Medicine in Society Flashcards Preview

DIT Phase II Chapters > Medicine in Society > Flashcards

Flashcards in Medicine in Society Deck (38):
1

Case-control study

  • Observational and retrospective
  • Compares group of people with disease to group without
  • ID's risk factors
  • Asks, "what happened"
  • Odds ratio

2

Cohort study

  • Observational
  • Groups according to risk factors and sees what happens to them
  • Looks to see if exposure increases likelihood of disease
  • Asks, "What will happen?" Rate of exposed/rate of unexposed
  • Relative risk (RR)

3

Cross-sectional study

  • Observational
  • Gives snapshot of disease at one point in time
  • Disease prevalence

4

Twin concordance study

  • Compares the frequency with which both monozygotic twins or both dizygotic twins develop a disease
  • Measures heritability

5

Adoption study

  • Compares siblings raised by biologic vs. adoptive parents
  • Measures heritability and influence of environmental factors

6

Clinical trial

Phases

  • Experimental study involving humans
  • Compares therapeutic benefit of 2 or more treatments or of treatment and placebo
  • Highest quality when randomized, controlled, double blinded

 

  • Phase I - Is it safe - safety, toxicity, pharmacokinetcs
  • Phase II - Does it work - Efficacy, optimal dosing, adverse effects
  • Phase III - Does it work better - compares new treatment to current standard of care.
  • Phase IV - Are there rare adverse affects? Postmarketing surveillance trial of patients after approval.

7

Meta-analysis

 

  • The systematic process of using statistical methods to combine the results of different studies
  • systematic, organized, and structured evaluation of a problem using information, commonly in the form of statistical tables, from a number of different studies of a problem
  • Need strict inclusion criteria and selection bias may creep in

8

Diagnostic Tests

  • Sensitivity = A/(A+C)
  • Specificity = D/(D+B)
  • PPV = A/(A+B);
  • PPV - tells you the probability that a person who tests positive actually has the disease
  • NPV = D/(C+D)
  • NPV - tells you the probability that a person who tests negative is actually free of the disease
  • Sensitivity and Specificity are the vertical columns
  • PPV and NPV are the horizontal columns

A image thumb
9

Which diagnostic tests change with prevalence of disease?

Sensitivity and Specificity don't change - static
PPV and NPV change depending on prevalence of disease in society.

  • ↑PPV with ↑prevalence of disease
  • ↓ prevalence will ↑ NPV

10

Prevalence vs. Incidence

  • Point prevalence = total cases in population at a given time/ total population at a given time
  • Incidence = new cases in population over a given time period/total population at risk during that time period
  • Incidence = new incidence

11

What is prevalence approximately equal to?

Prevalence ≈ incidence x disease duration

Prealence > Incidence for chronic diseases (diabetes)

Prevalence = incidence for acute disease (e.g., common cold)

12

Odds ratio vs. Relative risk

Odds ratio for case control studies:

  • Odds of having disease in exposed group divided by odds of having disease in unexposed group
  • Approximates relative risk if prevalence of disease is not too high

Relative risk (RR) for cohort studies:

  • Probability of getting a disease in the exposed group divided by the probability of getting a disease in the unexposed group

A image thumb
13

Attributable risk

AR = incidence of disease in the exposed group - incidence of disease in the unexposed group

Example: In a population of sexually-active people, 30% have HPV infection. In a population of people who are not sexually active, only 5% have HPV infection. The attributable risk of sexual activity to HPV is 25%.

14

Absolute Risk Reduction

The reduction in risk associated with a treatment as compated to placebo

ARR = C/(C+D) - A/(A+B)

Example:

People that got the disease on the drug = 5%
People that got the disease w/o drug = 20%

Absolute risk reduction is 15%

A image thumb
15

Number needed to treat

  • NNT = 1/absolute risk reduction
  • Number of patients you would need to treat in order to save/effect one life
  • Important number to help determine if a drug should be used or is cost effective
  • Example: If out of 10,000 patients that took t-PA during a STEMI, 100 were saved by the t-PA, then the NTT is 100. In other words, you would need to treat 100 patients in order to save/effect 1 life

16

Number needed to harm

  • NNH = 1/attributable risk
  • NNH=1/AR
  • (AR = incidence of disease in exposed group - incidence of disease in unexposed group)

17

Precision vs. Accuracy

Precision is:

  • The consistency and reproducibility of a test (reliability)
  • The absence of random variation in a test
  • Reduced by random error

Accuracy is the truness of the test measurements (validity):

  • Reduced by systematic error

 

 

 

18

Ways to reduce bias?

  • Blind studies (double-blind to limit influences of participants and researchers on interpretaiton of outcome
  • Placebo responses
  • Crossover studies (each subject acts as own control to limit confounding bias)
  • Randomization to limit selection bias and confounding bias

19

The referral centers for a trial of a new anticancer drug have more patients with end stage disease than early stage, so more patients with end stage disease are referred for the trial than early stage disease.

