Epidemiology (PDF)

# Anaemia ## Regenerative - Irresponsive of BM ## Non-regenerative - Either regenerative/non-regenerative ## Epidemiology ### Truusha Hadlinmafi ### Prevalence, Cumulative Incidence: - Percentage - Proportion #### Example: - 60% grower pigs, 5 developed diamond shaped skin lesions (2 in 1s...

# Anaemia ## Regenerative - Irresponsive of BM ## Non-regenerative - Either regenerative/non-regenerative ## Epidemiology ### Truusha Hadlinmafi ### Prevalence, Cumulative Incidence: - Percentage - Proportion #### Example: - 60% grower pigs, 5 developed diamond shaped skin lesions (2 in 1st week, 3+ in 3rd week) - Same last month, but not anymore. ### Incidence/Density/Rate/Risk #### Prevalence: - Sick animal (5) / Population at risk (50) = 0.1 or 10%. #### Numerator: - Not a part of a denominator. #### Cumulative Incidence: - Sick animal (5) / Population at risk (tested animals) (45) = 0.11 or 11%. #### Newly added cases: - Total cases. - # of new cases of disease discovered by the total # of individuals in the population at risk. #### Incidence Rate: - This is a proportion, not a true value. #### Incidence: - True rate. Incidence density. Incidence rate/incidence risk. - Incidence per unit time. #### Cases count with time periods. #### Example: - 5/(2 x 2 Wk + 3 x 3W + 40 x 4 Wk) - 5 / 178 Wks during 1 month = 3 per 200 pig weights. - Measurement of disease (P), Prevalence, Cumulative Incidence (CI), Incidence Rate (IR). ## Diagram: | 2 months | Beginning Period | Incident Rate | New Cases | |---|:---:|:---:|:---:| | | 12 dogs | | | | | 24 dogs | | | | | At the end 14 dogs got ill. | | | | | 4 dogs got ill. | | | | | 2 new additions | | | ## Multifactorial Nature of Disease - Host - Different shapes, uniqueness - Host - age, breed, gender, immunity, immunization. Fitness - Agent - Bacteria, fungi, virus - Protozoa, helminths, tapeworms, cestodes, antigenic shift & drift. Septicaemia. - Environment - Dry, humid, wet, windy, misty #### Example: - Haemorrhagic. - Tuberculosis. - Toxoplasmosis. - Trypanosomiasis - Randomized controlled clinical trials are the most powerful design used in medical research. # Necessary & Sufficient Cause - In cold weather, many catch a cold. Cold weather is necessary and/or sufficient cause to catch a cold. - Cold is common. - Annual vaccine outbreak is preventable. ## Pasteurella multocida - Cause HS. # Routes of Disease Transmission - Direct - Parvo - Vector - tick fever. 6 different types. - **Common dog:** haemoparasitic, Ehrlichia. - Wind - Influenza - Rain (water) - Leptospirosis - Drought # Exposure | Exposure | latent period | Tested | Detectable | Clinical signs | Live/dead | Clinical signs | |----------------|-----------------|--------|-------------|--------------|-----------|---------------| | | | | | | | | | Carriers | No clinical signs | | | | | | # Detecting Subclinical Diseases - **Example:** Mastitis. No recognizable clinical findings. Electrical conductivity. Cell count change. - Early detection will always be the style. - How to compare the old test with the new test. - Which one detect better. - Coliform mastitis test & electrical conductivity. - **Agreement & Disagreement:** will be interesting. #### Example - Radiograph for hairline fracture in bone. - One clinician may diagnose as fracture, whereas another one diagnose as not a fracture. - Agreement & disagreement. # How/Why to Prove/Disprove Cause? - Koch's postulates & its modification. - Henle-Koch's postulates for causation. - **A) Agent & disease:** Agent must occur in every case. - Tuberculosis. M.tuberculosis & M.bovis. - **B) Agent & disease:** Agent does not occur in other cases (non cases). - Tuberculosis. - **C) Condition of Agent:** Condition of agent can be induced by a culture. - **Modification:** due to difficulties & risk factors in place of agent. (Even environment & other causes). - In proving/disproving. ## Method of Agreement: - Same agent in different outbreaks (Influenza). ## Method of Difference: - One factor is different between diseased & not diseased basis for experiments (CCRDO). ## Method of Concomitant Variation: - Dose response relationship (Tuberculosis). ## Method of Analogy: - If similar to a known condition (Hanta virus). # Epidemiology & Epi - demos - logy - Occurrence & distribution of health, health determinants & health related conditions. - In populations or in populations. - Study of ds in a situation. - Production as a surrogate measure of health. - **Descriptive:** - When little is known. - Pattern according to age, geography, gender, etc. - **Analytic:** - To study causation. - Hypothesis testing. Cock's postulates. ## Diagram: | EPI | DEMOS | LOGY | |---|:---:|:---:| | Opon/on top of Population | | Studies | | Studying Regarding | | Populations | | | | | | | | | - Observation: Not only seen - brain + eyes (seen) + ears + other senses. - Health & disease interaction. - Surrogate & indirect. # + Test results & tests both Important ## Example: - Covid 19. - Biological & Statistical Concepts. ## Biology: - Agent, host, environment. - Multifactorial nature of conditions. - Infectious & non-infectious agents. - Metabolic conditions, nutrient deficiencies, no agents. - Necessary & sufficient cause (exception - rabies). - Route of transmission. - Direct, vector, wind, rain, drought. - Modification of host. - Vaccines, etc. - Incubation & latent periods. - Vaccination & treatments. - Detection of subclinical disease. ## Statistical Data (Information) - Type of statistical data. - Data transformation. - Normal, t, chi-square, F distribution. - Parameters, estimates. - Point estimate. - Variability. - Confidence limits - Pooled (paired t-test, ANOVA. - Multiple comparisons. - Proportions & chi-square. - Simple regression. - Multiple regression & correlation - Logistic regression. - Discontinuous variable (Yes/No). # Analytical Epidemiology - Measure of association. - Exposure & Outcome. - Incidence Rate - Rate (density). - Case fatality / 100 man years - unit - Cumulative incidence. - denominator difference. ## Risk Ratio: - Relative risk. ## Odd Ratio: - Relationship btw RR & OR. ## Attributable Fraction (CAF): - Estimate prevalence far away from true value. ## Sample size: - New cases. ## Incidence Rate: - Henle-Koch's postulates. - **A) Agent & disease:** Agent occurs in every case. Tuberculosis. - **B) Agent & disease:** Agent does not occur in other cases (non cases). Tuberculosis. - **C) Condition of agent:** Condition of agent can be induced by a culture. # FMD - **High morbidity**. - **Low fatality**. - **Low case fatality.** - Increase during rainy seasons. - Transmitted through wind. - In cloven-footed animals. - Case fatality - died animals. ## Diagram: | Sick animals | Case Fatality | |---|---| | | | | Children | | ## Conditions: - Similar but not for same virus. - **Not zoonotic.** - **From FMD only calves dies.** - Irritable, pant all, blisters occur in brown. - They cannot eat & drink. - They undergo starvation & thirst (hypoglycaemia). - Ultimately die. - Also cause myocarditis. - Adult animal can bare that pain & irritation. # BO - **Low morbidity** - **High case fatality** # Rinderpest - **High morbidity** - **High fatality.** - **High case fatality.** # Population At Risk (PAR) - The population that is exposed to the occurrence of a vital event. - Example: - Total population. - Legally married population in the case of the death. - Divorces in the case of dye suspected. - Some people dye due to particular, dye conditions whereas some are dye due to other reasons. #### Example: - During FMD outbreak - PMD, Sold, Stolen. - Covid outbreak - People dye due to covid 19, other ds i.e. pressure, sugar, chlolesterole, people cannot go to hospital due to current. - Some individuals in the population are protected. Eg: PMD vaccination, flu shots in cold seasons. - Those protected individuals, - Not counted as population at risk. ## Repeatedly occurring: - Also in epimeral fever - 3 day sickness. - But chronic ds - tuberculosis. Life time, existing as a population at risk. - Should add to calculate population at risk intermittently. # Then New Addition Animals into Population: - Newly added animals into population # Measurements of Diseases - To compare amount of ds between population. - Prevalence (P), Parameter (P), Estimate (P). - Also tested (PAR). # Point Prevalence: - Snap shot prevalence. ## Period prevalence: - Prevalence x time period. (Incidence x duration). - Chronic. #### Incidence, duration or born can be changed. # Strength