Medical Statistics: the MMR vaccination and the onset of autism - Research Paper Example

Critical Appraisal - Medical Statistics Article Smeeth et al. The paper was completed in order to assess whether there may be a link between the MMR vaccination and the onset of autism. This was identified as an important issue due to the concerns which had been raised over safety after the link was suggested. The objectives were to assess whether those with autism were more likely to have received the MMR vaccination and to determine whether there was any association between the timing of the vaccination and the onset of symptoms of autism. The paper presented a primary study as the research was new rather than an integration of previous studies. The data was collected through a survey which was distributed to all parents identified on the General Practice Register Database (GPRD) as having an autistic child. The study was also distributed to twice the number of parents with non-autistic children to provide a control sample. The study is not original in its intentions. The literature review at the beginning of the study highlights several previous studies which have been conducted into the possible link between autism and the MMR vaccine, all of which were recent to the study. The approach to the study is also not original as there were several of the studies identified in the review which adopted a similar approach to this study. The researchers felt that there was a need to conduct the study from a slightly different perspective as there is a great deal of controversy identified in the previous studies. For this reason the study sought to understand if there was a connection between the timing of the MMR vaccine and the onset of autism. This may add to the findings of other meta-analyses in the future, or may provide a step for further research if it can be identified that there is or is not a connection between the time at which the MMR vaccine is given and the time at which autism symptoms begin to show. The design of the study is appropriate to the purpose of the study, as the purpose is related to causation, and the survey may enable the researchers to establish a set of factors which are characteristic to the autistic cases. This would then reveal a potentially causative mechanism. The subjects were identified through the GPRD. The study does not explicitly describe whether the children came from across the UK or a selected area. The inclusion criteria were being born mid-1987 onwards, which ensured that the child was young enough to receive the MMR vaccine; children were first diagnosed when registered with practices participating in the GPRD. Autistic children were excluded if there was an alternative aetiology which had already been identified. Controls were selected on the basis of matched characteristics for age, sex and practice to the participants selected for autism. The study is of the case-control form, where there is a comparison made between the autistic children and the control children to which they are 'matched'. This is appropriate to the purpose of the study as it is the ideal strategy to use for studies which are concerned with the aetiology of a disease. The difficulty associated with case-control studies lies in the definition of the 'case'. This study presents a strong definition of autism, and it is unlikely that any child would be classed in the wrong group. Therefore it is unlikely that the results of the study would be affected by misrepresentation. The one flaw with using this design for the study is that it is not possible to show causality through a case-control study. Although the study can show that there is a potential link between factors such as the MMR vaccine and autism it cannot definitively conclude that the MMR vaccine causes autism, and likewise that it does not cause autism. Overall the structure of the study ranks low on the hierarchy of evidence, but is suitable to meet the objectives of the study. The measures which were used in the study were the onset of autism symptoms, and exposure to the MMR vaccine. Confounding factors were identified which may affect the uptake of the vaccine, such as social class and family size although it is unclear how this is relevant to the outcome of the study. Due to the lack of knowledge relating to factors which may affect the onset of autism, there was little consideration given to these in the analysis of the data. Therefore it is possible that this may have affected the level of bias in the data. This was unavoidable as the evidence is not available to allow for thorough removal of factors which bias towards autism development. The matching of the cases using five controls for each participant ensured that some factors were controlled for such as age, sex and practice, although it is possible that there are other factors affecting the outcome which were not controlled for. Overall the bias was controlled as far as was possible at that time. The total sample size which would be used would 400. This would be considered a large enough number from which to conduct accurate statistical analyses. The large number of participants should ensure that it is possible to identify any statistically significant links between the MMR exposure and onset of autism if it exists. Overall, the study was fit for the purpose and would be likely to achieve the objectives outlined. There were however some criticisms of the study. For instance the study is not original and as such does not present an original approach to the topic of study. Therefore it is unclear whether the study will contribute significantly to the field. There is also a lack of some data in the study which makes it difficult to assess the overall reliability and validity of possible results, such as the lack of demographic information. Overall I would consider the study to score 70 out of 100. Article 2 - Rudolf et al. It is unclear due to lack of information as to the exact purpose of