Selection bias - nonrandom assignment to study group (e.g., Berkson's bias, loss to follow-up)

20

Studies performed on patients that have been hospitalized

  • Berkson’s bias - type of selection bias
  • The result is that two independent events become conditionally dependent (negatively dependent) given that at least one of them occurs
  • classic illustration involves a retrospective study examining a risk factor for a disease in a statistical sample from a hospital in-patient population. If a control group is also ascertained from the in-patient population, a difference in hospital admission rates for the case sample and control sample can result in a spurious association between the disease and the risk factor.

21

Parents of autism patients having a more detailed recall of events and illnesses in theirchild’s first two years of life compared to parents of healthy controls. 

  • Recall bias - knowledge of presence of disorder alters recall by subjects

22

A study performed in China may not be generalizable to the US population.

  • Sampling bias - Subjects are not representative relative to general population
  • Results not generalizable
  • Sampe does not represent population

23

Sending a survey out to people diagnosed with a fatal illness 5 years after diagnosis will preferentially sample those with a low grade disease (or few comorbidities)

  • Late-look bias -  information gathered at an inappropriate time
  • e.g. using a survey to study a fatal disease (only those patients still alive will be able to answer the survey)

24

The positive benefit of a new drug during a study simply may have been due to thefact that study participants were required to attend clinic monthly, where they received extra disease education and counseling compared with the controls.

  • Procedure bias - subjects in different groups are not treated the same
  • e.g. more attention is paid to treatment group, stimulating greater compliance

25

Are asbestos miners more likely to have cancer because they mine asbestos or because they are more likely to smoke?

  • Confounding bias - occurs with 2 closely associated factors
  • the effect of 1 factor distorts or confuses the effect of the other

26

While test PSA-xyz may detect prostate cancer before it is detected by a traditional PSA, early detection using PSA-xyz does not increase cancer survival compared to traditional PSA.

  • Lead-time bias - early detection confused with increased survival
  • Seen with imporved screening (natural history of disease is not changed, but early detection makes it seem as though survival increased.

27

An orthopedic surgeon investigator who finds statistically significant benefit ofarthroscopic surgery when compared to non-invasive therapeutic strategies. A chiropractor-led study that finds significant benefit of the effects of cervical manipulation when compared to traditional medicine strategies.

Pygmalion effect - occurs when a researcher's belief in the efficacy of a treatment changes the outcome of that treatment.

28

When studying the effects that infection control education has on physicians, the investigator notes that both the experimental and the control groups improve their hand hygiene.

Hawthorne effect - occurs when the group being studied changes its behavior owing to the knowledge of being studied

  • "Dr. Hawthorne is watching you"

29

Terms that describe statistical distribution

  • Normal = Gaussian = bell-shaed (mean = median = mode)
  • σ = standard deviation; n = sample size
  • SEM = standard error of the mean = σ/(n)^.5
  • SEM decreases as n increases

30

Positive skew

Negative skew

  • Positive skew - mean>median>mode
  • Asymmetry with tail on right
  • Mode least affected by outliers in sample

 

  • Negative skew - mean
  • Asymmetry with tail on left

31

Statistical hypotheses

  • Null (Ho) - Hypothesis of no difference
  • Alternative (H1) - Hypothesis that there is some difference

A image thumb
32

What is a p value?

What is an alpha level?

Type I error (alpha)

Type II error (beta)

 

  • P value = tells you how compatible data is with null hypothesis - probability that study results occurred by chance alone given that Null hypothesis is true 
  • Alpha level - set by investigator at which p value is judged.
  • Type I error - "false-positive error" "saw" a difference that did not exist
  • Type II error - "false negative error"  - B = Blind to a difference that did exist, B= probability of making a type II error - failing to reject the null hypothesis when it is in fact false

33

Power (1-β)

  • Probability of rejecting null hypothesis when it is in fact false
  • likelihood of finding a difference if one in fact exists. It depends on:
    • Total number of end points experienced by population
    • Difference in compliance between treatment groups
    • Size of expected effect

34

Standard deviations

  • 1 - 68%
  • 2 - 95%
  • 3 - 99.7%

35

What does the shape of a positive skew graph look like?

  • asymmetry with curve shifted to left with tail on right; mean > median > mode

36

Confidence interval

Z values

  • CI = mean ± Z(SEM) 
  • range from [mean-Z(SEM)] to [mean + Z(SEM)]
  • Range of values in which a specified probability of the mean of repeated samples would be expected to fall
  • The 95% CI (corresponding to p = 0.05) is often used
    • For 90% CI, Z = 1.645
    • For 95% CI, Z = 1.96
    • For 99% CI, Z = 2.58

37

t-test vs. ANOVA vs. χ2

  • T-test checks difference between means
  • ANOVA checks difference between the means of 3 or more variables
  • χ2 (chi-square) = compare percentages (%) or proportions

38

Correlation coefficient (r)

  • Goodness of fit, how variables relate
  • always between -1 and +1