Practice Nurse

Evidence Based Medicine

Evidence-based practice is a growing component of our ability to deliver high quality clinical services. Based on our knowledge and interpretation of studies we are able to continually deliver care to the highest standards. The interpretation of the wealth of published clinical data is therefore important to ensure the correct incorporation of change into our practice. Thus, understanding the basics of the evidence that is produced from published clinical data, in essence, should enable us to remain contemporary and skilled.
This series of 10 questions focuses on the interpretation of typical clinical data that would be presented in clinical trials. Complete the 10 assessment questions, reflect on the questions and obtain a certificate offering one hour of CPD.

Understand, calculate and interpret typical terms used in data presentation such as sensitivity, specificity, Number Needed to Treat, Absolute and Relative Risk Reduction.

Interpret the presentation of data in graphical form such as Kaplan-Meier Curves, Forest Plots

Understand the types of trials used in clinical studies and their utility

Interpreting p values and statistical significance

Evidence-based practice is a growing component of our ability to deliver high quality clinical services. Based on our knowledge of and interpretation of studies we are able to continually deliver care to the highest standards. The interpretation of the wealth of published clinical data is therefore important to ensure the correct incorporation of change into our practice. Thus, understanding the basics of the evidence that is produced from published clinical data in essence should enable us to remain contemporary and skilled.

On completion of this module you should have a better understanding of the following:

The definition and interpretation of typical terms used in data presentation such as sensitivity, specificity, Number Needed to Treat, Absolute and Relative Risk Reduction.

The presentation of data in graphical form such as Kaplan-Meier Curves, Forrest Plots

The types of trials used in clinical studies and their utility

This Forest plot demonstrates the results of a meta-analysis of the effects of a specific treatment on macrovascular events. Which of the following conclusions can be drawn from this study?

This is a forest plot which provides a graphical representation of information gathered from studies. Here the studies have looked at the effect of a new treatment on macrovascular disease. The y axis (vertical) at 1 on the x axis (horizontal) means that there is no effect of treatment on macrovascular disease. To the left of 1 means that the new treatment is beneficial, whereas, to the right of 1 means that the treatment is detrimental. The lines represent each of the 4 studies and are referred to as the confidence intervals. In this case the meta-analysis of all the data (overall) is to the right of the y axis corresponding to approximately 2 indicating that the new treatment is associated with increased macrovascular risk such as cardiovascular disease.

Question 1

A study reports on the results of a large study of the primary prevention of stroke in a diabetic population using a new antiplatelet agent versus aspirin. The results of the study reveal that over a 5 year period the incidence of stroke in the aspirin treated group is 3% compared to a rate of 1.5% in the group treated with the new antiplatelet agent (p<0.001).
What is the absolute risk reduction in stroke associated with the new drug?

Absolute risk reduction is another important term in the interpretation of clinical data. It implies the absolute impact of a new treatment on an end point compared with another treatment or placebo. In this example, the incidence of stroke with aspirin is 3% but this is reduced to 1.5% with the new treatment over the 5 year study. Therefore, the absolute risk reduction is 3-1.5% = 1.5%

Question 2

A study of the effects of a new treatment reports on the results of a large randomised controlled trial of the secondary prevention of myocardial infarction in patients compared with aspirin. The results of the study reveal that over a 5 year period the incidence of MIs in the aspirin treated group is 5% compared to a rate of 4% in the group treated with the new antiplatelet agent (p<0.006).
What is meant by secondary prevention?

Primary prevention refers to the prevention of a first episode whereas secondary prevention refers to the prevention of a further event. This study relates to the secondary prevention of myocardial infarction. This means that the treatment is trying to prevent myocardial infarction in subjects who have already suffered a previous MI. Examples of such secondary prevention studies include the 4S study.

Question 3

A randomised controlled study of the effects of a new therapy in the treatment of osteoporotic fractures in the elderly reveals that the new treatment is associated with a 5% annual risk of fractures compared with a 13% risk of fractures with the standard therapy (p<0.005).
What is the approximate number needed to treat to prevent one fracture with the new agent?

Number needed to treat (NNT) is a commonly used term which gives an insight of the effectiveness and economic benefit of the use of a drug compared with a standard. The ideal would be one where everyone who is treated benefits. The lower the NNT the more effective the treatment. In this case we can calculate the number needed to treat quite simply by reference to simple calculations. If one hundred elderly people were treated with the new drug then one would expect 5 fractures (5%) compared with 13 fractures with the standard therapy. Thus the absolute difference would be 13-5 = 8 fewer fractures per 100 people treated with the new drug. Thus to prevent 1 fracture compared with standard treatment one would need to treat 12.5 people (100/8) with the new treatment.

