Research tools Flashcards
What is the standard error of the mean/ how to calculate?
Measures how much discrepancy is likely in a sample’s mean compared with the population mean
Standard deviation divided by square root of sample size
How do you calculate the chance of making a type 2 error/ what is a type 2 error?
Type 2 error = beta (failing to reject the null hypothesis when it is actually false i.e. false negative rate)
Sensitivity = power.
The higher the power/ sensitivity, the lower the chance of a type 2 error
1-power/sensitivity = P(type II error)/ beta
Power/ sensitivity = 1-P(type II error)/ beta
Incorrect acceptance of a null hypothesis
What is a type I error and how is it calculated?
Type I error = false positive rate
alpha (type I error) = 1- specificity
Incorrect rejection of a true null hypothesis
How to calculate true positive rate (TPR)
TRP = sensitivity = Power = 1- beta (FNR/ type II error)
How to calculate true negative rate (TNR)
TNR = specificity = 1-alpha (FPR/ type I error)
What is number needed to treat and how is it calculated?
Measure of effectiveness of an intervention: the number of subjects needed to receive an intervention for one event to be prevented/ occur
NNT= 1/AR
AR= absolute risk
AR = risk observed group - risk control group
Risk for each group= number of events occurring in that group divided by the total population.
The lower the NNT, the more effective the intervention.
Odds ratio
Odds of an event occurring in one group divided by the odds of it occurring in the other group
OR= odds observed groups/ odds control group (a/b)/(c/d)
a- number of subjects with event occurring in observed group
b- number of subjects without event occurring in observed group
c- number of subjects with event occurring in control group
d- number of subjects without event occurring in the control group
What is an ROC curve and what is on the axes?
Receiver operating characteristic curve: illustrates the performance of a binary classifier model at varying threshold values.
X axis: False positive rate (1-specificity)
Y axis: True positive rate (sensitivity)
The larger the area under the curve, the more accurate the test
Negative predictive value calculation
TN/ (TN+FN)
TN= true negative
FN = false negative
Term for condition in which statistical difference occurs purely by chance
Type 1 error
Relative risk calculation
Odd’s ratio calculation
Relative risk is the ratio of risk in an exposed group compared to a non-exposed group
RR = probability of an event when exposed/ probability of even in control group
Exposed & disease = a
Exposed & no disease = b
Control & disease = c
Control & no disease = d
RR = [a/(a+b)]/ [c/(c+d)]
OR = [a/b]/[c/d]
Incidence of ovarian cancer in UK
22 per 100,000
Risks associated with VBAC and statistics
2-3/10,000 additional risk of birth related perinatal death
8 in 10000 infant developing hypoxic ischaemic encephalopathy
22-74 in 10,000 risk of uterine rupture
1% additional risk of either blood transfusion or endometritis
Difference in risk of baby having breathing problems in VBAC vs repeat ELCS?
VBAC reduces the risk
Rates are 2-3% with VBAC compared to 3-4% with ELCS
Parametric vs non-parametric statistical tests
Parametric assume a normal distribution of population data. For example:
- Pearson (correlation test)
- T-test
- Analysis of variance (ANOVA)
- f-test
- z-test
Non-parametric can be used for populations that aren’t normally distributed.
For example:
- Spearman (correlation test)
- Mann Whitney
- Chi-squared
- Wilcoxon Signed Rank
- Fisher Exact Probability
- Kruskal Wallis
- Friedman
Levels of evidence
Ia- Evidence from meta-analysis of RCT
Ib- Evidence from at least one RCT
IIa- Evidence from at least one well designed controlled trial (not-randomised)
IIb- Evidence from at least one well designed experimental trial
III- Evidence from case, correlation and comparative studies
IV- Evidence from a panel of experts
Risk factors vs protective factors for ovarian ca.
Risk:
- Age, obesity, FHx, HRT (oestrogen only)
Protective:
- OCP, higher parity, breast feeding, hysterectomy, tubal ligation, statins, SLE
What is the success rate of VBAC following 1x CS
How does this change if she has had 1x successful VBAC
What lowers the success rate?
72-76%
87-90%
Induced labour
No previous vaginal birth
Obesity (BMI>30)
Previous CS for dystocia
What type of study is the most appropriate to assess treatment/ intervention, diagnostic test and prognosis
Treatment/ intervention: RCT
Diagnostic tests: cross sectional study or analysis
Assessing prognosis: Cohort study
Average age of diagnosis of uterine cancer in the UK
Age 60
Most cases seen in age 60-64
Incidence rate (cases per 100,000) is highest in the 70-74 age group
WHO definition of maternal death
Death of a woman whilst pregnant or within 42 days of termination of pregnancy
Miscarriage rates in relation to age at conception
20-24, 9%
25-29, 11%
30-34, 15%
35-39, 25%
40-44, 51%
>45, 93%
Maternal mortality rate of ectopic pregnancy in UK
2 per 1000 (0.2%)
WHO definition of perinatal mortality rate
Number of stillbirths and deaths in first week of life per 1000 births
Peak incidence for ovarian ca. vs decade with most cases
80-84 age group (number per 100,000)
60-69 age group comprises the most cases
Increasing sample size decreases what type of errors
Type 2
WHO definition of maternal mortality ratio
Maternal deaths per 100,000 live births
Direct vs indirect maternal deaths
Maternal mortality rate (MMBRACE)
Direct: those resulting from obstetric complications of pregnancy e.g. VTE
Indirect: deaths resulting from pre-existing disease or disease that developed during pregnancy which was not due to direct obstetric causes, but was aggravated by physiologic effect of pregnancy
Maternal mortality rate: deaths during pregnancy or within first 42 days following end of pregnancy per 100,000 maternities (for any cause related to or aggravated by pregnancy- not including accidental or incidental causes)
ICD-10 includes late maternal deaths occurring between 6 weeks and 1 year after childbirth.
Types of data
- Categorical
- Nominal: names or categories with no order e.g. eye colour, sex, ethnic group
- Ordinal: there is ranking within the categories, but not on a scale e.g. APGAR scores
- Quantitative
- Interval: there is ranking of the numbers, but on a scale and values are equally spaced e.g. temperature
- Ratio: as for interval, but 0 means the variable is absent e.g. length
Parametric vs non-parametric statistical tests
Parametric= for normally distributed samples
EXAMPLES
- T-test: independent when comparing 2 unpaired distributions or paired when comparing 2 paired distributions
- ANOVA (analysis of variance): used when multiple distributions are compared. Also assumes spread of each distribution is the same
Non-parametric = for samples that aren’t normally distributed
- Mann Whitney U test (equivalent of independent t-test)
- Wilcoxon’s signed rank test (equivalent of paired t test)
- Kristal-Wallis one way analysis of variance (equivalent of ANOVA)
- Friedman two way ANOVA
- Chi squared (for categorical data)
How to calculate variance
The average of the squared differences from the mean
(v. long equation)
How to calculate standard deviation
square root of variance
Objectives of phase 0 to IV clinical trials
Phase 0: pharmacokinetics: oral availability and half life
Phase I: dose-ranging on healthy volunteers
Phase II: efficacy and side effects
Phase III: Assess efficacy, effectiveness and safety
Phase IV: Post marketing surveillance in public
Positive/ negative likelihood ratios that indicate a moderately useful clinical test
PLR: 5-10 (higher the value, the better)
NLR: 0.1-0.2 (lower the better)
