WEBVTT 1 00:00:00.810 --> 00:00:02.790 Hello and welcome to module two 2 00:00:02.790 --> 00:00:04.563 of Environmental Public Health. 3 00:00:05.430 --> 00:00:07.890 Environmental and Occupational Epidemiology draws 4 00:00:07.890 --> 00:00:11.730 on the same tools as epidemiology generally. 5 00:00:11.730 --> 00:00:12.570 I understand that some 6 00:00:12.570 --> 00:00:15.660 of you have already taken epidemiology one, but for those 7 00:00:15.660 --> 00:00:18.780 of you who haven't, some of this material may be new. 8 00:00:18.780 --> 00:00:20.490 If you've already taken it, then this might be 9 00:00:20.490 --> 00:00:23.733 a good time to revisit some of the basics. 10 00:00:27.180 --> 00:00:29.760 There are a few keywords in this definition 11 00:00:29.760 --> 00:00:32.883 of epidemiology that I wanna go through really quickly. 12 00:00:33.900 --> 00:00:36.090 First, let's consider the distribution 13 00:00:36.090 --> 00:00:37.803 of the disease in a population. 14 00:00:39.030 --> 00:00:41.730 Distribution refers to who has the disease, 15 00:00:41.730 --> 00:00:45.990 which age groups, which sex and socioeconomic status. 16 00:00:45.990 --> 00:00:49.590 Where are cases of the disease geographically located? 17 00:00:49.590 --> 00:00:52.800 When did the cases of disease occur? 18 00:00:52.800 --> 00:00:55.950 These questions establish persons, place and time 19 00:00:55.950 --> 00:00:58.653 and make up what is known as descriptive epidemiology. 20 00:01:00.480 --> 00:01:04.140 Next, let's consider determinants of disease. 21 00:01:04.140 --> 00:01:07.020 Smoking is a determinant for lung cancer 22 00:01:07.020 --> 00:01:11.850 and sun exposure and sunburns are determinants for melanoma. 23 00:01:11.850 --> 00:01:13.320 These are often referred to 24 00:01:13.320 --> 00:01:16.083 as risk factors or causal factors. 25 00:01:18.270 --> 00:01:19.620 There are several different kinds 26 00:01:19.620 --> 00:01:22.230 of epidemiological studies. 27 00:01:22.230 --> 00:01:24.540 Descriptive studies describe the disease 28 00:01:24.540 --> 00:01:27.240 with regards to persons, place and time 29 00:01:27.240 --> 00:01:29.370 but they don't formally test any hypotheses 30 00:01:29.370 --> 00:01:31.923 about exposure and disease relationships. 31 00:01:33.600 --> 00:01:35.340 Ecological studies 32 00:01:35.340 --> 00:01:37.290 which are sometimes called correlational studies 33 00:01:37.290 --> 00:01:39.540 examine exposure-disease relationships 34 00:01:39.540 --> 00:01:41.340 at a population level. 35 00:01:41.340 --> 00:01:43.830 For example, a cancer epidemiologist might note 36 00:01:43.830 --> 00:01:45.270 that there's a higher rate of leukemia 37 00:01:45.270 --> 00:01:48.960 in the south of a city near a hazardous waste site. 38 00:01:48.960 --> 00:01:49.793 Does this mean 39 00:01:49.793 --> 00:01:52.290 that the hazardous waste site is causing leukemia? 40 00:01:52.290 --> 00:01:53.880 You really can't say because you don't know 41 00:01:53.880 --> 00:01:55.020 for certain whether the people 42 00:01:55.020 --> 00:01:58.470 with the risk factor are the same as those with the disease. 43 00:01:58.470 --> 00:01:59.940 You would need to do additional studies to 44 00:01:59.940 --> 00:02:02.130 assess this relationship. 45 00:02:02.130 --> 00:02:04.230 Ecological studies are particularly useful 46 00:02:04.230 --> 00:02:05.943 for hypothesis generation though. 47 00:02:07.490 --> 00:02:11.340 Etiological or analytical studies test a specific exposure 48 00:02:11.340 --> 00:02:14.190 disease hypothesis among individuals. 49 00:02:14.190 --> 00:02:17.493 They're divided into experimental and observational studies. 