of Association: - Exposure & Outcome: Risk Ratio , Relative Risk. <start_of_image> Xs - 80% diseased. - Risk of ds 80% among exposed. - 20% diseased. - Risk of ds - 20% among unexposed. ## Diagram: | Diseased | | |---|---| | | | | **Exposed:** 80 | 20 | | | **Non exposed:** 20 | 80 | | | | | | ## Risk Ratio (Relative Risk): - 80/100 ÷ 20/100 = 4. - (**80%**: exposed & died individuals/ **20%**: unexposed & died individuals) ## Odds Ratio: - a/ a+b ÷ c/ c+d = ad/bc. # Effect of Exposure on Individuals: ## Attributable Fraction: - Risk difference. - Amount of risk attributable to exposure. - Other reasons for the same disease, many causes for pneumonia. ## Diagram: | D+ | | |---|---| | E+: 10 | 90 | | E-: 5 | 95 | | | | ## AF: - (10/100 )-(5/100)/(10/100) - (a-b)/(a) - 5/ 100 - can AF be negative. ## Covid AF: - 67 %. only. ## Protective Factor: # Basic Study Designs: - At the end of this lecture you should know: selection or sampling & allocation of individuals. - To 1 or 2 groups. - Follow up, period, longitudinal, calendar time. - Types of data collection, measurement, study design & their analysis (brain mind). - **Strength of association:** between exposure & outcome. - **Interpretation of results**: limitations & extrapolation. ## Cross Sectional (Prevalence) Study: - Right budget. - An estimate of parameter & outcome. - Quick result. - Weak association between exposure & outcome. - Not good for rare disease. ## Diagram: | X | X | X | X | X | X | |---|:---:|:---:|:---:|:---:|:---:| | X | | | | | X | | X | | | | X | | | X | X | | X | | | | X | | X | | | | - **Select (Sample):** Not diseased / diseased among exposure. - **No. of diseased/ not diseased among non-exposed.** - **Only one sample.** - **Disregarding exposure or disease outcome status.** # Advantages: ## Establish Causation better: - Correct meaning of RR or OR, due to fallow up. ## Can be extrapolated meaningfully: - Eg 1: A study on exposure to fluoride in drinking water, dogs from NCP central province. - Eg 2: Tick fever & brown dog tick bite. Take dog 10 with ticks & 10 dogs with out ticks. - Observe over fixed period of time. - Check the outcome. ## Case - Control Study Design: - To be very sure of the cause - follow up. - Select 2 samples on outcome Cd & non Cd. - Explore into history retrospectively for exposure - Rare ds. - Example: Trypanosomiasis in dogs. - Eye problems. - Rare history. - Contact finding all possible. - Follow up present but back words. # Disadvantages: ## Conditions with long duration not practical to follow up: ## Require lot of time, energy, resources, ## exposure to fluoride in drinking water: - Dogs from NCP central province. ## Tick fever & brown dog tick bite: - Take 10 dogs with ticks & 10 dogs without ticks. - Observe over fixed period of time. - Check the outcome. # Advantages: - Can be done fast. - Current situation evaluation. - Acute/chronic can't be separated. - Not very strong evidence of causation. - Eg: mange in pig. - Snap shot. # Disadvantages: - **Condition "short duration" can be missed.** - **RR or do not be expressed (robust measure & can be used).** - **Favors chronic conditions.** - It exposure physically does not occur at all the time? - **Rare conditions need large sample size.** # Cohort Study Design (prospective) - **Select 2 samples: (Cohorts) on exposure** - Exposed & non-Exposed. - **Observe specified period & Count outcome (ds) or good association between exposure & outcome.** - Expensive, time consuming, expose over a period. ## Diagram: | Exposure | Time (Period) | Outcome | |---|---|---| | X | Time (prospective, longitudinal.......calender) Outcome | X | | X | X | X | | X | X | X | | X | X | X | | X | X | X | | non exposed | Time (Period) | Outcome | # Advantages: - **Establish causation better.** - Correct meaning of RR or OR, due to follow up. - **Can be extrapolated meaningfully.