EvidenceBased Practice David Pfleger NHS Grampian Nonmedical prescribing

Evidence-Based Practice David Pfleger NHS Grampian Non-medical prescribing conference 2011

Definition “ Evidence-based medicine is the process of systematically reviewing, appraising and using clinical research findings to aid the delivery of optimum clinical care to patients. ” Rosenberg and Donald 1995

Evidence Based Practice Evidence based practice is the integration of best research evidence with clinical expertise and patient values Best evidence invalidates and replaces previously accepted treatments, tests, practice and replaces them with better ones Clinical expertise - clinical experience and skills to ID patient’s individual situation Patient values - patients preferences, concerns and expectations

How – according to Sackett Step 1 – converting the need for information (about prevention, diagnosis, prognosis, therapy, causation etc) into an answerable question. Step 2 – sourcing the best evidence in order to answer the question Step 3 – critical appraisal of the retrieved evidence concentrating on validity (closeness to the truth), impact (magnitude of the effect) and applicability (usefulness to the clinical situation) Step 4 – integration of the results of appraising the evidence with clinical expertise and the patient’s (individual or population) Step 5 – evaluation of effectiveness in completing steps 1 -4 and reflecting on needs and formulating a plan for improvement.

What is the Evidence? I Strong evidence from at least one systematic review of multiple, well-designed, randomised, controlled trials II Strong evidence from at least one properly designed, randomised, controlled trial of appropriate size III Evidence from well designed trials without randomisation , single group pre- or post(treatment), cohort studies, time series, matched, casecontrolled studies IV Evidence from well designed , non-experimental studies from more than one centre or research group V Opinions of respected authorities based on clinical evidence, descriptive studies or reports of expert committees

Why Are Systematic Reviews Needed? l Around 20, 000 journals and 2. 5 million biomedical articles published each year l Impossible to keep up with in detail, or even to interpret at times l Results of trials may be unclear, confusing or downright contradictory l Taken together, a clearer picture can emerge

How Are Systematic Reviews Conducted? - I l Define the question - intervention, patient group, settings l Search the literature - seeking an unbiased assessment, including non-English sources, and trying to avoid publication bias l Assess the studies – eligibility – study quality – independent assessors

How Are Systematic Reviews Conducted? - II l Combine the results - sometimes qualitative, but usually numerical, using meta-analysis l Put the findings in context – quality of the studies – differences between them – impact of bias and chance l Can be very substantial undertakings - they cost money

Assessing the studies: Study Design Descriptive studies – Correlation, case reports, case series Analytical studies – Observational Case – Control Cohort – Intervention / Experimental RCT – Meta analysis

Case Control Studies Time Direction of Enquiry Exposed Not Exposed Cases (people with the event / outcome) Population Exposed Not Exposed Controls (people without the event / outcome)

Case Control Studies Positive Rare outcomes Small sample sizes Good follow up Can look at multiple risk factors / exposures Relatively fast, easy and cheap Negative Selection of controls must be representative of exposure in general population - selection bias Retrospective data on exposure - recall bias

Cohort Studies Time Direction of Enquiry Event / outcome Population People without outcome Exposed No event / outcome Unexposed Event / outcome No event / outcome

Cohort studies - Strengths / Weaknesses Positive Subjects can be matched Time sequence Validity of data Multiple outcome measurement Negative Large cohorts for rare events Long follow up? cost? Loss? Confounders? No randomisation No blinding

Randomised Control Trials The Gold Standard A population is selected and randomised in to two groups. One group receives the intervention whilst the other does not (or receives placebo). The outcomes for each group are then compared.

