Evidence Based Medicine

alternative hypothesis

"Something happened," or "there was an effect" or "there was a difference." Often stated as Ha or H1. Refers to the population (not the sample)

Null Hypothesis (H0)

"nothing happened" or "no effect" or "no difference." Often stated as Ho Makes a statement about the population (not the sample)

What is the probability that a randomly selected individual has extreme pain?

# with extreme pain = 14; # in the sample = 85; marginal probability = 14/85 = 0.1647

Conditional Probability: Given that an individual lives in North East, what is the probability that the individual suffers from extreme pain?

# with extreme pain and in North east = 3# living in North east = 11Conditional Probability = 3/1

What is the probability that a randomly selected individual living in the northeast reports extreme pain?

# with extreme pain and lives in North east = 3; # in the sample = 85; Joint probability = 3/85 = 0.0353

What are the four scales of measurement

1. Nominal 2. Ordinal 3. Interval 4. Ratio

How to calculate standard deviation

1. calculate mean using x values from data set 2. subtract mean from each data point to get x' 3. calculage difference between x and x' for each data point 4. square the difference for each data point 5. sum the square values of the differences 6. divide sum by n-1 7. take sq. root of previous step

sample size

A sample is a part of the population chosen for a survey or intervention, or experiment. Ideally, the investigator desires a small sample size because the use of a large sample size wastes money, time, and effort, and it may be unethical when it comes to research on safety. The use of too few subjects may lack power and miss a scientifically important response to the treatment. This also wastes resources and could have serious consequences, particularly in safety assessment.

one-tailed test

A statistical test in which the critical area of a distribution is one-sided so that it is either greater than (>) or less than (<) a certain value. When using a one-tailed test, you are testing for the possibility of the relationship in one direction and completely disregarding the possibility of a relationship in the other direction. Example 1: Null hypothesis: mean = XXXX; Alternative hypothesis: mean > XXXX; test statistic is in the top 5% of its probability distribution using a significance level of 0.05 Example 2: Null hypothesis: mean = XXXX; Alternative hypothesis: mean < XXXX; test statistic is in the bottom 5% of its probability distribution using a significance level of 0.05

Which of the following statements is false?

A statistically significant difference in survival is always clinically significant.

You are a researcher studying medication adherence in adults living in your local community. Medication adherence is measured as a quantitative variable thanks to a validated survey instrument you are using. Your patients vary greatly in age, ranging from 23 to 91, and you wonder if there is a relationship between medication adherence and age. What statistical test is most appropriate to test this research question?

Correlation; Because both variables are quantitative, and we just want to see if there is an association between them, we can use a correlation test.

Example of strength of associations

According to Cohen (1988, 1992) d= 0.2 is small effect size d = .5 is medium effect size and d = .8 is large effect size. d= mean 1-mean 2/standard deviation

After making a scatterplot, you believe medication adherence and age have a linear relationship. You run a correlation assuming a significance level of 0.05 and find that your p-value = 0.04, and r=-0.75. What conclusions can you draw about the correlation between age and medication adherence? Check all that apply.

Age and medication adherence are negatively correlated The correlation between age and medication adherence is statistically significant The correlation between age and medication adherence is strong

What is the absolute risk reduction (ARR) for MI in the Aspirin group compared to placebo group?

Approximately 1%

Which of the following statistical method help understand the association between nominal variables? (choose the best answer)

Both Odds ratio and relative risk

You are involved in a clinical trial studying a new prescription pain medication. You notice many of the patients reporting stomach pains as a side effect are women, and would like to use the data you have collected so far to see if there is an association between reported stomach pains and se

Chi-square test for independence (This test looks for an association between two categorical variables.)

Which of the following direct to consumer advertising are suggested remedies (select all that apply)? Correct! Include Quantitative Information Correct! Improve Patient Comprehension Correct! Include Drug Cost Information Include Patient Compensation

Correct! Include Quantitative Information Correct! Improve Patient Comprehension Correct! Include Drug Cost Information Include Patient Compensation

The Physicians' Health Study is a classic study undertaken to learn whether aspirin in low doses (325 mg every other day) reduces mortality from cardiovascular disease. The participants in this clinical trial were 22,071 healthy male physicians who were randomly assigned to receive aspirin or placebo and were evaluated over an average period of 60 months. A total of 139 cases of MI were observed out of 11,037 subjects in the Aspirin group. A total of 239 cases of MI were observed out of 11,034 subjects in the Placebo group. What is the relative risk (RR) of Aspirin on MI compared to placebo?

