public health

F:\VEB\teach\osu\h100\winter 2013\h100 first slide banner.gif

H100, Introduction to Public Health

Epidemiology exercise, fall 2017

Due: 11 October, uploaded to Canvas through assignment function

Use the tables and response boxes for answers; the first two pages are background—no responses needed

Introduction

This assignment is designed to allow you to employ the basic tools of epidemiology, using a real example that has elements which cross public health, child development, exercise and physical activity, and nutrition, as well as other fields.

As we discussed in class, and as you have learned through your readings and online material, epidemiology concerns itself with the “distribution and determinants” of health related matters in populations, with the goal of using the findings to improve health and prevent or cure disease. Epidemiology is most typically thought of as applying to large geographically defined populations (e.g. a state or country), and dealing with death and disease (e.g. number of people contracting influenza).

The tools of epidemiology, though, can be applied in any setting where there is a) a defined population “at risk” for some event, and b) a count of the number of events that occur in that population in a given time. For example, it is possible to treat passing H100 as an epidemiologic problem. The population is everyone in the course, and the event is passing the course. To calculate prevalence of passing, therefore, you simply make a fraction of those passing divided by those enrolling. If 340 students of 350 enrolled pass, the prevalence of passing is 97.1% (i.e. 340/350).

In this exercise, you will use real data from the American College Health Association report on As part of the exercise, you will calculate measures of frequency (e.g. prevalence) and of association (e.g. relative risk). We will go over ACHA and the NCHA in class. Before you do the exercise yourself, I suggest you review the class materials on epidemiology. For those of you with greater interest in the topic of college student health, the link to ACHA is on the Canvas site and embedded in hyperlinks here.

Background

Because we will have spent time in class on ACHA/NCHA, this background will be specific to the assignment. Sample surveys are available at the ACHA website, as are results from recent surveys.

Data

We will be using data abstracted from the 2015 NCHA survey. Below are the characteristics of the institutions participating in that effort. You should refer to the information below when answering the question about confounding and bias.

With the increased availability of opiates, both prescription and otherwise, their use has become a concern for campus leaders across the US. During the following exercise, you will explore what proportion of US college students report using opiates, and comment on what action, if any, is justified to address potential public health concerns.

Exercise

The following activities are designed to be conducted in order—answers from the first section feed into items in the second, and so on. Sections 1 and 2 ask you to conduct calculations similar to those discussed in class, in narrated slides, and in your text. Section 3 asks you to think about ways to interpret the results, including the possibility that they are misleading, and what you might do with the information.

Section 1. Prevalence

Prevalence is simply the proportion of those in a given population which have the condition of interest at the time of assessment. For instance, if 23 of 350 H100 class members have colds at our first class meeting, the prevalence (point prevalence, to be technical) is 23/350, or 6.6/100 or 6.6%. In this section, you will calculate the prevalence of strength training for college men and women, using the data table provided to you below.

1. NCHA participants were asked whether they had ever used opiates and if so, how recently. The numbers in the table below divide respondents into “ever used” for simplicity. Using those numbers, calculate the prevalence of opiate use for men, women, and overall, and enter your results in the table below. Prevalence should be reported to three decimal places (e.g. 0.094), rather than as a percentage, though you can interpret results in item 2 as percentages.

Group Ever used opiates Participants Prevalence
Men

Women

141

98

5150

10952

Total 239 16,102

2. In your own words, and in no more than a couple of sentences or bullets, provide a summary of the prevalence pattern you see here; in other words, what is the overall prevalence and does it seem to differ for women and men? Don’t worry for now about the interpretation or implications of the patterns you see (i.e. whether there is some statistical or causal association between gender and opioid use).

Section 2. Relative risk

Using the methods illustrated in the narrated slides, in class, and in your text, calculate the relative risk below using the prevalence estimates you calculated above (show the fractions you create as well as the relative risk). The prevalence estimates from the table above are your best estimates of absolute risk for each gender category; in this instance, it is the absolute risk of ever using opiates. (“Risk” is used here in the epidemiologic sense rather than common usage—“risk” does not necessarily refer only to health outcomes, but for any event; in this case we are dealing with the “risk” of opiate use.)

1. Using the prevalence for opiate use in women as the denominator, calculate the relative risk of opiate use for men relative to women. So, for the table below, you will calculate one relative risk. In the center column, provide the fraction you use to calculate relative risk (i.e. the two numbers you use to calculate RR). In the right column, provide the relative risk that comes from that fraction, to one decimal place.

Measure Fraction Relative risk
RR for men vs women

2. Using your own words, how do you quantitatively interpret the results? That is, how would you translate the relative risk for someone who didn’t know the term “relative risk”? You must interpret the actual relative risk you found (i.e. as a number), not just comment about how to interpret relative risks generally, and not just qualitatively (i.e. not just saying it is big or small or some such).

Section 3. Interpretation and action

As our session on epidemiology emphasized, getting a result is only part of conducting an epidemiologic study. The results must be interpreted correctly to be of use. One possible interpretation of results is that they are causal—that is, that the predictor (or exposure) causes the outcome (or event). For instance, you may look at the results on gender and opiate use and determine that the association is somehow causal (i.e. that gender—in the broadest definition, including social definitions—is really associated with opiate use via some causal mechanism). You then have the task of explaining how that could be so—what cause or causes lead to the connection? There are other reasons that two variables (e.g. gender and reported opiate use) might be associated but without one causing the other. A third variable might be leading to opitate use, but also be associated with gender, for instance. A variable statistically associated with gender may be the cause of opiate use, and not gender itself (“confounding”). Or, it could be that the students in the survey were either sampled in a way that led to misleading results, or there were problems with the way the survey collected data that led to misleading results (i.e. bias); you may not know this to be a fact, but you may speculate about it.

Your task in this section is to look at the results, and try to decide whether you think they are causal, that is, do you think that gender (or a component of gender) actually influences, in some way, the likelihood of opiate use? If you think it is causal, explain how—if not, give a plausible alternative to causation.

1. Do you think gender is somehow causally related to opiate use? Answer one of the following below, depending on how you want to explain any association you see in the results.

a. If yes, how could gender be causally associated with opiate use? Explain how, in a way consistent with the results you found.

b. If no, how might the results be the result of confounding or bias (i.e. the result of a third variable or a problem with the way the survey was conducted)? Provide one explanation; again, be sure your answer is consistent with the data and results you found earlier.

Epidemiology is an applied science. Ultimately, the goal of the field is to find results that allow public health organizations (and others) to take action to improve health. Based on everything you have calculated and reported for this exercise, in the box below briefly (2 bullets or sentences) say 1) whether you would recommend action based on these results, and 2) what that action would be?

2. Would you recommend to policy makers and leaders (e.g. university presidents) that they take public health action based on these results? If so, what? If not, what would make you unwilling to take action at this point? Be sure your responses are consistent with the data and results, but also take into account larger issues of ethics, policy, and so on.

Well done! This is what epidemiologists, along with other public health, clinical, and other professionals do all the time—take data, try to gain insights into what determines health-related outcomes, and determine whether public health action should be recommended. If you want to learn more about how these college surveys are done, visit the ACHA website; there is a lot out there on the epidemic of opioid use, and the CDC is a good place to start.