Epidemiological Literature PDF

ERIC NOTEBOOK SERIES Second Edition Common Measures and Statistics in Epidemiological Literature...

ERIC NOTEBOOK SERIES Second Edition Common Measures and Statistics in Epidemiological Literature For the non-epidemiologist or non- health outcome over a 5-year period Second Edition Authors: statistician, understanding the of time. statistical nomenclature presented Lorraine K. Alexander, DrPH in journal articles can sometimes be Risk is generally measured in prospective studies as the population challenging, particularly since Brettania Lopes, MPH at risk can be defined at the start of multiple terms are often used the study and followed for the interchangeably, and still others are Kristen Ricchetti-Masterson, MSPH development of the health outcome. presented without definition. This However, risk cannot be measured notebook will provide a basic Karin B. Yeatts, PhD, MS directly in case-control studies as the introduction to the terminology total population at risk cannot be commonly found in epidemiological defined. Thus, in case-control literature. studies, a group of individuals that Measures of frequency have the health outcome and a group of individuals that do not have the Measures of frequency characterize health outcome are selected, and the the occurrence of health outcomes, odds of developing the health disease, or death in a population. outcome are calculated as opposed to These measures are descriptive in nature and indicate how likely one is calculating risk. to develop a health outcome in a specified population. The three most common measures of health outcome or frequency are risk, rate, Rate and prevalence. A rate, also known as an incidence Risk rate or incidence density, is a Risk, also known as incidence, measure of how quickly the health cumulative incidence, incidence outcome is occurring in a population. proportion, or attack rate (although The numerator is the same as in risk, not really a rate at all) is a measure but the denominator includes a of the probability of an unaffected measure of person-time, typically individual developing a specified person-years. (Person-time is defined health outcome over a given period as the sum of time that each at-risk of time. For a given period of time individual contributes to the study). (i.e.: 1 month, 5 years, lifetime): A 5-year risk of 0.10 indicates that Thus a rate of 0.1 case/person-years indicates that, on average, for every an individual at risk has a 10% 10 person-years (i.e.: 10 people each chance of developing the given followed 1 year or 2 people followed ERIC at the UNC CH Department of Epidemiology Medical Center  ERIC NOTEBOOK PA G E 2 for 5 years, etc.) contributed, 1 new case of the health excess risk is due to the exposure of interest. A positive outcome will develop. risk difference indicates excess risk due to the exposure, while a negative result indicate that the exposure of Prevalence interest has a protective effect against the outcome. Prevalence is the proportion of a population who has the (Vaccinations would be a good example of an exposure health outcome at a given period of time. Prevalence is with a protective effect). This measure if often utilized to generally the preferred measure when it is difficult to define determine how much risk can be prevented by an onset of the health outcome or disease (such as asthma), or effective intervention. any disease of long duration (e.g. chronic conditions such as arthritis). A limitation of the prevalence measure is that it Risk ratio and rate ratio tends to favor the inclusion of chronic diseases over acute Risk ratios or rate ratios are commonly found in cohort ones. Also, inferring causality is troublesome with prevalence studies and are defined as: the ratio of the risk in the data, as typically both the exposure and outcome are exposed group to the risk in the unexposed group or the measured at the same time. Thus it may be difficult to ratio of the rate in the exposed group to the rate in the determine if the suspected cause precedes the outcome of unexposed group interest. Thus a population with a heart disease prevalence of 0.25 indicates that 25% of the population is affected by heart disease at a specified moment in time. Risk ratios and rate ratios are measures of the strength of the association between the exposure and the outcome. A final note, risk and rates can also refer to deaths in a How is a risk ratio or rate ratio interpreted? A risk ratio of population and are termed mortality and mortality rate, 1.0 indicates there is no difference in risk between the respectively. exposed and unexposed group. A risk ratio greater than Measures of association 1.0 indicates a positive association, or increased risk for developing the health outcome in the exposed group. A Measures of association are utilized to compare the risk ratio of 1.5 indicates that the exposed group has 1.5 association between