Statistics and Pharmacology Flashcards
Different data type. Give example of each:
- Categorical
- Nominal
- Ordinal
- Integer
- Discrete
- Continuous
- Ranked
- Categorical: Gender, Race, late preterm, early term etc.
Nominal = Categorical - Ordinal: ROP stages, APGAR scoring,
- Integer: count. number of accidents, number of hospital re-admissions
Discrete = Integer = Count - Continuous: weight, gestational age.
- Ranked: assigned a number rank. ranked list of leading causes of death. or leading causes of pediatric hospital admission.
Mean, Median and Mode
In right skewed distribution, where is the peak of the curve? where is each of the above from L to R
Mean
Median: values at which 50% of values are greater and 50% are
smaller.
Mode: most common value
Peak of the curve to Left
Mode (peak of the bell curve)
Median
Mean
Normal bell curve
Give percentage of the following:
within 1 SD
within 2 SD
within 3 SD
68 % (68.2%)
95 % (95.4%)
99.8%
Box Plot
Outlier:
Far Outlier:
Median
Qu
QL
Inner Fence
Outer Fence
outlier: between inner and outer fence
far outlier: outside outer fence
median: 50% ile
Qu: 75% ile
QL: 25% ile
Inner Fence: QL + 1.5x IQR = 95%
Outer Fence: QL + 3x IQR = 99%
What is Null Hypothesis (Ho)?
There’s no difference between the two groups.
What is Type 1 Error?
What is Type 1 Error also called?
Reject null hypothesis when it is true
False positive
Alpha Error
What is Type 2 error?
What is Type 2 error also called?
Accept the null hypothesis when it is false
False negative
Beta error
what is power ？
how to calculate power?
Probability of rejecting the null hypothesis when null hypothesis is false
(probability of detecting a difference if it exists)
1- Beta error
Calculate sensitivity/specificity/PPV/NPV and draw the 2x 2
Populations of 1000 babies. 100 babies have a disease
180 babies tested positive, among them 100 do not have the disease
What is the sensitivity, specificity, PPV, NPV.
Top: Disease : +, -
Left: Test: +, -
Sensitivity:
80 / 100
(portion of ppl who has the disease and tests positive)
–> needed for screening test.
Specificity:
800 / 900
(portion of ppl who does NOT have the disease and test negative)
PPV:
80 / 180
probability of having the disease when test is positive.
NPV:
800 / 820
probability of not having the disease when test is negative.
what are parametric tests for continuous data? (how to compare continuous data?)
2 groups:
- unpaired t-test
- paired t-test (compare two groups before and after)
3 or more groups:
- analysis of variance (ANOVA)
- multiple regression
What are parametric Tests for categorical data?
(how to compare categorical data?)
2 groups:
- chi-square
- McNemar’s (two paired groups)
3 or more groups:
- chi-square
- logistic regression
how is p-value related to type I error
p-value IS type I (alpha) error.
false positive rate
The probability of observing a difference by chance alone
Confidence interval:
Give some examples of how to explain confidence interval.
Based on the sample’s standard deviation and its size (i.e. given its standard error), we are 95% confident that the limits cover the true value for the population mean
OR, if we draw samples of the same size 100 times, 95 of those will include the true population mean
OR, ‘the results are accurate to within +/-2SD 950 times out of 1000’
(if cannot figure out, it’s fine)
case reports, case series and cross-sectional study
you just look at cases, exposure and outcome
case-control study
cases + controls (chosen by analyst)
you know the outcome,
you look at exposures
cohort study
cases + controls (chosen by analyst)
some has exposures and some no exposures.
You look at the outcome.
Relative Risk Reduction
what is it? how to calculate it?
what about Relative Risk?
“Risk of adverse outcome in the experimental group is reduced by this proportion relative to controls”
2 x 2
Top: Disease (+, -)
Left: exposure/intervention (+, -)
A, B
C, D
Control event rate: C/ (C+D)
Experimental event rate: A / (A+B)
Relative Risk or releative risk ratio. = experimental event rate / control event rate.
RRR= 1- Relative Risk
(measures of effect)
Odds Ratio
what is it? how to calculate it?
