Mediterranean Diets and Cardiovascular Disease Analysis Coursework

By January 3, 2019 Academic Papers

v:* {behavior:url(#default#VML);}
o:* {behavior:url(#default#VML);}
w:* {behavior:url(#default#VML);}
.shape {behavior:url(#default#VML);}




/* Style Definitions */
{mso-style-name:”Table Normal”;
mso-padding-alt:0cm 5.4pt 0cm 5.4pt;
{mso-style-name:”Table Grid”;
border:solid windowtext 1.0pt;
mso-border-alt:solid windowtext .5pt;
mso-padding-alt:0cm 5.4pt 0cm 5.4pt;
mso-border-insideh:.5pt solid windowtext;
mso-border-insidev:.5pt solid windowtext;



Total marks for this assignment: 60


This assignment counts 30% of the total grade for the subject.



Word limit: 1500 words.  Please include a word count.


For every 150 words (10%) you exceed the word limit, you will be penalized 10%.


Notes on completing this assignment:

       Please use 12-point font and format your document in portrait mode

       Unless instructed otherwise, justify all your answers

       Your answers should be brief.  You can use dot points where appropriate

       Don’t repeat the questions in your completed assignment.  Just number each answer

       Please answer the questions in order – it simplifies the marking process

       Write the answers in your own words.  If you must use text from the paper, put it in quotes and write something like: Rosata et al. stated that “the sky is blue”.


The following questions relate to this systematic review and meta-analysis:


Rosata V, et al.  Mediterranean diet and cardiovascular disease: a systematic review and meta-analysis of observational studies.  European Journal of Nutrition 2017;




1.     Search strategy                                                                                                [2 marks]


Copy the authors’ search strategy into the search field in PubMed.  Here is the link to PubMed:  The search field is at the top of the page.



When you have done that and the search results appear, click on “Custom Range” under publication dates and in the “to date field” enter 2016/08/31 as shown below and then click apply.  The date is required because they state in the paragraph that they sought “studies published up to August 2016”.



How many publications does this give?  Is this the same number of publications as reported by the authors in the flow diagram and at the beginning of the results?


2.     Based on the previous answer, does this indicate that the authors provided a full electronic search strategy for at least one database, including any limits used, such that it could be repeated?                                                        [1 mark]


3.     Would you consider the search exhaustive in terms of databases?                    [1 mark]


4.     Were the eligibility criteria for studies clearly specified?                                   [1 mark]


5.     Does the flow diagram give reasons why studies identified from the search were excluded?       [1 mark]


6.     Was each study assessed for risk of bias?                                                         [1 mark]


7.     The authors state that their review was better than some other systematic reviews that only included cohort studies because they also included case-control studies.  Many epidemiologists argue that if there are sufficient cohort studies reported, case-control studies should not be included in a systematic review[1].  Why? Your answer must refer to diet (i.e., it should not be theoretical only)         [4 marks]


8.     Table 1: Characteristics of the studies                                                            [10 marks]


What additional information would be useful in Table 1 to assist the reader assess each study’s risk of

selection bias        

information bias with respect to measurement of Mediterranean diet



Please enter your answers into the following table.  The number of marks for each section is given in the table and corresponds to the number of points you need to mention.



Cohort studies

Case-control studies

Selection bias

1 mark


3 marks


Information bias (measurement of diet)

1 mark


2 marks



1 mark


2 marks



9.     Variables adjusted in the analyses                                                                   [4 marks]


In the methods, the authors state: “The RRs adjusted for the largest number of confounding factors were included in all the analyses”.  What this means is that if a paper presented results from two analyses, one adjusted for three variables for example, and the other for four variables, they extracted and used the latter.


What is the potential bias resulting from extracting and using the estimates from the analyses with the most number of variables? Draw a DAG (causal diagram) that illustrates the issue. {Hint: recall the lecture on confounding and the tutorial on systematic reviews} 


10.  Heterogeneity                                                                                                  [4 marks]


For each outcome, copy the I2 value from the article into the table below and describe the amount of heterogeneity. 




Description of amount of heterogeneity




Stroke: unspecified



Stroke: Ischemic



Stroke: Haemorrhagic



CVD unspecified




11.  Sources of heterogeneity                                                                                 [3 marks]


Table 2 shows the results of analyses designed to explore reasons for heterogeneity in the associations between MD and CVD.  Ignore the comparisons between cohort and case-control studies because there are too few case-control studies to be informative.  Also ignore comparisons for stroke because there are few studies on each type of stroke.  Otherwise, do the RRs for each category of the factors considered (incidence vs mortality, women vs men, etc) vary materially?  What does this tell you about whether these factors were sources of the heterogeneity across studies.


12.  Publication bias.                                                                                              [4 marks]


For these questions, consider the outcome CHD/AMI only.


The authors used statistical testing to assess the potential for selection bias.  We will ignore the testing.


(For those interested, the statistical tests are designed to test the (null) hypothesis that there is no publication bias.  The numbers (e.g., 0.034 from Egger’s test for CHD/AMI) are called p-values.  A small p-value provides evidence against this hypothesis.  The smaller the p-value, the stronger the evidence against the hypothesis of no publication bias.  For unspecified CVD, there is little evidence against the hypothesis, while for CHD/AMI there is moderate to strong evidence depending on the test performed.)


The authors did not include funnel plots in the publication.  Shown below is the funnel plot for the outcome CHD/AMI.  What does this funnel plot indicate about the potential for publication bias for this outcome?  In which direction would the observed pooled estimate of the RR be biased?  Recall that the pooled estimate of the RR for this outcome was 0.70.


Figure: funnel plot showing study-specific RRs and standard errors of the log(RR) for the outcome CHD/AMI.


Bradford Hill guidelines for assessing causality


13.  Consistency                                                                                                     [5 marks]


How consistent is the evidence that adhering to a Mediterranean Diet is associated with reduced risk of cardiovascular disease?  Is the degree of heterogeneity consistent with a causal interpretation of the association?  Use the Forest plots and your answers to the questions on heterogeneity and sources of heterogeneity above in responding to this question.


14.  Temporality                                                                                                     [3 marks]


How strong is the evidence that the exposure precedes the outcome?


15.  Strength of the association:                                                                              [3 marks]


How strong do you consider each of thefive associations to be? Use the pooled estimates of the relative risks for CHD/AMI, unspecified stroke, ischaemic stroke, haemorrhagic stroke and unspecified CVD.  Are the associations strong enough that you could confidently rule out the possibility of selection bias and confounding as explanations?


16.  Dose-response relationship                                                                             [3 marks]


What evidence does the paper provide of a dose response relationship between Mediterranean diet score and risk of cardiovascular disease?


17.  Biological plausibility                                                                                      [2 marks]


Do the authors attempt to make a case for the biological mechanisms by which adhering to a Mediterranean diet might reduce risk of cardiovascular disease?                                                                                                                                


18.  Evidence to action.                                                                                           [8 marks]


The Australian Government is considering a new health promotion campaign to reduce the incidence of cardiovascular disease.  Because all health promotion campaigns have costs and some have unexpected adverse consequences, the decision to introduce a program should be based on firm evidence.


Based only on this systematic review and your responses to the previous questions, do you think the evidence that the Mediterranean diet reduces the risk of cardiovascular disease is of sufficient quality for the Australian Government to promote this type of diet? If yes, why? If not, why not?  Please write between 100 and 150 words summarising the evidence and stating your conclusion.   {Assume that the results of the studies would be applicable to Australia.}


[1]For example, this is the approach adopted by the World Cancer Research Fund in its reviews of the literature pertaining to diet and risk of cancer.