0.58

choosing one tailed test

Example: The consequences of choosing a one-tailed test Ho: A new drug developed for pain relief is as effective as an existing drug (difference in pain relief when comparing the two drugs = 0) Ha: The new drug developed for pain relief is an improvement over an existing drug (difference in pain-relief when comparing the two drugs > 0 - i.e., the new drug provides better pain relief. A one-tailed test may optimize the ability to detect the improvement. However, by failing to test the possibility that the new drug is less effective than the existing drug, the study findings may have caused harm. When is choosing a one-tailed test NOT appropriate? For the sole purpose of attaining significance Decide on a one-tailed test after running a two-tailed test (because the two-tailed test did not attain significance)

Type 2 error (false negative)

Failing to reject H0 when H0 is really false, denoted by ("beta"). = P(Type II error)

Authors who have previously worked with pharmaceutical industry should not be included in guideline development.

False

Safety is a topic that will always be reported in the journal club template.

False

True or False. DTCA is defined as good education for patients that don't influence providers' prescribing practices.

False

From your previously calculated line of best fit for the relationship between age and medication adherence: Y = 8 - 0.5*X What conclusions can you make? Check all that apply

For every one-year increase in age, there is a decrease of 0.5 in the value of medication adherence As age increases, medication adherence decreases

p value is greater than alpha

If the p-value is larger than 0.05, we cannot conclude that a significant difference exists. fail to reject the null

p value is less than .05

If the p-value is less than 0.05, we reject the null hypothesis that there's no difference between the means and conclude that a significant difference does exist.

Techniques that allow us to study samples and then make generalizations about the populations from which the samples were selected.

Inferential Statistics

Average temperature in January at the place where you live

Interval

Effect size

It comes from subject-area knowledge, pilot studies, published findings from similar studies, educated guess based on informal observations, and knowledge of the field. A quantitative, standardized, scale-free measure of the relative size of the effect of an intervention Can either measure the size of associations or the size of differences between means or proportions Interpretation of standardized effect sizes can be problematic when a sample does not come from a Normal distribution. The 2 major effect size families include the d family and the r family. The "d family" effect sizes: a measure of the amount of a difference between groups. The most prominent examples of the d family include mean or median differences, percentage differences, risk ratios, odds ratios, and risk differences. The "r family" effect sizes: a measure of the strength of a relationship between variables. Most of these r family measures are variations on the Pearson correlation coefficient. Example: Differences in means: The effect size is computed as the standardized mean difference between the two groups.

Which organization has been responsible for publishing guidelines related to hypertension management?

Joint national committee

simple logistic regression

Logistic regression is simply a transformation of the linear regression equation that allows us to interpret the predicted probabilities in a situation where linear regression won't work well. However, interpretability is not the same. The linear regression equation y = A + Bx relies on several assumptions that are not met when the outcome is binary. The logistic regression equation, then, accounts for this by transforming y into a "log-odds" using the probability of success. ln[p/(1-p)] = A + BX a still represents the constant, and B still represents a version of the slope, however, because y is now log-odds, you cannot simply say that "a one-unit change in x results in a B-unit change in y." Instead, you would conclude that "a one-unit change in x results in a B-unit change in the log-odds of y" or, alternatively, you could exponentiate the slope exp(B) and obtain the odds ratio, which will be covered in greater detail next week.One categorical response variable and one continuous or categorical explanatory variable

Which of the following represents a trail clinical endpoint? Sudden Death MI A1c All are correct

MI and sudden death

Which of the following DOES NOT contribute to publication bias? A. Medline (PubMed) is less likely to include studies with negative results in their database B. Journal editors and peer reviewers may be hesitant to allow negative findings to be published C. Authors may be hesitant to publish negative results D. Industry sponsors may be reluctant to publish negative findings

Medline (PubMed) is less likely to include studies with negative results in their database

A phase 3 clinical trial comparing a new drug and placebo for prophylactic efficacy (preventing infections) of HIV-1 infections. Results indicated that the relative risk estimate was 0.71 with 95% confidence intervals (CI) 0.58-1.5. What would be the correct conclusion from the study?

New drug was not effective in preventing HIV-1 infections as CI include 1

A variable that has names such as Alex, Janet, Dennis, Usha is called --- variable.