“Odds of adverse outcome in the experimental group is reduced by this proportion relative to controls”
Intervention Odds = A/B
Control Odds = C/D
Odds Ratio = Intervention Odds / Control Odds = (A/B) / (C/D)
OR will always overestimate the effect size compared to RR.
(measures of effect)
Absolute Risk Reduction
what is it? how to calculate it?
“Risk of adverse outcome in the experimental group is reduced by this absolute percentage.”
Absolute Risk Reduction = control event rate - experimental event rate
(ARR = CER - EER)
[ a / (a+b)] - [ c/ (c+d)]
ARR is aka attributable risk (AR) or risk difference (RD).
(measures of effect)
Numbers Needed to Treat
what is it? how to calculate it?
“Need to treat this many patients to avoid one adverse outcome”
NNT = 1/ Absolute Risk Reduction
(measures of effect)
Run Chart
what’s x-axis and what’s y-axis
what are 4 rules
what is the “run”
Y-axis: measure of interest
X-axis: unit of time
median line.
“run” = one or more consecutive data points on one side of the median
4 rules:
1. trend (5 or more)
2. shift (6 or more)
3. too many or too few runs (median data line crosses only once)
4. an astronomical data point.
Control chart elements:
x and y axis
x-axis: period
y-axis: precent
central line, upper control limit, lower control limit
SMART Aim
what does each letter represent
Specific
Measurable
Achievable
Relevant
Time Limited
Three component in QI
process
outcome
balancing measures
4 moral principles:
(Belmont Report)
Respect for persons > consent
Beneficence > risk & benefit
Justice > population
nonmaleficience > do no harm
Fetal death at what week is considered miscarriage?
Fetal death at what week is considered:
“NCHS Still Birth”
“WHO still Birth”
Early Fetal Death
Late Fetal Death
THIS WAS ON BOARD
< 20 weeks –> miscarriage
“NCHS Still Birth” > 20 weeks
“WHO still Birth” > 28 week
Early Fetal Death: 20-27 weeks
Late Fetal Death > 28 weeks
Definition of Infant’s Death:
What are 2 subcategories of infant death?
Among them, one has 2 sub-sub categories.
A live birth that results in death within the first year (<365 days) is defined as an infant death.
0 to < 28 days = Neonatal death
* 0 to <7 days = Early neonatal death
* 7 to <28 days = Late neonatal death
28 to 364 days = Post-neonatal death
3 definition of perinatal death
Definition I
fetal death ≥ 28 weeks through infant death < 7 days
Definition II
fetal death ≥ 20 weeks through infant death < 28 days
Definition III
fetal death ≥ 20 weeks through infant death < 7 days
Fetal Mortality Rate
Fetal mortality rate = fetal death in 1 year / (live births + fetal death) x 1000
Fetal death: fetal death >/= 20 weeks (NCHS guideline)
Neonatal Mortality Rate
(Death from birth to <28 days / Live births) x 1000
Perinatal Mortality Rate
(Fetal death >/= 28 weeks + neonatal death < 7 days) / (Live birth + fetal death > =28 weeks) X 1000
Infant Mortality Rate
(Death from Birth to < 365 days / Live births) x 1000
postnatal mortality rate
Post-neonatal death (28 day to 364 days) / live births x 1000
For gentamicin, trough is used to assess/ monitor what?
Toxicity risk.
For vancomycin, trough is used to assess what?
Therapeutic effectiveness. (Vancomycin has time-dependent killing )
Most common mechanism of trans-placental drug transfer
Simple diffusion
How does prevalence change PPV and NPV?
What about sensitivity and specificity?
Increase Prevalence increase PPV and decrease NPV.
No change in sensitivity and specificity.
P450 inducer (p450) cytochrome system inhibitor)
1 D, 2P, 1 R
Dexamethasone
Phenobarbital
Phenytoin
Rifampin
P450 inhibitor
Ranitidine
Omeprazole
Methadone
Indomethacin
Fluconazole
Erythromycin
Cimetidine
Chloramphenicol
0 order kinetics.
What does the half life depend on?
what does time vs. log of drug concentration curve looks like?
vs. 1st order kinetics.