** - Eg-1: A study on exposure to fluoride in drinking water, dogs from NCP central province. - Eg 2: Tick fever & brown dog tick bite. Take dog 10 with ticks & 10 dogs with out ticks. - Observe over fixed period of time. - Check the outcome. # Disadvantages: - Conditions with long duration not practical to follow up. - Require lot of time, energy, resources. # Case-Control Study Design: - To be very sure of the cause - follow up. - Select 2 samples on outcome Cd & non Cd. - Explore into history retrospectively for exposure. - Rare ds. - Example: Trypanosomiasis in dogs. - Eye problems, rare history. - Contact finding all possible. - Follow up present but a backwords. # Advantages: - Establish causation better. - Correct meaning of RR or OR, due to follow up. - Can be extrapolated meaningfully. - Eg-1: A study on exposure to fluoride in drinking water, dogs from NCP central province. - Eg-2: Tick fever & brown dog tick bite. Take 10 dogs with ticks & 10 dogs with out ticks. - Observe over fixed period of time. - Check the outcome. # Disadvantages: - Conditions with long duration, not practical to follow up. - Require lot of time, energy, resources. # Rare ds. - Example: Trypanosomiasis in dogs. - Eye problems - rare history. - Contact finding all possible. - Follow up present but a backwords. # Observational Period: - Rare ds (outcomes). - Some can't be raced, matched with control, collect over period of time. #### Example: - Dog, mega oesophagus in UTH, examine separation anxiety as a potential cause. # Advantages: - Can be used on new, rare conditions. # Disadvantages: - Improper meaning of RR & OR. - Therefore, not used. - Require lot of questioning, in depth history. - Not practical at times. - No strong evidence of causation. # Incidence: - Diseases due to other reason! - Recurrences: ## Diagram: - 1. Incidence, new additions. - 2. Prevalence - burden of ds - represent. - 3. Cure. - 4. Die due to ds or other reasons (evaporation). - 5. Vaccinations. - 6. No water in tab. - 7. At beginning: prevalence 0, mortality 0, recurrent 0. - 8. Recurrence. - 9. Community. - 10. Mortality or cured. - 11. Die due to ds. # Prevalence Rate: - No unit measurement, proportion included in the denominator. - Numerator is # of population at risk. - Denominator is CF (No. tested). # Incidence Rate: - Animal time is the denominator. - Only new cases. - No of new cases. - Summation of risk periods by individuals at risk (PAR). # Measure of Disease Frequency: - Prevalence rates. - Incidence rates. - Crude. - Cumulative # Sampling in Epidemiology: ## Population: - Too large. - Several characters. - Practical problems. ## Sample: - Sample size determined. - Small, easy to deal. - Should be large enough for extrapolation. - Some may not cooperate - Some may be lost. ## Vaccine Efficacy: - **RR, AF.** - **Vaccine efficacy:** P exposed - unexposed / P exposed - **Foot rot outbreak in August.** ## Diagram: | | FR + | FR - | Total | |---|:---:|:---:|:---:| | **Vaccination (January)** | | | | | **Non-vaccinated:** | 94 | 328 | 422 | | **Vaccinated:** | 21 | 296 | 317 | ## Follow up study (cohort - clinical trials): - **Vaccination efficacy:** (94/422) - (21/317)/ (94/422). - 0.22 - 0.06 / 0.22 - 0.16/ 0.22 - 72.7%. # Vaccine (+) & Tests for Influenza Study: - **Vaccine (+):** Cases (test (+) for influenza). - **Vaccine (-):** Subject with rest results. - **Vaccine (ti):** Control according to the post test results - lab. - **Vaccine (+):** Lab - confirm influenza (+) - **Vaccine (+):** Lab - confirm influenza (-). - **Vaccine (-):** Lab - confirm influenza (+) - **Vaccine (-):** Lab - confirm influenza (-) ## Source Population: - Those who visit medical institutions due to ILI during influenza season. # Attributable Fraction (CAF) is important: - Vaccine efficacy probability. - Individual deaths due to (attributable) covid 19. # Clinical Trials: - Systematic study. - **Prophylactic** or **Therapeutic**. - Assessment of drugs: - **Pharmacological** & **toxicity** trials on target animals. - **SP** or **Laboratory.** - **Therapeutic effect** & **safety** trials on target animals. - **SP** of rabbits & mice (in safety trials). - **Clinical evaluation** of efficacy - in hela cells. - **Clinical evaluation** or efficacy - in field trials. - **Operational conditions.** - **Post authorization surveillance**. - **AU (both clinical & field trials)** can be therapeutic & be conducted on diseased or healthy individuals. ## Outline of basic clinical trials: | Basic | Clinical trial | Outcome | |---|---|---| | Population with condition | Sample & allocation | Non-experimental intervention | | | experimental intervention | | # Randomized Controlled Clinical Trials: - Can controlled form (no control). - an controlled clinical trial. - Control group can receive a standard treatment (+ control) or placebo (- control). ## Random allocation of study subjects - resembles cohort study. - **Community trial (mostly in human medicine eg: Fluoridation of water supplied to a community).