Nominal

Names: Alex, Constance, Dennis, Usha

Nominal

In preparation for your journal club presentation, you need to calculate which of the following if not provided by the authors?

Number needed to treat

Which of the following is an appropriate risk estimate for a case-control study?

Odds ratio

Power Analysis

One needs to combine statistical analysis, subject-area knowledge, and availability of units to derive the optimal sample size for the study. There are four inter-related concepts in conducting a power analysis. If three of these values are fixed, the fourth is completely determined. By increasing one, you can decrease another. Alpha Power Effect size Sample size

The order in which runners cross the finish line is an example of

Ordinal

Regarding the PICO model for guideline develop, the acronym represents which of the following? A. Patients, Interventions, Comparison, Outcome B. Patient, Influence, Collaboration, Outcomes C. Patents, Interventions, Colleagues, Outcomes D. Patients, Interventions, Comparison, Outstanding

Patients, Interventions, Comparison, Outcome

Which definition best describes meta-analysis?

Quantitatively combines the results of studies that are the result of a systematic literature review. Capable of performing a statistical analysis of the results.

Your body Weight (in pounds)

Ratio

Type 1 error (false positive)

Rejecting H0 when H0 is really true, denoted by ("alpha") and commonly set at .05. = P(Type I error)

Which of the following is a nominal variable?

Sex

After finding that age and medication adherence are negatively correlated, you decide to try and determine if there is more than just a correlation, and would like to see how changes in age could affect changes in medication adherence. What statistical test would be most appropriate to use in this situation?

Simple linear regression

significance level (alpha)

The acceptable level of error selected by the researcher, usually set at 0.05. The level of error refers to the probability of rejecting the null hypothesis when it is actually true for the population.

interval scale

The interval scale is an ordered scale. Numbers are assigned along a continuum. The space between each interval on a scale is equivalent. For example, the difference between 90 degrees Fahrenheit and 100 degrees Fahrenheit is the same as 110 degrees Fahrenheit and 120 degrees Fahrenheit. Interval scales have no true zero points and can represent values below zero. For example, you can measure temperature below 0 degrees Celsius, such as -10 degrees. Any measurement of interval scale can be ranked, counted, subtracted, or added, and equal intervals separate each number on the scale. In the absence of absolute zero, we cannot multiply or divide interval values with each other. These measurements don't provide any sense of ratio between one another.

With interval scale we know: The order and exact difference between values The order only Exact difference between values only

The order and exact difference between values

Alpa

The probability of finding significance where there is none; false positive; the probability of type I error; usually set to 0.05

Which of the following is FALSE regarding systematic reviews?

The search strategy and databases that were used may not be provided

After running your test with a significance value (alpha) of 0.05, you obtain a large test-statistic and find that your p-value=0.07. What conclusions can you draw from these results? Check all that apply.

There is not a significant association between reported stomach pains and sex in this sample We fail to reject the null hypothesis

Ratio scale

This is the highest order of measurement. Ratio scales can be meaningfully added, subtracted, multiplied, divided (ratios). The scale tells us about the order The scale tells us the exact value between units, AND The scale has an absolute zero

Help-seeking ads describe a disease or condition but do not recommend or suggest specific drugs.

True

p-value

Used in hypothesis testing Used to provide evidence to reject or fail-to-reject a null hypothesis Smaller the p-value the stronger the evidence against null hypothesis the p-value is the area in the tail of a distribution Typically, the p-value is compared to the selected alpha level (<.05, <.01 etc); alpha level is determined by the researcher; if the researcher wants to be 98% percent confident in the findings, the alpha level would be .02 ( 1 - 0.98)

One-way ANOVA

Used to compare the means of more than two independent groups (one-way between groups analysis of variance) More efficient than using multiple t-tests (one test vs many) and reduces likelihood of type-1 error (false positive) compared to multiple t-tests Assumes: Samples are independent Response variable is normally distributed or all sample sizes are at least 30 Population variances are equal across responses for the group levels H0: All population means are equal Ha: Not all population means are equal Test-statistic: ANOVA uses the F-statistic, which is a ratio of variability between and within groupsEssentially measures how much individuals in different groups vary from each other, over how much do individuals within groups vary from each other Drawing conclusions: if p <= alpha, then you can reject the null hypothesis and conclude that there is at least one significant difference in population means of the groups being tested