the study. Given the details of the methodology it is assumed that the purpose of the study was to assess whether there is an increase in obesity prevalence in children when compared to previous years. The lack of an abstract, introduction or literature review makes this difficult to assess. The study is a primary study, where the researchers have collected all of the information themselves by directly measuring the children. The study adopted the form of a cohort study in order to assess basic differences between the population today and the population several years ago from known data. The research design was appropriate for the assumed purpose of the study. Measurement of physical characteristics is suitable for direct comparisons between two population samples and is suitable for constructing statistical analysis of proposed differences between the samples. It is unclear from the study whether the research is original. The reference to a different study which showed that obesity prevalence was higher than expected would suggest that there has been previous research conducted into the levels of childhood obesity but it is unclear from this how similar the research was to the paper presented. The lack of a literature review suggests that the research was entirely original although this is unclear. The use of one specific area for the study suggests that even if the research is not wholly original it is likely to add to future meta-analysis by providing detailed data from that area. The study used participants from 10 primary schools in Leeds who were aged between 7 and 9 years old. The demographics of the participating population were given, with 1-42% of children from ethnic minorities in each school and 7-29% receiving free school meals. The wide ranges of this data suggest that there schools would create a broad representation of children of this age in Leeds. Overall 608 children participated in the study, which were all but 21 children from this age group in the schools. This was due to lack of parental consent rather than intentional exclusion from the study. There were no medical or other interventions required for the study therefore it was carried out at the actual schools themselves. This would not have affected the outcomes of the study, however, which was concerned purely with physical measurement. The outcomes measured in the study were the physical dimensions of the students. Measurements were taken of the children's height and weight. The triceps measurements of each child were taken. Finally, Body Mass Index (BMI) was calculated for each child. These measurements were validated through the precise description of methods in the study. The equipment which was used was named, as was the technique used in BMI calculation. This ensures that should follow up study be required the same methodology can be followed to produce comparable results. The measurements which were taken were objective in nature, and the supply of information regarding their derision ensures that bias is removed from the study. The measurements were taken three times, annually between 1996 and 1999. The results were compared to the standards which had been constructed in 1975 to assess the level of change. The selection of a comparable control group is one of the crucial elements in a cohort study. In this case there were two possible comparisons which were identified. The first is the set of standards which was constructed in 1975. The study suggests that these may have been based upon overweight children although there is no clear evidence that this is the case. Even so, there is no clear evidence whether the other characteristics of the group are the same as in this cohort study, such as the ethnic mix in the original study. This is important as factors such as this are known to impact upon obesity levels within a given population, allowing bias to present in the study. The unclear nature of the match between the two samples may lead to incomplete or inaccurate conclusions through comparison with unsuitable previous studies. The study also mentions the set of data which was constructed in 1990, although it is unclear whether the data from this study was compared to that sample. If this was not performed, it is unclear what factors led the researchers to conclude that this was not an appropriate sample and also to conclude that the 1975 sample was a suitable comparison. It may be that the researchers did not consider there to have been a sufficient lapse of time since 1990 for effective analysis and that may be why they selected 1975 for comparison, as changes are likely to be more pronounced over a longer time period. The sample size may be large, but when considering the overall number of children of that age group within the UK it is debatable whether the sample size is large enough to draw statistically significant conclusions relating to the whole of the UK. The use of participants only from one city also makes the study unsuitable for drawing conclusions about the entire UK as there may be other socio-demographic factors involved in obesity which bias any sample drawn only from Leeds. Overall the cohort study ranks in the middle of the hierarchy of evidence, which would suggest that it would be likely to present better evidence than a case study. In this cohort study there is however a lack of critical information which renders the cohort study less useful than it could potentially be. Overall I would rate the cohort study as only 40 out of 100 due to the severe lack of information. References Rudolf, M.C.J., Sahota, P., Barth, J.H. and Walker, J. (2001) Increasing prevalence of obesity in primary school children: Cohort study. BMJ, Smeeth, L., Hall, A.J., Fombonne, E., Rodrigues, L.C., Huang, X. and Smith, P.G. (2001) A case-control study of autism and mumps-measles-rubella vaccination using the general practice research database: Design and methodology. BMC Public Health, 1, doi: 10.1186/1471-2458-1-2. Read More