Epidemiology is the study of distribution and determinants of diseases and health related states in populations and use of that knowledge in order to prevent illnesses and promote health of populations (cite??). The word "Epidemiology" is derived from "Epi" = upon, "demo" = population, and "logos" = discourse. Thus, from its source, Epidemiology would mean that it is the study of everything that is related to populations. Within the larger field of Epidemiology, environmental epidemiology is  a branch where the exposure variables are present in our environment. For example, if you were to conduct an epidemiological investigation of the relationship between air pollution and asthma, then this would be an environmental epidemiological study. Similarly, if you were to conduct a study on tassociatoin association between job stress and hypertension or diabetes, then this kind of study would be labelled as occupational epidemiological study. 
In this module, we will cover Environmental and Occupational Epidemiology along the following sequence:
We will focus on environmental factors as risk factors or exposures. 
Imagine an elderly woman slipping on an icy street and breaking her hip. What do you think would be the cause?
Sufficient causes are those models which incorporate a number of different possible causes within the same model and all component causes interact with each other. The causes interact with each other and result in the outcome that we discuss. The causes however, do not need to interact all at the same time.
// Sufficient causal models here
Figure 1. Sufficient causal models. Note how component causes, sufficient causal models, and necessary causes are illustrated in this model
Causal thinking is part of being human and most people have a notion of one-one notion of .cause and effect. According to Rothman (2014), an infant, for example, cries in order to get milk or attention from his or her mother and this is an example of an early formation of our caus thinking (citation??). Likewise, when a boy grows up, he thinks that the flipping of the switch _causes_ the bulb to glow. However, as we have
Causal thinking grew out of a long history. Initially, people used to rely on the words of the wise people. Hippocrates was an early era physician who also introduced the concept of environment and the need for accurate history  
Inductive thinking was the next step. Here, from particular events or facts, the scientist would deduce generalisable truths. For example, 
In about 1700s, a new line of thinking emerged, referred to as "deductive thinking" and falsification of hypothesis. David Hume and others argued that inductive logicians assumed that the initial conditions of inductive logic remained invariant. For example,
In falsification of hypothesis, we would start with a theory and then formulate a hypothesis. Instead of finding out facts to further substantiate my theory, we would look for counter-examples or examples or facts that would make the hypothesis wrong. For example, . If we are able to identify an example or an instance which refutes my theory, or a hypothesis derived from my theory, then that would disprove the theory itself. This would mean that we would have to search for another theory to explain the phenomenon.
The Vienna school of philosopher, Karl Popper, who was also an academic at the University of Canterbury, in 1934, provided an extension of the falsification of the hypothesis. He referred this as the theory of conjecture and refutation. According to the theories of conjecture and refutation, the scientist would frame a theory, and then, based on the theory, he would frame a number hypothesis. He would observe the facts first. He would then explain the fact by building a theory. There would be more than one theories to explain the fact. These theories may rival each other. Then, he would gather or search for "facts" that would falsify the hypotheses. The hypothesis that would remain would explain the phenomenon unless or until another alternative explanatory theory and a fact would be found to falsify it.
In health sciences, we use the theories of conjecture and refutation by framing null and alternative hypotheses. The alternative hypotheses are the conjecture based hypotheses which we use to explain facts and using the theories, we set up predictions or explanations. We use the null hypotheses to refute these hypotheses.
Table xxx. Null and Alternative Hypotheses
//Insert table here to show null and alternative hypotheses
As can be seen in Table xxx, type I errors are those situations where we have falsely rejected the null. Type II errors are those where we have falsely failed to reject the null hypothesis (also referred to as beta errors).
In establishing a cause and effect framework, we have discussed that we can use a model of sufficient and necessary causes. We have also discussed that component causes interact. In order to establish cause and effect relationships, we would also have to ensure that the associations would need to be valid. We would lable any association between X and Y as an internally valid association if it meets the following three criteria
  1. We can rule out the play of chance
  2. We can eliminate any bias 
  3. We can control for confounding variables
We can rule out chance by specifying the alpha and beta errors so that in the beginning of a study, we would acknowledge that the possibility of random error, so that even if we would have rejected the null hypothesis based on what we have observed either on the basis of how the events occurred in nature or on the basis of our experiments, the error would be less than 5%. This is conventional. Then, after the collection of data, we would test the likelihood of occurrence of the observations if the null hypotheses would be true. If the probability of occurrence of the facts if the null hypothesis would be true were to be less than 5%, then we would conclude, on the basis of the observation of facts alone, that such occurrence could not have occurred due to chance alone. The lower the probability, the stronger would our claims be.