** # Epidemiology & Statistics - **May scientific conclusion is important:** Scientific procedures - hypothesis testing. - **Scientific proceeds - conclusions:** Statistical & hypothesis testing. - **No subjective, conclusion:** It will ... . may rain. - **Probability of rain:** 50%. **Type I & Type Error** - Type error, select correct size representative sample cross section of population. - Sample: random, simple, stratified. - Convenient/judgment sampling with interval. # Parameter (P) & Estimate (P) # Why Scientific Procedure? - Only a sample is studied not the entire population. - **Population (parameter P) is often too large and has several characters. Practical problems.** ## Sampling: - **Sample:** - Vary from sample to sample. - Sample size determination; - Small, easy to deal. - Should be large enough for extrapolation. - Some may not cooperate. - Some may be lost. # Requirements for Scientific Conclusions: - **Appropriate experimental design / sampling technique.** - Homogeneous unit & lab animal. - Heterogeneous unit. - Stratified random sampling - different variables. - **Completely randomized & field experiments. Block design. ** # Homogeneous (F/M) Population: - **Simple random sampling - students in class on their own.** # Homogeneous Population: - **Simple random sampling - students in class on their own.** # Is adequate, no. of replicates/sample size: - Adequate to estimate experimental error (least df or using the correct formula for sample size). ## Diagram: - 0 - S # Factors for sample size: - Variability, precision (level of accuracy). - Level of significance (a) (usually 0.05). - Power, Ci (P) (usually 0.9). # Statistical Method: - Based on scale of measurement of data: - **Nominal/Categorical:** Least amount of measurement, no order between values. Value itself is not meaningful (eg: gender, CH/D). - **Ordinal/Rank:** Some amount of measurement, value itself is not meaningful, no order, but there is amount of measurement between values (eg: taste, severity, marks). - **Interval:** Exact measurement, there is order, interval is meaningful (eg: temperature). - **Ratio/Continuous:** An exact measurement. Order is meaningful, interval is meaningful, value itself is meaningful (eg: height, weight). ## Diagram: | Scale | Expected Distrubution | Method | |---|---|---| | Nominal | Binominal (if 2 categories) | Categorical data analysis methods | | Ordinal/Rank | Multinominal (if > 2 categories) | | | Interval | Normal | | | Ratio | Normal | Parametric methods | # Scientific results are reported: - **Probability (level of significance).** - **Confidence interval.** - **K statistics, fishers test, fishers exact test.** ## Comparing categories and continuous variables: - One sample (paired data) - paired t-test. - Two sample - pooled t-test, correction. - Contingency table - chi-square. - Agreement or disagreement - Kappa statistics. In some situation it's correction. - One way analysis - of variance - in recall memories. - Comparing categories on dichotomous - t-test (usually in cases). - Examining relationship between two continuous variables. - chi-square. - Examining a variable. - linear regression, correlation. - Multiple regression. # Impaing categories on continuous variables: - Log Br regression. - Paired t-test. - Sample paired with a known value, (population) - t-test. - Eg: compared pulse rate (at 40/m) in horses (debu hon) with other horses (28-40/m). - t-test (1d or > 1) - Co-0.05 (level of significance). # Comparing categories: - **Paired, t-test, student/t-test.