Week 1 - stats Flashcards
What is analytical biochemistry?
- The study of biochemical components found in a cell or other biological sample
- Uses a broad range of techniques for separation, identification, quantification and functional characterisation of biological molecules
Chromatography
- separates components of a mixture based on their chemical properties separates and identifies components of a complex mixture.
Examples–HPLC, GC, TLC
Electrophoresis
separates charged molecules based on their size and charge separates and identify proteins, DNA, and RNA
Spectroscopy
uses the interaction of light with matter to study the properties of biological moleculesused to study the structure and function of proteins, nucleic acids, and other biological molecules
WHAT IS CLINICAL BIOCHEMISTRY?
- generally concerned with analysis of bodily fluids for diagnostic and therapeutic purposes
- an applied form of biochemistry
General or routine chemistry
Commonly ordered blood chemistries (e.g., liver and kidney function tests)
Special chemistry
Elaborate techniques such as electrophoresis, and manual testing methods
Clinical endocrinology
The study of hormones, and diagnosis of endocrine
Toxicology
The study of drugs of abuse and other chemicals
Therapeutic Drug Monitoring
Measurement of therapeutic medication levels
Urinalysis
Chemical analysis of urine for a wide array of diseases
DIAGNOSIS OF A MYOCARDIAL INFARCTION
- A patient is diagnosed with myocardial infarction if 2 (probable) or 3 (definite) criteria:
→ Clinical history of ischaemic type chest pain lasting for more than 20 minutes
→ Changes in serial ECG tracings
→ Rise and fall of serum cardiac biomarkers, e.g. creatine kinase-MB or troponin
CARDIAC MARKERS
measure of damage to the cardiac muscle
Where is CREATINE KINASE (CK) found?
Present in muscle (skeletal & cardiac) and brain (not liver) Dimer, with two different subunits: Muscle (M), Brain (B)
May exist as CK-MM, CK-MB, CK-BB
WHAT IS NEAR PATIENT TESTING?
- May also be called point of care testing
- Performed near or at site of patient–result may change care pathway for patient
Near Patient Testing - Advantages
- Generally faster than traditional lab testing
- May require smaller sample
Near Patient Testing - Disadvantages
May be less accurate–clinical staff
APPLICATIONS OF NEAR PATIENT TESTING
- Diagnosis of infectious diseases
- Monitoring of chronic diseases
- Detection of drug abuse
- Guide to other problems
NEAR PATIENT TESTING–DEVICES
- Blood glucose meters
- Pregnancy Tests
- Rapid Strep Tests
- Covid Tests
NEAR PATIENT TESTING - URINALYSIS
Visual’ exam
- urine appearance
- clear
- cloudiness
NEAR PATIENT TESTING - URINALYSIS - Colour
- Watery
- Yellow
- Orange
- Red/brown B/R
- Milky white
Unusual odour
NEAR PATIENT TESTING–URINALYSIS - DIPSTICK
Leukocytes number
small number
NEAR PATIENT TESTING–URINALYSIS - DIPSTICK
What does nitrates/Leukocyte Esterase presence indicate?
sign of infection or inflammation
NEAR PATIENT TESTING–URINALYSIS - DIPSTICK
Urobilinogen
breakdown product of bilirubin, not normally present
NEAR PATIENT TESTING–URINALYSIS - DIPSTICK
Bilirubin
0-85% from haemoglobin released by breakdown of senescent red blood cells; 12-20% breakdown myoglobin, bone marrow, not normally present
NEAR PATIENT TESTING–URINALYSIS - DIPSTICK
Is protein [albumin] normally present?