RCT Positive Confounders distributed randomly Blinding more likely Randomisation makes statistical analysis simpler Negative Expensive, time & £ Volunteer bias Ethical issues Loss to follow up Artificial questionable advantage over prestudy

PICO-T – Helping formulate the question PICO-T – Population/Patients – Intervention – Comparison – Outcome – Time

The RAMMbo acronym: assessing study bias Recruitment P P Allocation E E C C Measurement of outcomes blind O T Maintenance T O or objective

Interpreting Risk 200 subjects aged 59 years or older, with previous heart disease and type 2 diabetes randomised to two groups: – 100 receive the experimental treatment – 100 receive the control treatment Follow-up is a mean of 5 years Endpoint is a composite of all of the CHD deaths and non-fatal MIs

Results Number of subjects Number of CHDdeaths or non-fatal MI Probability of CHDdeaths or non-fatal MI Treatment group 100 20 Control group 100 30 0. 2 (20%) 0. 3 (30%)

Risks to be considered Rate of CHD-deaths or non-fatal MI Treatment group 0. 2 (20%) Control group 0. 3 (30%) RR = 0. 2 / 0. 3 = 0. 67 – With treatment were 0. 67 times as likely to die as the control group i. e. risk reduced Attributable Risk = 0. 3 - 0. 2 = 0. 1 – Your actual risk of dying from CHD-related causes or suffer a nonfatal MI has dropped from 30% without treatment to 20% with i. e. the actual risk has reduced by 10% due to the treatment. Number needed to treat = 1 / 0. 1 = 10 – We would need to treat ten people for one of them to avoid dying from CHD-related causes or suffer a non-fatal MI

What if the baseline risk is lower? Number of subjects Number of CHDdeaths or non-fatal MI Rate of CHD-deaths or non-fatal MI Treatment group 100 2 Control group 100 3 0. 02 (2%) 0. 03 (3%) RR = 0. 02 / 0. 03 = 0. 67 ARR = 0. 03 - 0. 02 = 0. 01 NNT = 1 / 0. 001 = 100

Estimation Rare that we observe the entire population Normally we use a representative sample Ideally the mean, risk or whatever we are looking at would be the same in the sample group as the entire population Samples are not ‘perfect’ therefore this is not the case

Confidence Intervals A range within which the real population mean or risk is likely to fall We usually use the 95% confidence interval 95% CI says that we are 95% confident that the true mean falls within those limits Important to see – How spread out those limits are – What clinical significance those limits have

Statistical inference Hypothesis testing – How likely are observed differences a result of sampling errors rather than real differences between the groups States that any differences between two groups are due to chance i. e. there are no real differences between the groups Normally one group is the study group and one is the control Looking for a real difference between the groups i. e. rejection of the null hypothesis

P values Probability As p 1 then it is more likely that something is going to happen As p 0 then it is less likely that something is going to happen We normally use the p value to measure how likely a result is due to chance p value approaches 0 tells us that it is less likely that something is going to happen due to chance

If we reject the null hypothesis saying that it is statistically significant at the 5% level I. e. p < 0. 05 then…. . We are saying that the rejection of the null hypothesis carries a 5% risk of being incorrect.

Before you run off to do a systematic review Sackett says: For situations encountered every day it is key that the practitioner is up to date at all times so time spent on steps 1 (question forming), 2 (searching) and 3 (appraising) will be well spent. For situations that are encountered less frequently it may be appropriate to use time to look for completed critically appraised evidence in the area of interest so time will be spent on steps 1 and 2 rather than 3. Emphasis will be on how that evidence has been derived and how it applies to practice. For situations that are outwith normal practice, where recommendations / guidance are applied ‘blindly’. Sackett describes this as the replicating mode emphasising the problems with this style of practice when the evidence base, upon which such guidance has been generated, is unknown.

Coping with the overload: three possible things you might try A. Read an evidence-based abstraction journal (and cancel other journals) B. Keep a logbook of your own clinical questions C. Run a case-discussion journal club within your practice /with peers

Take Home EBP needs the basic skills and use of those skills Needs to be part of your everyday practice Needs to be used in the context of your individual patient Is one tool to use rather than an ends to itself