Ruling out chance alone would be insufficient to establish the internal validity of the association. We would also need to ensure that our observations were free of biases. In observational epidemiological studies, such biases could arise from the following sources:
We can only eliminate biases from studies at the stage of designing the study. We can carefully train the observers to collect data, so that we can eliminate selection or observation bias. We can also use objective measurements for the biological variables or outcomes we study, so that we can eliminate or minimise reporting bias or response bias. We can blind the investigators of the study so that we can eliminate selection bias or we can use strict protocols such that prior to the study the investigators would not be able to influence the selection of individuals.
Finally, we should control for confounding variables to ascertain that our association is a valid association. Confounding variable indicates a variable is associated BOTh with the exposure variable and the outcome variable but this variable does not come in the causal pathway between the exposure and the outcome (Figure xxxx)
// Insert here a diagram or figure that illustrates confounding variable
Figure xxxxx. Confounding variables
We can control for confounding variables either at the stage of planning the study or at the stage of data analysis. We must select confounding variables on the basis of concepts or what we know about the substantive association between the exposure and the outcome variable. For example, . The three ways in which we can control for confounding variables are:
Matching and restriction works during the planning stage of the study design. Stratified analysis and multivariate analysis works in the stage after the data collection. We will review these concepts in the sections on study design and data analyses.
In summary, establishment of a valid association through ruling out chance, eliminating biases, and controlling confounders is great, but this does not guarantee that the nature of such association is "causal". Not all associations are causal; for example, several studies are designed to examine if the association between X and Y can be explained by invoking biases. That is, would the results still hold if these biases were not to exist or offer alternative explanations?
You can partly appraise the nature of this association, that is, whether the association is one of cause and effect, by referring to a series of considerations that Sir Austin Bradford Hill introduced in 1965. Although these are referred to as "Hill's criteria", the use of the word "criteria" is a misnomer. These considerations are as follows:
Finally, when causes interact, they interact in a biological sense. While the component causes together make a sufficient causal mechanism, it is not mandatory that their strengths or contribution would add up to 100%. They may add up to more than 100%, in which case, it would mean that they have overlaps (some people would have more than one cause working simultaneously, for example smoking and exposure to asbestos dust for lung cancer); they can also add up to less than 100%, and then the rest of the factors might remain unknown. Genes, for example account for 100% of all cancers, so would environment as a risk factors. 
Measures in Epidemiology
Risks and Rates
Time-person
Incidence Rates
Prevalence
Incidence
Age-specific prevalence and incidence
Age-standardisation
Direct standardisation
Indirect standardisation
Application to an Environmental Health Situation
Standardised Mortality and Morbidity Ratio
Concepts of Determinants of Disease
Rule out chance
Eliminate Bias
Control for Confounding Variables
How to Rule out chance
Hypothesis testing
Sample Size and Power Calculation
How to Eliminate Bias
What is bias
Selection Bias
Response Bias
Must eliminate bias at the stage of study design
Role of questionnaires and piloting questionnaires
Objective measurements
Example of an objective measurement in an environmental health setting
Example of an objective measurement in an occupational health setting
Example of selection bias in environmental health
Example of Response bias in environmental health
How to Control for confounding variables
Stratified analysis
Example of a stratified analysis with gender as a confounding variable
Matching
Matched Case Control study: example in an enviromental health setting
Multivariate analysis
Example of how to control for confounding variables in an Environmental Health study
Causal Inference
Hill's Criteria
Counterfactual theories of Causation
Necessary and sufficient causes - component cause model
Rothman's Pie
Measures of Determinants of Diseases
Relative and Absolute Risks
Population Attributable Risk and PAR%
Example of PAR% in Environmental Health
Absolute Risk
Relative Risk
Odds and Odds Ratio
Hazard Ratio
Epidemiological Study Designs
Ecological Study
Example of ecological study design in environmental health
Case Series
Example of case series in environmental health: surveillance
Cross-sectional survey
Example of cross-sectional survey in an occupational setting
Case Control Study
Example of case control study in environmental Health: arsenic exposure and cancer
Prospective Cohort study
Example of a prospective cohort study in environmental health context
Retrospective Cohort study
Example of a retrospective cohort study in the context of occupational health
Systematic Review and Meta Analysis
Example of systematic reviews and meta analyses in environmental and occupational health
Advantages, disadvantages, and indications for different study designs
Summary 
References