a specific exposure and health outcome, times the risk of having the outcome as compared to the They can also be used to compare two or more populations, unexposed group. Rate ratios can be interpreted the typically those with differing exposure or health outcome status, to identify factors with possible etiological roles in same way but apply to rates rather than risks. health outcome onset. Note that evidence of an association does not imply that the relationship is causal; the association A risk ratio or rate ratio of less than 1.0 indicates a may be artifactual or non-causal as well. Common measures negative association between the exposure and outcome of association include the risk difference, risk ratio, rate ratio in the exposed group compared to the unexposed group. and odds ratio. In this case, the exposure provides a protective effect. For example, a rate ratio of 0.80 where the exposed group received a vaccination for Human Papillomavirus Risk difference (HPV) indicates that the exposed group (those who received the vaccine) had 0.80 times the rate of HPV Risk difference is defined as compared to those who were unexposed (did not receive the vaccine). One of the benefits the measure risk difference has over the risk ratio is that it provides the absolute difference in risk, information that is not provided by the ratio of the two. A risk ratio of 2.0 can imply both a doubling of a very small or large risk, and one cannot determine which is the The risk difference, also know as the attributable risk, provides case unless the individual risks are presented. the difference in risk between two groups indicating how much ERIC at the UNC CH Department of Epidemiology Medical Center  ERIC NOTEBOOK PA G E 3 Odds ratio confidence interval of 1.7 to 4.0, we would reject the null hypothesis. The alternative hypothesis could be Another measure of association is the odds ratio (OR). The expressed in two ways: 1) children of smoking mothers formula for the OR is: will have either a higher or lower incidence of asthma than other children, or 2) children of smoking mothers will only have a higher incidence of asthma. The first alternative hypothesis involves what is called a "two-sided test" and is used when we simply have no basis for predicting in which direction from the null value exposure is likely to be The odds ratio is used in place of the risk ratio or rate ratio associated with the health outcome, or, in other words, in case-control studies. In this type of study, the whether exposure is likely to be beneficial or harmful. The underlying population at risk for developing the health second alternative hypothesis involves a "one-sided test" outcome or disease cannot be determined because and is used when we have a reasonable basis to assume individuals are selected as either diseased or non- that exposure will only be harmful (or if we were studying diseased or as having the health outcome or not having a therapeutic agent, that it would only be beneficial). the health outcome. An odds ratio may approximate the risk ratio or rate ratio in instances where the health Measures of significance outcome prevalence is low (less that 10%) and specific The p-value sampling techniques are utilized, otherwise there is a tendency for the OR to overestimate the risk ratio or rate The "p" value is an expression of the probability that the ratio. difference between the observed value and the null value has occurred by "chance", or more precisely, has occurred The odds ratio is interpreted in the same manner as the simply because of sampling variability. The smaller the risk ratio or rate ratio with an OR of 1.0 indicating no "p" value, the less likely the probability that sampling association, an OR greater than 1.0 indicating a positive variability accounts for the difference. Typically, a "p" association, and an OR less than 1.0 indicating a negative, value less than 0.05, is used as the decision point, or protective association. meaning that there is less than a 5% probability that the The null value difference between the observed risk ratio, rate ratio, or odds ratio and 1.0 is due to sampling variability. If the "p" The null value is a number corresponding to no effect, that value is less than 0.05, the observed risk ratio, rate ratio, is, no association between exposure and the health or odds ratio is often said to be "statistically significant." outcome. In epidemiology, the null value for a risk ratio or However, the use of 0.05 as a cut-point is arbitrary. The rate ratio is 1.0, and it is also 1.0 for odds ratios and exclusive use of "p" values for interpreting results of prevalence ratios (terms you will come across). A risk epidemiologic studies has been strongly discouraged in ratio, rate ratio, odds ratio or prevalence ratio of 1.0 is the more recent texts and literature because research on obtained when, for a risk ratio for example, the risk of human health is not