Question 4

In a study of cardiovascular risk over a five year period, 1800 patients with type 2 diabetes were randomly assigned to a new glucose lowering agent together with metformin or metformin alone. In the study there was an extra 0.5% reduction in HbA1c with cardiovascular events occurring in 122 patients of the 1000 patients treated with the new drug vs 132 in the 800 treated with metformin alone.
What is the approximate relative risk reduction of cardiovascular events associated with the new drug over the duration of this study?

Relative risk reduction is another important term used in the interpretation of clinical data and evidence based medicine and provides an idea of how effective a treatment is in comparison with another. In this study the rate of cardiovascular events was 122/1000 (0.122) with the new treatment versus 132/800 (0.165).
Therefore, the relative risk reduction of the new therapy can be calculated as:
Absolute risk reduction = risk with old treatment minus risk with new treatment
= 0.165-0.122 = 0.023 ie Absolute risk reduction is 2.3%
Relative risk reduction is absolute risk divided by risk with old treatment = 0.023/0.165 = 14% - so approximately 15%

Question 5

In a review of glycaemic control associated with the use of a continuous insulin infusion device, a group investigates the published data relating to controlled studies in type 1 diabetes using the device and conclude that there is an effective reduction in HbA1c of approximately 1% vs the use of standard subcutaneous insulin bolus regimes. What is this type of study called?

This type of study where the researchers study all the publications that relate to a particular treatment or device is called a meta-analysis. Meta-analyses are very useful for pooling together all the data that relate to certain treatments that in themselves may be too small to provide meaningful data yet when analysed together can provide a clear indication of the usefulness or not of the treatment.

Question 6

A study reports on the effect of a new antihypertensive in the prevention of stroke in an elderly population. The study compares the new antihypertensive with standard antihypertensive therapy and concludes that there is an 8% relative risk reduction of stroke with the new therapy compared with the old therapy which is statistically significant p=0.01. What does p=0.01 mean?

Statistical significance ie where the result is not considered likely to have occurred merely by chance, is assumed in the vast majority of papers to occur at a p (probability) value of 0.05 ie a 5% chance of the results occurring by chance is 1 in 20. Clearly, the lower the p value the less likely it is that the results purely occur as a matter of coincidence. In our case the results are highly significant at a p value of 0.01 ie a 1% chance of coincidence being responsible for the results is 1 in 100.

Question 7

This Kaplan-Meier Curve reveals the survival characteristics of two groups of patients treated for breast cancer. Group 1 is treated with standard therapy and group 2 is treated with a new treatment designed to target a novel receptor. The data is collected over a 35 month period and there is a statistically significant difference between the groups (p<0.0001).
Which of the following can be determined from the data presented?

The Kaplan-Meier survival curve is an excellent graphical representation of the impact of interventions upon survival/outcomes over time. In this case we are faced with the data relating to a new treatment of breast cancer (Group 2) compared with standard treatment (Group 1). In this case we can see that at the end of the 35 month study approximately 40% of Group 1 are alive versus none in Group 2. This is significantly different p<0.0001 and indicates that there is a reduced survival in patients treated with the new drug

Question 8

A study looking at the effect of a new medical intervention on weight loss compared with diet alone reveals that there was an average 1.1kg weight loss associated with the medication (95% confidence interval -0.5. What does this confidence interval mean?

Confidence intervals are another term frequently used in evidence based medicine. On reviewing confidence intervals it is worth considering two things.

Does the confidence interval include 0? If it does then it is unlikely that the data is statistically significant. In our case the 95% confidence interval is quite wide - indicating that 95% of people in this study lost between 5.6kg in weight or gained 0.5 Kg in weight with the average weight loss being relatively trivial at 1.1kg

Is the data of practical significance and how wide are the intervals? In this case the 1.1kg weight loss is really rather small over the study period. Similarly the width of the confidence interval implies insufficient numbers in the study.

Question 9

A diagnostic test has been developed to supersede the oral glucose tolerance test (OGTT) as a means of diagnosing patients with diabetes. The following data reveal the results of this study in a group of 100 obese volunteers who have undergone both tests. The results reveal those patients that were diagnosed with diabetes on both tests as testing 'positive'.

Which of the following is the approximate sensitivity of the new test in diagnosing diabetes?

Sensitivity and specificity are terms frequently used in the interpretation of clinical data

 is the proportion of people with the disease who have been positively diagnosed with the disease using the new test. In our case we are using the oral glucose tolerance test as the gold standard diagnostic test.
Sensitivity = a/a+c = 12/14 = 86%


 relates to the tests ability to identify true negatives. In this case the specificity isNumber true negatives with new test (84)/No True negatives (84) + false positives(2) = 98%

Evidence Based Medicine

title

title

title

title

title

title