PUBH2001-Epidemiology From Principles to Practice

1 PUBH2001 Tutorial 2 Guide Q1. The word epidemiology comes from the Greek words epi , meaning on or upon, demos, meaning people, and logos, meaning the study of. Many definitions have been proposed, but the following captures the underlying principles and public health spirit of epidemiology: Epidemiology is the study of the distribution and determinants of health -related states or events in specified populations , and the application of this study to the control of health problems. Epidemiology is the basic s cience of disease prevention and plays major roles in the development and evaluation of public policy. It is now used together with lab research to identify environmental and genetic risk factors for diseases and to shed light on the mechanism s involved in pathogenesis. Keeping this in mind c omplete the following statements by choosing from the given words : (l) Investigating epidemics and new diseases; (2) Studying the biologic spectrum of disease; (3) Instituting surveillance of community health interv entions; (4) Setting disease control priorities; (5) Helpful in the diagnosis, treatment, and prognosis of clinical disease; (6) Improving health services research; (7) Providing expert testimony in courts of law. Note: M ore than one word can be suit able for some spots. Q2. Descriptive Epidemiology is concerned with describing the general characteristics of the distribution of disease particularly in relation to time, place and person. What are the 4 W words that tell us what descriptive epidemiology is about & what is the ‘W’ word for analytic epidemiology ? What, who, where and when ( Why for analytic epidemiology) Q3. Write two research questions that can be answered using descriptive approach and two examples of research questions that can be answered using analytic approach. Descriptive approach: What was the p revalence of type 2 diabetes among Australian adults in 2017? What is the incidence of colon cancer among Australian females 60 years and older. Analytic approach: Is there an association bet ween obesity and the risk of uterine cancer? Does breastfeeding reduce the risk of breast cancer? Analytic Epidemiology is the process of testing a specific hypothesis mainly about association between potential exposures (or determinant s) and outcome of interest. A study is performed to examine the relationship between the exposure to a risk factor and the outcome, typically a disease but can be disability, injury or any health related state. 2 Q4. Table 1.1 Water sources and cholera mortality in London, 9 th July to 26 th August 1854. Source of Water Total no of households No of cholera deaths Deaths per 10,000 house hol S & V Company 40, 046 1263 315 Lambeth Company 26, 107 98 38 Rest of London 256, 423 1422 55 During 19 th century two private companies that obtained water directly from river Thames supplied water in London. Southwark & Vauxhall (S&V) and Lambeth as listed in the table. Between 1849 and 1853, when London was free of cholera, Lambeth Company moved its water source to upstream, an area outside London and less polluted, while the S&V Company continued to draw water from a downstream source in London. Snow collected data on the number of houses supplied by each company and when cholera epidemic recurred in 1954 he collected data on sources of water for households of those who had died of cholera. The above table shows number of cholera deaths per 10,000 households, stratified by water source, during the first 7 weeks of the epidemic. Note: In case you are wondering how t hey got the numbers in final column labelled as ‘Deaths per 10,000 households’ which are Cholera deaths per 10,000 households the formula they used was: No of deaths due to Cholera x 10,000 Number of households supplied Deaths due to cholera in househol ds supplied by S&V Company = (1263/40,046) x 10,000 = 315 Deaths due to cholera in households supplied by Lambeth Company = (98/26,107) x 10,000 = 38 Deaths due to cholera in households for rest of London = (1422/256,423) x 10,000 = 55 a. How many death s happened in households supplied by S&V, Lambeth and rest of London (calculate separately for each of the three and express as percentage with 2 decimals) Deaths due to cholera in households supplied by S&V Company = (1263/40,046) x 100 = 3.15% Deaths due to cholera in households supplied by Lambeth Company = (98/26,107) x 100 = 0.38 Deaths due to cholera in households for rest of London = (1422/256,423) x 100 = 0.55 Please note there are more than one ways of doing this (as you can see below) but I sugge st going with the original data (information given as number of Cholera deaths and number of households supplied), rather than using the final column, number of deaths per 10,000 households because some rounding up or down have been done in this column. Using a calculated answer from a previous step may sometimes give you slightly different final answer (not in this case though). I still prefer if you go with the method shown above rather than the one below) Deaths due to cholera in houses supplied by S&V Company = (315/10,000) x 100 = 3.15% Deaths due to cholera in houses supplied by Lambeth Company = (38/10,000) x 100 = 0.38% Deaths due to cholera in houses supplied by S&V Company = (59/10,000) x 100 = 0.55% b. Do the answers in ‘ part a’ support Snow’s hypothesis that cholera is transmitted through water? Give reasons for your answers. 