** - **One way analysis of variance (ANOVA):** Non-parametric, 2 groups compared. - **chi-square:** - Used when two categories are not equal (unequal variance in two categories). - Eg 1: Male & female, differences on height. # Paired t-test: - Before & after treatment or paired observational t-test on a sample of one. - Differences (eg: erythropoetin injection (4000 IU) increase PCU in dogs. - Continuity correction & fishers exact test. - Least squares method, regression. - It's 1 variable dichotomous, other continuous variables. # One way analysis of variance: - Examining relationship & correlation between 2 continuous variables. - Linear regression & correlation. - Spearman? Logistic, non gaussian distribution only. - Strength of association between Exposure & Outcome. - Probability attributable to individuals. - Can be extrapolated to population. - Non parametric distribution of data. - Cross sectional data collection & analysis in studies. # Sampling, data collection & analysis, in cohort studies: - Sampling, data collection & analysis in case controlled studies. # Clinical trials - field, clinical, & raceme trials. # Randomised Controlled Clinical Trials: - Open (an controlled form - no control) & controlled clinical trial. - Control group can receive no treatment (+control) or a placebo (-control). - Random allocation of study subjects resembles cohort study - Community trial (Mostly in human medicine eg: fluoridation of water supplied to a community). # Epidemiology & Statistics: - May scientific conclusion is important. - Scientific procedures - hypothesis testing. - No subjective, conclusion; it will ... . may rain. - Probability of rain: 50%. # Type 1 & Type Error - **Type error:** selecting correct size representative sample cross section of population - **Sample:** random, simple, stratified. - **Convenient/judgment sampling** with interval # Parameter (P) & Estimate (P) # Why Scientific Procedure? - Only a sample is studied, not the entire population. - Population (parameter P) is often too large, with several characters. Practical problems. # Sampling - **Sample:** - Vary from sample to sample. - Sample size determination: - Small, easy to deal. - Should be large enough for extrapolation. - Some may not cooperate. - Some may be lost. # Requirements for Scientific Conclusions - Appropriate experimental design/sampling technique. - Homogeneous units & lab animal. - Heterogeneous unit. - Stratified random sampling, different variables. - Completely randomized & field experiments. Block design. # Homogeneous (F/M) Population - Simple random sampling, students in class on their own. # Homogeneous Population - Simple random sampling, students in class on their own. # Is adequate, no. of replicates/sample size - Adequate to estimate experimental error (least df or using the correct formular for sample size). - Precise & reliable estimate. ## Diagram: - 0 - S # Factors for sample size: - Variability, precision (level of accuracy). - Level of significance (a) (usually 0.05). - Power, Ci (P) (usually 0.9). # Statistical Method - Based on scale of measurement of data: - Nominal/Categorical: Least amount of measurement, no order between values. Value itself is not meaningful (eg: gender, CH/D). - Ordinal/Rank: Some amount of measurement, value itself is not meaningful, no order, but there is amount of measurement between values (eg: taste, severity, marks). - Interval: Exact measurement, there is order, interval is meaningful (eg: temperature). - Ratio/Continuous: An exact measurement. Order is meaningful, interval is meaningful, value itself is meaningful (eg: height, weight). ## Diagram: | Scale | Expected Distrubution | Method | |---|---|---| | Nominal | Binominal (if 2 categories)| Categorical data analysis methods | | Ordinal/Rank | Multinominal (if > 2 categories) | | | Interval | Normal | | | Ratio | Normal | Parametric methods | # Scientific results are reported: - Probability (level of significance). - Confidence interval. - **K statistics, fishers test, fishers exact test.** ## Comparing categories and continuous variables: - One sample (paired data) - paired t-test. - Two sample - pooled t-test, correction. - Contingency table - chi-square. - Agreement or disagreement - Kappa statistics. In some situation it's correction. - One way analysis - of variance - in recall memories. - Comparing categories on dichotomous - t-test (usually in cases). - Examining relationship between two continuous variables. - chi-square. - Examining a variable. - linear regression, correlation. - Multiple regression. # Impaing categories on continuous variables: - Log Br regression. - Paired t-test. - Sample paired with a known value, (population) - t-test. - Eg: compared pulse rate (at 40/m) in horses (debu hon) with other horses (28-40/m). - t-test (1d or > 1) - Co-0.05 (level of significance). # Comparing categories: - **Paired, t-test, student/t-test.