatment and disease progress. Since this would have the effect of some patients having slow healing rate than others and thus this will affect the final result. Null hypothesis in the two groups cannot produce nil since the two groups are not equal in size. Allocation concealment within the study used code break randomization envelopes containing one of the two options. It was not stated what the envelopes looked like. If the small scale salary format cannot be opened without breaking the seal limiting opportunities to tamper with them. The researchers were present at the randomisation but it does not record that they were blinded leading me to presume they were part of the process. Also they were involved in allocation of the envelopes so bias cannot be ruled out. The system is not fool proof, if the researchers are not blinded from the treatment allocation they can cheat by holding the envelopes up to a bright light (Gotzsche, 2008). Moreover, many participants on the same day could mean that several envelopes are opened at the same time, so the envelopes could be switched to favour bias (Gotzsche, 2008). Once the envelopes were opened they were taken back to the statistical centre for security. In this study the days and the number of envelopes opened on each day was not stated. Allocation concealment is a process favoured whenever none of the parties can be in a position of knowing the process beforehand what will happen and how the process will take place. Facts are that some researchers have the ability to manipulate results by choosing some groups that have strong candidates than the rest. Hence allocation concealment seeks to minimise information known to researchers so that during sampling, there will not be hiding of specimen considered weak. In the research I am reviewing in this paper, like mentioned earlier, the process did not meet, ideal situation criteria. Researchers had already done randomization on the envelopes, and the envelope code criteria are wanting. If randomisation is done correctly bias is eliminated among the two groups upon ingress to the trial as well as when the patient is assigned the treatment. Randomisation helps blind the identity of the treatment allocation from the participants, investigators and the assessors (Pocock, 1983). The result is that the outcome between the two treatment groups is due to chance following all the directions above (Pocock, 1983). Omitting statically data leaves me suspicious as to whether the authors were covering up the data. Despite this the results did not find any significant regarding ulcer size or reduction, so there was no difference between the RD and the TCC The second element to be discussed is blinding which in research contributes towards the outcome measures within each research done. Blinding is important because it minimises bias but maximizes the validity of the research. (Schulz and Grimes, 2002) The concept of a blind trial is that the participant does not know which treatment they are receiving. The ultimate ‘gold standard’ is the double-blind test where the researcher and participants know nil about the kind of treatment to be received (Schulz and Grimes, 2002) Participants who are not blinded to the treatment may alter their behaviour in the trial, this then may go on to affect the outcome measure (Karachicilas et al, 2010). Subject effect means the participants try to be good subjects trying to see what the researcher and adjust their behaviour accordingly. (Chow and Liu, 2014 ). In the case of the chosen study the trial could not be a blind trial as there were physical treatments used that would make it clear which was being used. In the faglia research however it is possible that the reactions towards wearing the RC could have had higher compliance than is recorded in other studies. Other papers record there has been lower compliance with RD and is not as highly regarded as the gold standard TCT (Armstrong, 2001). The outcome in this study reveal that there were no differences in ulcer healing between the two groups. It would be wise and conclusive to do the same study where participants were wearing the same device but not aware that they were part of a study blinded, this however would be unethical (Chow and Liu, 2014) Blinding has also to be considered for the researchers. Researchers if blinded are less likely to transfer bias on to participants (Karahicilas et al, 2010). In this study, as the RD were produced and given to the researchers for the purpose of the trial it is possible that their bias to the RD was passed to participants. Moreover, this conflict of interests was stated at the end of the article. For this reason the results could be affected by bias in favour of the RD. In this study researchers were not blinded from any stage of the intervention. Because many RD trials are funded by large research bodies they can dictate the research agenda (Greenhalgh, 2010). The conclusion from the paper was that the RD was just as effective as the TCC regarding the reduction in ulcer size and total healing rate. If the company had not supplied the RD the information could have been less biased .Also if the researchers had been blinded to the participants this may have had different results (Greenhalgh, 2010) Absence of blinding can also affect the gathering of the data. The data collector and the outcome measurer are normally the same person. (Karachicilas et al 2010) Blinding this person is important to ensure that the data collected and the outcome measures will be unbiased (Karachicilas et al 2010) However, I found it interesting is that less than 10% of trials actually blind the assessors in research. (Karachicilas et al 2010) .If assessors cannot be blinded then assessors should be used to be as objective as possible in their approach to the outcome data. In this study the data collector was not blinded, rendering the outcomes questionable. However, as the devices were removed at the weekly visits I consider it would have been possible for the assessor taking photographs and measuring the ulcer using the Visitrak system, to be blinded. In an ideal situation blinding targets a minimum of five groups. In the case of faglia, blinding should have happened to clinicians, data collectors, participants and result analysts. Blinding assists in making the participants an aware of the assignment, so that they may not alter their behaviour, if blinding does not occur properly this leads to participants altering their behaviour or the researchers manipulating the outcome of the result to favour their research. As the researchers were involved in the statistics, there is the possibility that the tests selected would have been more likely to favour the RD. If the researchers were blinded to the statistical outcomes the research would have been less biased (Pocock, 1983). For example the study used the Mann-Whitney test to examine the ulcer size surface reduction for each group was two P Read More