3 Yes the above data supports Snow’s hypothesis that cholera is transmitted through water but they do not prove it. For example no information on other possible modes of transmission is included. c. Are these data adequate to conclude that cholera mortality is higher in houses supplied by the S&V Company than in houses supplied by the Lambeth Company? Give reasons for your answer. No. You can see that the number of deaths were 8.29 times more in households supplied by S&V Company in comparison to Lambeth Company (3.15/0.38 = 8.29) and 5.73 times compared to the rest of London (3.15/0.55 = 5.73 ). Although the data suggest that cholera deaths were higher in households supplied by S&V Company but they do not prove the causality and are not adequate to make firm conclusions that cholera mortality is higher in households supplied by S&V Company. d. What further questions you might ask before reaching any conclusions based on these data? Before reaching any conclusions we should consider whether the number of people per household, their socio-economic status, and other potential factors associated with the risk of cholera are comparable (similar) between the populations we are comparing. For example, the S&V Company might have supplied water to multiple occupancy buildings while Lambeth supplied individual family houses. If this were the case, then the risk of cholera death per house between these populations will not be comparable since the average number of people per house would differ between them. Also since S&V Company was drawing water from downstream, it is possible that the households supplied by the company would have been in downstream areas and might be poorer than households upstream. Therefore, although the data appear to support Snow’s hypothesis, more information is needed to be convincing. Q5. What is the name of the following graph? Write a short interpretation for the following graph (what information is being conveyed) . What do the ‘error bars’ represent and what is their importance ? Figure1: Prevalence of self -reported diabetes by age group and gender (dark bands males & light for females) 4 It is a ‘Bar chart’ which is suitable for displaying categorical data (number of people with diabetes per 100 in the population within each gender and age group) . Males have higher prevalence of self -reported diabetes than females in all age groups and the prevalence of diabetes increases with age. Error bars represent the 95% confidence intervals (a confidence interval reflects the uncertainty of the population estimate) and the height of each bar shows the accuracy of prevalence estimates. If 95% CIs for any two groups overlap it means there is no real (or no statistically signi ficant difference) in prevalence between the two groups. In the above graph 95%CIs for males and females overlap for all the age groups therefore we can say that though prevalence of self -reported diabetes is slightly higher among males but this difference is due to some random chance and there is no statistically significant difference in prevalence among males and females. Q 6a. Fill in the blanks: The two main types of data are Categorical (i.e. Nominal & Ordinal) and Continuous (Interval and Ratio). Q 6b. Label the following variables (or data) with correct measurement scale ( Nominal (including Binary/dichotomous, having two values or categories only), Ordinal , Interval or Ratio). For those who need a refresher p lease read on: A measurement scale refers to how data or a variable of interest has been measured ( Why you need to know on what scale a variable has been measured ? Because it determines what sorts of analysis or statistical tests we can do with them) Nominal scale: Categories or groups with names only, no information regarding size or magnitude or relationship between categories e.g. name of the units you are studying this semester . It does not tell us much other than how many are in each category (percentage or proportion) . Ordinal scale: categories with some form of ‘order’ or ‘ranks’ in terms of size or magnitude (we can say one is bigger than the other) but the