conducted to reach a decision point disease among the exposed is equal to the risk of disease (a "go" or "no go" decision), but rather to obtain evidence among the unexposed. that there is reason for concern about certain exposures Statistical testing focuses on the null hypothesis, which is or lifestyle practices or other factors that may adversely a statement predicting that there will be no association influence the health of the public. Statistical tests of between exposure and the health outcome (or between significance, (such as p-values) were developed for the assumed cause and its effect), i.e. that the risk ratio, industrial quality-control purposes, in order to make a rate ratio or odds ratio will equal 1.0. If the data obtained decision whether the manufacture of some item is from a study provide evidence against the null hypothesis, achieving acceptable quality. We are not making such then this hypothesis can be rejected, and an alternative decisions when we interpret the results of research on hypothesis becomes more probable. human health. For example, a null hypothesis would say that there is no The lower bound of the 95% confidence interval is also association between children having cigarette smoking often utilized to decide whether a point estimate is mothers and the incidence of asthma in those children. If statistically significant, i.e. whether the measure of effect a study showed that there was a greater incidence of (e.g. the ratio 2.5 with a lower bound of 1.8) is statistically asthma among such children (compared with children of different than the null value of 1.0. nonsmoking mothers), and that the risk ratio of asthma among children of smoking mothers was 2.5 with a 95% ERIC at the UNC CH Department of Epidemiology Medical Center  ERIC NOTEBOOK PA G E 4 Measures of precision of deaths in a calendar year divided by the average population for that year. This may be an appropriate Confidence interval measure in certain circumstances but could become problematic if you want to compare two or more A confidence interval expresses the extent of potential populations that vary on specific factors known to variation in a point estimate (the mean value or risk ratio, contribute to the death rate. For example, you may want rate ratio, or odds ratio). This variation is attributable to to compare the death rate for two populations, one of the fact that our point estimate of the mean or risk ratio, which is located in a high air pollution area, to determine if rate ratio, or odds ratio is based on some sample of the air pollution levels affect the death rate. The high air population rather than on the entire population. pollution population may have a higher death rate, but you For example, from a clinical trial, we might conclude that a also determine that it is a much older population. As new treatment for high blood pressure is 2.5 times as older individuals are more likely to die, age may be driving effective as the standard treatment, with a 95% the death rate rather than the pollution level. To account confidence interval of 1.8 to 3.5. 2.5 is the point estimate for the difference in age distribution of the populations, we obtain from this clinical trial. But not all subjects with one would want to calculate an adjusted death rate that high blood pressure can be included in any study, thus the adjusts for the age structure of the two groups. This estimate of effectiveness, 2.5, is based on a particular would remove the effect of age from the effect of air sample of people with high blood pressure. If we assume pollution on mortality. that we could draw other samples of persons from the Adjusted estimates are a means of controlling for same underlying population as the one from which confounders or accounting for effect modifiers in analyses. subjects were obtained for this study, we would obtain a Some factors that are commonly adjusted for include set of point estimates, not all of which would be exactly gender, race, socioeconomic status, smoking status, and 2.5. Some samples would be likely to show an family history. effectiveness less than 2.5, and some greater than 2.5. The 95% CI is an interval that will contain the true, real (population) parameter value 95% of the time if you Practice Questions repeated the experiment/study. So if we were to repeat the experiment/study, 95 out of 100 intervals would give Answers are at the end of this notebook. an interval that contains the true risk ratio, rate ratio or odds ratio value. Remember, that you can only interpret 1. Based on the following table, calculate the requested the CI in relation to talking about repeated sampling. Thus measures. Also provide the definition for each measure in we can also say that the new treatment for high blood one sentence. pressure is 2.5 times as effective as the standard treatment, but this measure could range from a low of 1.8 a) The risk ratio comparing the exposed