Professional Documents
Culture Documents
23-BIOSTATICS, FORMULAE
categorical/qualitative data-percentage%
freq of occurrence(comparision of magnitude)
bar diagram
pie chart/sector diagram
pictogram
map diagram/spot map
quanTitative data-meanSD(unskew), median(range)(skew)
distribution(not related to time)
hisTogram
freq polygon
freq curve
line diag(time trend)
scatter plot(study relation b/n 2variable)
measure of central tendency
outlier, skewed, extreme
most affected-mean
least affect-mode
most preferred-median
geometric mean=(n)(x1x2xn)
harmonic mean=1/(1/x1+1/x2+1/xn)
meas of dispersion/deviation/variation
range=maxmin
mean deviation=|meanx|/n
SD=[(|meanx|)/n]
SD(<30)=[(|meanx|)/(n1)]
variance=SD
CV=(SD/mean)100
SE=SD/n
precision=1/SE=n/SD
for bimodal data
mode=3median2mean
z score/relative deviate/critical ratio
z=|meanx|/SD
(quant)SE=SD/n
(qualit)SE=(PQ/n)
(quant)95%CI=mean2.5SE
(qualit)95%CI=proportion2.5SE
(quant)sample size=ZSE/L
(qualit)sample size=ZPQ/L
Z=4, P=prevalence, Q=1P, L=100-CI
probability
p(event)=n(experiment)/n(total event)
odds=n(event occur)/n(event not occur)
Poisson distribution-discrete probability distribution for random event
+ve/rt skew
mean>median>mode
ve/lt skew
mean<median<mode
(Hint-arrange in alphabetical order and look at the direction where the (<,>) is
pointing)
Pearson skewness coefcient=(meanmode)/SD
std normal curve
area=1
mean=mode=median=0
SD=1
mean1SD=68.27%
mean2SD=95.45%
mean3SD=99.73%
test for pRecision(Reliability, Reproducibility, Repeatability)
Range chart
R chart
CRonbach
test for accuracy
mean chart
LJ(Levy Jenning) chart
Shewart control chart
null hypothesis(
reject )
rejected-p value<0.05-p value signicant
accepted-p value>0.05-p value insignicant
p value=prob of declaring diff when actually not=prob rejecting null hypothesis when
true
ecological
yield=new case d/t screening
screen time=time(1st possible Dx nal critical Dx)
lead time=time(1st possibl Dx usual time Dx)
serial interval=time(prim case sec case)
generation time=time(receipt of inf by host max infectivity)
T interval=no. of d bed of hosp remain vacant
incubat periodgeneration time
RR/risk ratio=incid(exposed)/incid(non exposed)
AR=[incid(exposed)incid(non exposed)]/incid(expos)
popul AR=[incid(population)incid(not exposed)]/incid(population)
type of bias-technique
selection-randomisation
surveillance/detection
Neyman survival/incidence prevalence
referral/volunteer
response
Berkesonian(hosp admission)
recall(memory)
bias d/t confounding-matching, multivariate analysis
information-single blinding
interviewer/observer-double blinding
statician-triple blinding
sensitivity=(true+ve/all ds)100
specicity=(trueve/all not ds)100
+ve predictive value=(true+ve/all+ve)100
ve predictive value=(trueve/allve)100
diagnostic test-predictive value
false+ve%=(FP/total not ds)100=1sPecicity
falseve%=(FN/total ds)100=1seNsitivity
efciency(accuracy)=all true/all pt
likelihood ratio for +ve test=sensitivity/(1specicity)
likelihood ratio ve test=(1sensitivity)/specicity
pretest probability=prevalence+clinical assessm
posttest probability=pretest probabilitylikelihood ratio
sensitivity-TP-PPV
specicity-TN-NPV
sensitivity,specicity-criterion validity
multiple Dx test
series-sensitivity, specicity, PPV
parallel-sensitivity, specicity, NPV
best test to compare new&old test-Bland&Altman analysis
sampling
random/probability/nonpurposive-chance of being selected-same&known
simple
systematic
stratied(heterogenous)
multistage
Cluster(homogenous-Cost effective)
nonrandom/nonprobability/purposive
convenient
quota/targetted
snowball/network
judgement
PQLI=IMR+life expectancy at age 1y+literacy rate, range=0-100, India=65(rank-63)
measure social, economic, politic policies
does not measure economic growth
HDI=GNI per capita+ life expectancy at birth+ (mean y of school, expected y of
school), range0-1, India=0.545(rank-136)
top-Norway, Australia, USA
bottom-Congo, Nigeria
human poverty index1=long&healthy life(probability at birth not surviving 40y)+
knowledge(adult literacy rate)+ std of living(%of populat using improved H2O source+
%child underwt for age), India=31.3%
DALY(Disability Adjusted Life Y)=y of life lost+y lost to disability
burden of ds in populat
effectiveness of intervention
DFLE(Disability Free Life Expectancy)/active life expectancy=no. of y expected free of
disability if current pattern of mortality& disability continue to apply
HALE(Health Adjusted Life Expectancy)=life expectancytime spent in poor health
QALY(Quality Adjusted Life Y)=no. of y added to life by intervention
type of data(NOIR)
midtriceps(best)+biceps+subscapular+suprailiac 50mm(fem),40mm(mal)
midtriceps18mm(mal), 32mm(fem)
waist circumf(cm)
World102(mal), 88(fem)
India90(mal), 80(fem)
waist/hip ratio>1.0(mal), >0.85(fem)
waist/ht ratio(best CVS ds risk)0.5
contracept effectiveness
Pearl index
potency of contraception=(total accident preg/no. of pt observedmth of use)1200
life table analysis-best
prot quality assessm
digestibilty coefcient=(N2 absorbed/N2 intake)100
biological value=(N2 retained/N2 absorbed)100
net protein utilization=(N2 retained/N2 intake)100=(biological valuedigestibility
coefcient)/100
protein efciency ratio=wt gain(g)/prot intake(g)
AA(chemical) score=[{(no. of AA/g prot)}/{(no. of same AA/g egg prot)}]100
prot quantity assessm
prot energy ratio=(energy fr prot/total energy fr diet)100
1g prot=6.25g N2
norm NPU=50-80
relative humidity=H2O vapour content/H2O vapour capacity
Box Whisker plot
quartile
forest plot
represent result in systematic review& metaanalysis(best study design)
funnel plot
assess publication bias
Kaplan Meier curve/product limit estimation
survival fn fr lifetime data-#pt living for certain time after Rx
X-time, Y-%of survival
cox-proportion hazard model to control confounding factor in survival analysis
Levy Jenning chart
accuracy, quality monitoring