- half life depends on the dose
- curved
(vs. 1 order kinetic is straight)
1st order kinetics: certain percentage, proportional to serum drug concentration. not depend on dose
Phase I
Phase II
Phase III
Phase IV
of clinical trial
I: small, safety and dose, side effect.
II: slightly larger, safety, effectiveness
III: large cohort, RCT, treatment vs. no treatment.
IV: post marketing,
one-compartment model of drug distribution and equilibration
two- compartment model of drug distribution
One compartment: time and log drug concentration straight line.
Gentamicin
Two compartment: two straight lines: distribution phase and elimination phase
Vancomycin
if you calculate NNT as decimal (like 2.1), what is NNT?
round up,
NNT = 3
How to increase study power?
- increase significance criteria aka p-value (statistical significance value)
increase power:
reduce risk of Type II error.
increase risk of Type I error (p-value). - increase magnitude of effect
magnitude of effect: the size of expected observed change
increase magnitude of effect would reduce the risk of type II error (false-negative) - increase the sample size
- decrease standard deviation (decrease heterogeneity)
Power: probability of rejecting the null hypothesis when it is indeed false.
if Type I error decreases, what happens to the required power
power needs to increase.
greater amount of power is needed to detect smaller effect size.
peak concentration of drug are dependent on what?
trough concentration of drug are dependent on what?
peak concentration of drug are dependent on infusion rate and dosage
trough concentration of drug is dependent on interval of drug administration
who has higher volume of distribution?
Premie or full term baby?
what does this mean in term of clearance for preterm infant
Premie has HIGHER volume of distribution
Therefore, SLOWER Clearance
Volume of distribution calculation?
Vd (L/kg)
Vd = total amount of drug in body (mg) / plasma concentration of drug (mg/L) x weight (kg)
Phase I reaction in drug metabolism
Cytochrome P450 metabolism
oxidation, reduction, hydrolsysis, demthylation
Most common Phase II reaction in drug metabolism
Glucuronidation
(slow in neonate, important for morphine and acetaminophen metabolism)
Dose and Steady state calculation
Dose 1 / [SS] 1 = Dose 2 / [SS] 2
5 half life = steady state
what drugs have 0 order kinetics
diazepam, ethanol, aspirin, chloramphenicol
Loading dose calculation
Loading Dose = Vd x Cp (mg/L) / (S x F)
Cp: plasma concentration
S: fraction of drug which is active
F: fraction of drug which is bioavailable
S and F can be 1.
Three formulas for:
Clearance of drug (2 formulas)
Elimination rate constant (Kel)
Clearance of Drug (L/hr-kg) = Elimination rate constant (Kel) x Volume of distribution
Clearance of Drug = (Dose/Interval) / Average steady state concentration
Kel = Clearance of Drug / Volume of distribution
steady state formula
Steady State = Infusion Rate / Clearance of Drug = Infusion Rate / (Elimination rate constant x Vd)
5 half lives to reach steady state
Elimination half-life aka Half-life calculation
Half life = 0.693 / Kel
= 0.693 / (Clearance of drug / Vd)
= (0.693 x Vd) / Clearance of Drug
ACE Inhibitor - adverse effect on fetus?
Oligohydramnios
Renal failure
Lung hypoplasia
Skull ossification defects
Beta blockers - adverse effect on fetus?
Fetal bradycardia
Hypoglycemia
Possibly fetal growth restriction (not macrosomia)
Carbamazepine - adverse effect on fetus?
Neural tube defects
Craniofacial defects (cleft palate)
Growth restriction
Hemorrhagic disease of the newborn Developmental delay
Methimazole - adverse effect on fetus?
Choanal atresia
Esophageal atresia
Hypothyroidism, goiter
Cutis aplasia
Tetracyclines (Doxycycline, Minocycline, and Tetracycline)
Slowed bone growth
Enamel hypoplasia
Permanent yellowing of the teeth
Drug administration modes from greatest to least effect on maternal blood levels
IV > Paracervical > IM > Epidural > Spinal
Drugs that displace bilirubin from albumin
Sulfonamides
Ceftriaxone
Chloral hydrate
Ibuprofen
Which medications interfere with vitamin K oxidation?
Phenytoin
Phenobarbital
Isoniazid
Rifampin
Warfarin
Cephalosporins