and the to a high of 3.5. unexposed study participants The confidence interval also provides information about b) The risk difference between the exposed and the how precise an estimate is. The tighter, or narrower, the unexposed study participants confidence interval, the more precise the estimate. c) The prevalence of the disease among the entire study Typically, larger sample sizes will provide a more precise estimate. Estimates with wide confidence intervals should sample, assuming the disease is a long-term, chronic be interpreted with caution. disease with no cure and assuming no study participants have died. Other terms Crude and adjusted values Has Does not Total disease have There are often two types of estimates presented in disease research articles, crude and adjusted values. Crude Exposed 651 450 1101 estimates refer to simple measures that do not account for other factors that may be driving the estimate. For Unexposed 367 145 512 instance, a crude death rate would simply be the number Total 1018 595 1613 ERIC at the UNC CH Department of Epidemiology Medical Center  ERIC NOTEBOOK PA G E 5 Acknowledgement 2. Interpret the following risk ratios in words. The authors of the Second Edition of the ERIC Notebook would like to acknowledge the authors of t he ERIC N ot eb ook, Firs t Edit ion: Michel Ib rahim , MD, PhD, a) A risk ratio= 1.0 in a study where researchers Lorraine Alexander, DrPH, Carl Shy, MD, DrPH and examined the association between consuming a Sherry Farr, GRA, Departm ent of Epidemiology at certain herbal supplement (the exposure) and t he Univers it y of N ort h Carolina at Chapel Hill. The First Edit ion of the ERIC Notebook was produced b y developing arthritis. t he Educat ional Arm of t he Epidem iologic Res earch b) A risk ratio= 2.6 in a study where researchers and Informat ion Cent er at Durham , N C. The funding examined the association between ever having texted for the ERIC N ot eb ook First Edition was provided while driving (the exposure) and being in a car b y the Departm ent of Vet erans Affairs (DV A), V et erans Healt h Administ rat ion (V HA), Cooperative accident. St udies Program (CSP) to prom ot e the s t rat egic c) A risk ratio = 0.75 in a study where researchers growt h of the epidemiologic capacit y of t he DV A. examined the association between ≥ 30 minutes of daily exercise (the exposure) and heart disease. Answers Continued References 1c) Prevalence= Total # people with the disease / total # of people in the study population = 1018/1613 = 0.63 Dr. Carl M. Shy, Epidemiology 160/600 Introduction to Prevalence refers to the proportion of the population Epidemiology for Public Health course lectures, 1994- studied that has the disease at a given time. 2001, The University of North Carolina at Chapel Hill, Department of Epidemiology 2a) A risk ratio of 1.0 means there is no difference in risk for the health outcome when comparing the exposed and Rothman KJ, Greenland S. Modern Epidemiology. Second unexposed groups, i.e. the herbal supplement was not Edition. Philadelphia: Lippincott Williams and Wilkins, associated in any way with the development of arthritis 1998. 2b) A risk ratio of 2.6 means there is a positive The University of North Carolina at Chapel Hill, Department association, i.e. there is an increased risk for the health of Epidemiology Courses: Epidemiology 710, outcome among the exposed group when compared with Fundamentals of Epidemiology course lectures, 2009- the unexposed group. The exposed group has 2.6 times 2013, and Epidemiology 718, Epidemiologic Analysis of the risk of having the health outcome when compared with Binary Data course lectures, 2009-2013. the unexposed group. In this example, the risk ratio of 2.6 means that people who had reported ever texting while Answers to Practice Questions driving had 2.6 times the risk of being in a car accident 1.a) Risk ratio= risk exposed / risk unexposed = when compared with people who reported never having (651/1101 ) / (367/ 512) = 0.82 texted while driving. The risk ratio reflects the ratio of the risk of the disease in 2c) A risk ratio of 0.75 means there is an inverse the exposed study participants compared with the risk of association, i.e. there is a decreased risk for the health the disease in the unexposed study participants. outcome among the exposed group when compared with the unexposed group. The exposed group has 0.75 times 1b) Risk difference = risk exposed - risk unexposed = the risk of having the health outcome when compared with (651/1101 ) - (367/512 ) = -0.13 the unexposed group. In this example, the risk ratio of 0.75 means that people who exercised at least 30 The risk difference indicates how much excess risk is due minutes per day had 0.75 times the risk of developing to the exposure studied. heart disease when compared with people who did not exercise at least 30 minutes a day. ERIC at the UNC CH Department of Epidemiology Medical Center