Nominal Scale

Used to label variables in distinct groups/classifications. It is also known as categorical variable scale. The categories have no meaningful order. The values serve only as labels. It is also known as categorical variable scale. The categories have no meaningful order. The values serve only as labels.

simple linear regression

Uses one explanatory/independent variable (x) to predict one response/dependent variable (y) There is also multiple linear regression that uses more than one explanatory variable (but will not be covered here) Can only be used if x and y are both quantitative variables x is always the explanatory/independent variable, y is always the response/dependent variable (unlike correlation, where it doesn't matter) How it works Linear regression uses data from a sample to construct a "line of best fit." The equation for this line is generally written as:

two-tailed test

When using a two-tailed test, regardless of the direction of the relationship you hypothesize, you are testing for the possibility of the relationship in both directions. Example 1: Null hypothesis: mean = XXXX; Alternative hypothesis: mean is not equal to XXXX; This means that .025 is in each tail of the distribution of your test statistic.

Medication adherence is measured quantitatively and ranges from 0-10. You run a simple linear regression and calculate the line of best fit for the relationship between age and medication adherence: Y = 8 - 0.5*X

Y(medication adherence) 8 (constant) -0.5 (slope) X(age)

ordinal scale

a scale of measurement in which the measurement categories form a rank order along a continuum Categories are ordered in some fashion. The order of the values is important. But the distance (difference) between categories is not really known.

Rating of movies

ordinal

Your experience of shopping for fresh foods in store near you (very unsatisfied, unsatisfied, neutral, satisfied, very satisfied)

ordinal

statical significance

represents whether or not the differences observed between two groups (example: treatment versus control group) are real or due to chance. P-values and confidence intervals are the most commonly used measures of statistical significance. Statistical significance alone is not sufficient in making decisions to adopt a new treatment.

clinical/practical significance

the emergence of "large" and "big-data" has brought a renewed focus on the concept of clinical/practical significance. "Clinical significance refers to the practical or applied value or importance of the effect of an intervention-that is, whether the intervention makes a real (e.g., genuine, palpable, practical, noticeable) difference in everyday life to the clients or to others with whom the clients interact." - Kazdin 1999 Clinical significance relies on "size and scope of an effect." It's a critical tool for decision-makers in pharmaceutical, psychological, and medical research.

Joint Probability

the probability of two events occurring together

statical power

the probability that a test of significance will pick up on an effect that is present the probability of correctly rejecting the null hypothesis -- the probability of rejecting the null hypothesis when it is false Type I and type II errors deal with making incorrect decisions; power is about making the correct decision Beta (β) is the probability of making a Type II error and has an inverse relationship with statistical power (1 - β) It is a conditional probability -- the null hypothesis makes a statement about parameter values, but the power of the test is conditional upon the values of those parameters Power ranges from zero to one (power of .5, power of .80, power of .90) Higher the power level, the lower the probability of a Type II error Power of 0.90 implies:there is a .90 chance of rejecting the null hypothesis0.10 chance of failing to rejectFg the null hypothesis0.10 is the risk of committing a Type II error (β) Power of 0.50 implies:there is a 0.50 chance of rejecting the null hypothesis0.50 chance of failing to reject the null hypothesis.50 is the risk of committing a Type II error (β)

Conditional Probability

the probability that one event happens given that another event is already known to have happened

Marginal Probability

the values in the margins of a joint probability table that provide the probabilities of each event separately

Those who oppose DTCA advocates that DTCA: A. DTCA leads to inappropriate prescribing B. Includes too much regulation and regulation should be reduced C. Encourages drug under-utilization D. Decreases costs

DTCA leads to inappropriate prescribing

There are two aspects to conducting power analysis:

Determine the minimum sample size for a specified power Compute power, given specific sample size.

why is statical power important

Determine the sample size needed before beginning a particular studyAlmost all research studies are based on samples;An appropriate sample size is crucial to any research investigation.Statistical power analysis guides the tradeoffs of a large sample size that has a high probability of detecting minimal effects to be useful and a small sample size that has a low probability of detecting an important effect. Occasionally used after the study is conducted to determine if the reason an effect was not significant was insufficient power. It is an essential part of any research plan and is required in research grant applications.

Which of the following is TRUE regarding narrative reviews?

Does not follow strict systematic methods to locate and synthesize articles