** - **One way analysis of variance (ANOVA):** Non-parametric, 2 groups compared. - **chi-square:** - Used when two categories are not equal (unequal variance in two categories). - Eg 1: Male & female, differences on height. # Paired t-test: - Before & after treatment or paired observational t-test on a sample of one. - Differences (eg: erythropoetin injection (4000 IU) increase PCU in dogs. - Continuity correction & fishers exact test. - Least squares method, regression. - It's 1 variable dichotomous, other continuous variables. # One way analysis of variance: - Examining relationship & correlation between 2 continuous variables. - Linear regression & correlation. - Spearman? Logistic, non gaussian distribution only. - Strength of association between Exposure & Outcome. - Probability attributable to individuals. - Can be extrapolated to population. - Non parametric distribution of data. - Cross sectional data collection & analysis in studies. # Sampling, data collection & analysis, in cohort studies: - Sampling, data collection & analysis in case controlled studies. # Clinical trials - field, clinical, & raceme trials. # Randomised Controlled Clinical Trials: - Open (an controlled form - no control) & controlled clinical trial. - Control group can receive no treatment (+control) or a placebo (-control). - Random allocation of study subjects resembles cohort study - Community trial (Mostly in human medicine eg: fluoridation of water supplied to a community). # True ds (D) status & Diagnostic Test (T) results - Gold standard. - True prevalence D+/N - Apparent prevalence T+/N. - How close D+ & T+. ## Diagram: | **D+** | **D-** | |---|---| | **T+** | True + False+ | | **T-** | False- True- | ## Estrus & Estrus - - Inserminating false (+) = waste of semen. - Inserminating false (-) = waste of days. ## Heat Detection: - Mounting & testing. - Redness, vulva & discharge. ## Prevalence: - Chance to detect ds. - Following. ## Diseased/Non Diseased - Type 1 test - straight forward. - Type 2 test - cut off decided: ## Diagram: | | Diseased (Number) | Non disease (Number) | Total (Chamber) | |---|---|---|---| | **Test results:** | | | | | **Positive (Number)** | True positive A | False positive B | T test positive | | **Negative (Number)** | False negative C | True negative D | T test negative | | | | | | | | | | | | | T disease | | | | | T non-disease | | | | | Total | | | - T disease & T non-disease - does not change. ## Sensitivity: - Probability of a test detecting a true positive (a/ a+c). ## Specificity: - Probability of a test detecting a true negative (d/ b+d). ## (+) Predictive value: - Probability of a test (+) being a true positive (a / a+b). ## (-) Predictive value: - Probability of a test (-) being a true negative (a / c+a). ## Sensitivity & Specificity: - Best if N is lower. ## Multiple tests: - Parallel reading improves sensitivity. - All tests used on all individuals. - At least one of the test need to be (+) for individuals to be positive. - Report maximum rate of pregnacy when DF semen is freely available. ## Serial reading: - Improves specificity - subsequent tests are used only on (+) individuals - one of the test to be (+) for individuals to be negative (babesiosis in dogs in a peripheral clinic). # Testing Strategies - Parallel Testing: - 2 screening tests performed at the same time, the results are subsequently combined. - Higher sensitivity but lower specificity. - Serial testing: - 2nd screening test performed only if the result of the first screening test is positive. - Improve specificity but the cost of lower sensitivity. - **More than one test.** - Clinical & field findings. - Using all tests on all suspected individuals - tick fever diagnosis. ## Diagram: | **Test 1** | **Test 2** | **Test 3** | **Test 4** | **Test 5** | **Diagnosis** | |---|---|---|---|---|---| | **Depression** | **Pale** | **Bounding** | **Hepatome** | **Dark pulse** | **Strongly suggestive** | | + | + | + | + | + | | | + | + |