not normally present
NEAR PATIENT TESTING–URINALYSIS - DIPSTICK
Glucose
present in small amount
NEAR PATIENT TESTING–URINALYSIS - DIPSTICK
Specific Gravity
= weight of urine/weight of water
NEAR PATIENT TESTING–URINALYSIS - DIPSTICK
Ketones
appear in urine as a consequence of accelerated fat metabolism; when in large amount: fruity odour “pear drops”
LIVER FUNCTION TESTS (LFT) - Total Protein (serum)
→ Albumin
→ Globulins
→ A/G ratio (albumin-globulin)
→ Protein electrophoresis
→Urine protein
LIVER FUNCTION TESTS (LFT) - bilirubin
direct + indirect = total
LIVER FUNCTION TESTS (LFT)
Total protein (serum)
Bilirubin
Aspartate transaminase (AST)
Alanine transaminase (ALT)
Gamma-glutamyl transpeptidase (GGT)
Alkaline phosphatase (ALP)
BASIC METABOLIC PANEL (UREA & ELECTROLYTES)
- Four electrolytes:→ sodium (Na+)→ potassium (K+)→ chloride (Cl-)→ bicarbonate (HCO3-)
- blood urea (blood urea nitrogen, BUN)
- creatinine
- glucose
COMPREHENSIVE METABOLIC PANEL
General tests
- Serum glucose
- Calcium
COMPREHENSIVE METABOLIC PANEL
Kidney function assessment
- Blood urea nitrogen (BUN)
- Creatinine
Liver function assessment
- ALPALTAST Bilirubin
COMPREHENSIVE METABOLIC PANEL - Electrolytes
- Sodium
- Potassium
- Chloride
- Carbon dioxide (CO2)
COMPREHENSIVE METABOLIC PANEL - Protein
- Albumin
- Total protein
Inferential Statistics
- relationship between variables
- is based of a hypothesis
- needs a formal statistical test to guide policy making, health care choices etc
- tests depending on the nature of the data etc. Mann-whitney U-test, t-test
- evaluation of test results in terms of statistical significance
Steps in Performing a Statistical Test
- formulate null and alternative hypothesis
- evaluate the data and choose an appropriate statistical test for the data
- perform the statical test
- obtain test statistical significance of the results
- reject or accept the null hypothsois
Null hypothesis
H0
Alternative Hypothesis
H1
Why do we aim to reject hypothesis?
rejecting a hypothesis is often more feasible task than proving it
null hypothesis
a null hypothesis is the statement that is considered to be true unless the data prides sufficient evidence to reject it. one does not prove null hypothesis
statistical significance
the observed results is not by chance
p-value
probability of observing the result or more extreme result (test statistic) given the null-hypothesis is true
note about the use of p-value
- a large p-value does never prove the absence of an effect but allows us to conclude that there is no sufficient evidence of the effect to be present
- the cut off is 0.05 is quite arbitrary
statistical test depends on
→ type of data
→ research question
→ sample size
test of correlation
- Pearson
- spearman
Test of frequency
test whether the observed data deviates from the expected
Chi-Squared
- represents how the observed data (o) data/frequency deviates from the expressed (E)
- x^2 is another probability distribution and ranges from 0 to infinity
Fisher’s Exact Test
- to be used when sample size is small (cell size <5)
- p-value is calculated exactly as opposed to an approximation
Test of frequency examples
Chi-Squared
Fisher’s Exact Test
Quantitative Normally Distributed Data
Student T-Test
Why is it called a student t-test?
the test statistic t follows student t-distribution with n-2 degrees of freedom
Student T-Test
a comparison of group means between two groups (2 sample t-test)
Student T-Test Assumptions
assumes the data are normally distributed
assumes equal variance (for unequal variance Welch’s t-test)
Testing for Normality
visual inspection eh. using histogram
provides a single p-value
Testing for Normality Examples
d’agostino and Pearson normality test
Evaluation Testing for Normality
evaluates the skewness and kurtosis of the distribution to quantify how far the distribution is from normal distribution in terms of asymmetry and shape
Examples Quantitivative Non-Normal Data
Mann-Whitney U-test
More than 2 groups: ANOVA - analysis of variance
Mann-Whitney U-test
- a non-parametric test for non-normal data
- a comparison of group means between two group
- also called wilcoxson rank-sum test
One-Way ANOVA
effect of one categorical variable on the continuous outcome variable
Two-Way ANOVA
effect of two categorical variable incombination, on the continuous outcome variable
Post-Hoc Testing
- find out which groups are different from one another
- multiple comparison between group - this is controlled by these test (adjusted p-value are produced)
Tukey’s Honestly Significant Difference (HSD)
- often referred to as Tukey’s method
- the most commonly used
- compares all the levels against each other
- assumes equal variance across groups. if not statisfied, see eg. Games Howell Post Hoc Test
More than 2 groups: Kruskal-Wallis Test
- a non-parametric test
- an extension of Mann-Whitney U-test is for 2 groups
- based on ranking the data
Post-Hoc Testing Example
Tukey’s Honestly Significant Difference (HSD)
More than 2 groups: Kruskal-Wallis Test
What is ANOVA?