interval or difference between one category and the next is not the same e.g. units you are studying as to ‘dislike, like a little, like a lot’ we can make some sense here but we can’t really say how much exactly you like or dislike and how different one category is from another in terms of size or magnitude. A nother example may be ‘ disease severity’ measured as (mild, moderate and severe) we know t he disease is getting worse from mild to severe but the difference between mild and moderate, and between moderate to severe is not known or equal. Interval scale: Magnitude or size can be measured precisely and ‘intervals’ are equal and can be identified e.g. we can say A - B = B - C, if A, B, and C are IQ scores and A = 150, B=100 and C=50 then it will be true that A - B = B – C however, we cannot say that A=3C or A is three times more intelligent than C . Interval scale does not have meaning ful zero ( an IQ of zero does not mean complete absence of intelligence rather it simply means that a person has severe intellectual deficit and cannot cope with the procedure or the tests for measuring the IQ ). Another example to clarify equal intervals is e.g. if Claudia, Sam and Hamish weigh 60kg, 66kg and 72kg respectively 6kg difference is ‘same’ whether it is lighter or heavier i.e. difference in terms of body weight is same between Claudia and Sam as well as between Sam and Hamish. Remember in Interva l 5 scale ‘zero’ does not mean ‘complete absence of that characteristic’ e.g. 0 degree Centigrade does not mean complete absence of heat, as temperature can go down below zero such as - 5 or -40 etc . (therefore in statistics lingo we say ‘zero is not absolute ’) Ratio scale: It has ALL the properties Interval scale along with a meaningful or ‘absolute zero’, so a value of zero on ratio scale actually means ‘complete absence of that characteristic’ e .g. zero number of heart beats per minute would actually mean ‘no heart beat’  therefore it is a variable with ‘Ratio scale’ so is ‘Money in your wallet’ now if ‘money in your valet = 0 it will actually mean there is complete absence of money (yes no coins either  ). Following table shows all four scales and their characteristics/properties. Scale/Level of Measurement Characteristic Nominal Ordinal Interval Ratio Distinctiveness x x x x Ordering in magnitude x x x Equal Intervals x x Absolute zero x Distinctiveness = Different numbers (labels) assigned to different properties Ordering in magnitude = Larger values represent more of the property Equal intervals = same distance between points on a scale. Absolute zero = I think you already know this now.  Let’s see if the following answers make perfect se nse: a. The number of heart beats per minute: Ratio. b. Platform numbers at a railway station: Nominal. c. Finishing order in a horse race: O rdinal. d. Self -rated anxiety levels on a five- point scale: Ordinal. e. Assignment 1 marks out of 20: Ratio. f. Subjects’ height in centimeters: Interval. g. Body weight measured in kilograms: Interval h. Percentage change in serum cholesterol levels: Ratio (as zero means ‘no change’). i. Patient identification number: Nominal . j. Weight measured as percentage overweight in relation to healthy weight: Ratio k. Subject status after 2 years of follo w up measured as dead or alive: ( Binary/dichotomous: a nominal variable with 2 categories only) Please note: I f ‘k’ was a multiple choice question, between the two choices ‘Nominal’ and ‘Binary/dichotomous’ the correct answer will be ‘Binary /dichotomous’ since it is more precise/detailed than ‘Nominal’. Q 7. Describe the 3 levels of prevention with an example each. 1) Primary prevention: denotes an action taken to prevent the development of a disease in a healthy person not having the disease in question. Or, if a disease is environmentally induced, we can prevent a person’s exposure to such environment to prevent the development of the condition. For example, we can immunize a person against certain diseases . By getting people to stop smoking we can prevent 70- 80% of lung cancer in human beings. Primary prevention is our ultimate goal. 6 2) Secondary prevention: denotes the identification of people with disease at an early stage in the disease’s natural history, through screening and early intervention. For example, breast cancer screening , colon cancer screening (occult blood test). The rationale is that if disease can be identified earlier, the intervention measures would be more effective, less costly and the m ortality or complications could be significantly reduced. 3) Tertiary prevention: Reduction of disability and promotion of rehabilitation e.g. surgical intervention to prevent a complication (amputation of lower limb in case of gangrene in a diabetic patient to prevent septicemia and death).