analysis of variance
ANOVA
- a comparison between groups mean between mutilple groups
- compares the variance between groups to variance within groups
What does ANOVA tell you?
tells only whether there is a difference in the mean levels of the outcome between the groups but does not tell in which group
What is a Statistical Model?
a simplified representation of reality
a model for the association structure in the data
most common ones are regression models
What does a regression model represent?
a regression model represents how a dependent (outcome) variable Y depends on one or more independent variables (covariates) X
Linear Regression (Simple and Multiple)
- a generalisation of t-test, ANOVA
- to only for differences in group means (categorical variables) but also for association between continuous variable
- can be included just one covariate (simple linear regression) or many (multiple linear linear regression)
- mathematical formula
Linear Regression (Simple and Multiple) Assumptions
- linearity
- = normality and homoscedasticity (constanct variance) of the errors
- independence of the errors
Confidence Intervals (CI)
it is the point estimate beta that an error (SE) associated within telling about its variability if we repeated the test multiple times (remember SEM) - 95%
What can we use SE bar to estimate?
we can use the SE to estimate where the point estimate is with a certain confidence if we were to repeat the test in multiple samples from the population
What does it mean if Cl does not include 0?
if 95% CI does not include 0, there is a statistically significant association ie. p<0.05
Standard Curve in Analytical Chemistry
- in analytical chemistry, standard curve is use to determine the concentration of an unknown quantity given your data
- this is obtained via linear agression
Logistic Regression
- for categorical outcomes Y, eg. disease status yes/no
- the predictor variable X can be categorical or continous
- again, can have one or multiple predictors (covariates) in the model
Logistic Regression - How are the results given?
Results are given as odds ratio (OR)
→ approximate how much more likely (or unlikely) it is for the outcome to be present (event to occur) among group A than among reference group B
Logistic Regression - What does an OR of 1 suggest?
OR = 1 means that there is no increased odds
Logistic Regression - if the 95% CL includes 1, what does that mean?
if the 95% Cl includes 1, the association is not significant
Odds Ratio (OR) vs Risk Ratio (RR) - if in the rare event where RR is close to OR, what does that mean?
(a and c are close to 0)
RR is easier to interpret but OR is more widely used because R cannot be used in designed case-control studies because we don’t know the real total number of exposed individuals/events
Testing for Causation
- randomised controlled trails - the golden standard
- direct acyclic graphs eg, path analysis
- mendelian randomisation - uses genetic data, very popular
- the two later involve several assumptions
Sensitivity
if I have disease X, what is the likelihood I will test positive for it?
true positive / (true pos + false negative)
Specificity
if I do not have disease X, what is the likelihood I will test negative for it?
True Negatives / (True Negatives + False Positives)
Positive Predictive Value (PPV)
is the proportion of those with a POSITIVE blood test that have Disease X.
“If I have a positive test, what is the likelihood I have Disease X?”
PPV = True Positives / (True Positives + False Positives)
Negative Predictive Value (NPV)
is the proportion of those with a NEGATIVE blood test that have Disease X
“If I have a negative test, what is the likelihood I do not have Disease X?”
PPV = True Negatives / (True Negatives + False Negatives)