50 00:02:18.390 --> 00:02:20.040 Randomized control trials fall 51 00:02:20.040 --> 00:02:21.930 under the experimental study category 52 00:02:21.930 --> 00:02:24.810 and these are the gold standard for evidence 53 00:02:24.810 --> 00:02:27.390 and are frequently used in clinical medicine to determine 54 00:02:27.390 --> 00:02:29.343 if one treatment outperforms another. 55 00:02:30.180 --> 00:02:34.140 However, they're rarely used in environmental epidemiology 56 00:02:34.140 --> 00:02:36.180 because it isn't ethical to randomize people 57 00:02:36.180 --> 00:02:38.460 to environmental hazards. 58 00:02:38.460 --> 00:02:40.320 The exception to this is randomizing people 59 00:02:40.320 --> 00:02:43.620 to intervention trials in which a proposed method 60 00:02:43.620 --> 00:02:47.613 of reducing exposure or a protective factor is assessed. 61 00:02:49.020 --> 00:02:51.180 Observation studies are the most frequently used 62 00:02:51.180 --> 00:02:54.660 in environmental epidemiology, but they're subject 63 00:02:54.660 --> 00:02:58.590 to some confounding, and that's an inherent limitation. 64 00:02:58.590 --> 00:03:00.210 I'll walk through different types 65 00:03:00.210 --> 00:03:02.853 of observational studies in the next few slides. 66 00:03:04.410 --> 00:03:06.900 So first I wanna talk about cohort studies. 67 00:03:06.900 --> 00:03:10.440 Cohort studies are useful when separate groups 68 00:03:10.440 --> 00:03:14.193 of exposed and non-exposed individuals can be assembled. 69 00:03:15.150 --> 00:03:17.010 Some studies follow large cohorts 70 00:03:17.010 --> 00:03:19.650 of people over time to assess a variety 71 00:03:19.650 --> 00:03:22.260 of exposure-disease relationships. 72 00:03:22.260 --> 00:03:23.970 These are prospective studies 73 00:03:23.970 --> 00:03:25.350 in which the groups are assembled 74 00:03:25.350 --> 00:03:28.140 and then followed forward in time. 75 00:03:28.140 --> 00:03:29.580 These are typically expensive 76 00:03:29.580 --> 00:03:31.260 and take a long time to conduct 77 00:03:31.260 --> 00:03:33.510 because epidemiologists have to wait 78 00:03:33.510 --> 00:03:35.880 for the disease events to occur. 79 00:03:35.880 --> 00:03:38.400 However, they can provide more accurate information 80 00:03:38.400 --> 00:03:41.610 on exposure, especially if you're taking measurements 81 00:03:41.610 --> 00:03:43.473 at the start of the study. 82 00:03:45.240 --> 00:03:48.510 Cohort studies can also be done retrospectively. 83 00:03:48.510 --> 00:03:50.730 In that case, epidemiologists would assemble groups 84 00:03:50.730 --> 00:03:52.710 of people with a past exposure 85 00:03:52.710 --> 00:03:55.050 and assess their current health status. 86 00:03:55.050 --> 00:03:57.330 These types of studies are usually less expensive 87 00:03:57.330 --> 00:03:59.490 and don't take as long to conduct 88 00:03:59.490 --> 00:04:02.409 but sometimes some pieces of information may be missing 89 00:04:02.409 --> 00:04:04.653 like the exact exposure level. 90 00:04:05.700 --> 00:04:08.130 Cohort studies as a rule of thumb 91 00:04:08.130 --> 00:04:10.650 can be best for when the exposure is rare 92 00:04:10.650 --> 00:04:13.503 but the outcome or the disease is common. 93 00:04:15.450 --> 00:04:17.220 In contrast to cohort studies 94 00:04:17.220 --> 00:04:19.500 case-control studies tend to be good 95 00:04:19.500 --> 00:04:23.370 for when the exposure is common, but the disease is rare. 96 00:04:23.370 --> 00:04:26.940 So in a case-control study, epidemiologists select cases 97 00:04:26.940 --> 00:04:30.210 with a disease and cases without that disease 98 00:04:30.210 --> 00:04:33.810 and look back in time to assess prior exposure. 99 00:04:33.810 --> 00:04:36.300 Exposures are then compared between the two groups 100 00:04:36.300 --> 00:04:39.270 so between the cases and the controls to determine 101 00:04:39.270 --> 00:04:41.430 if a particular exposure is more common 102 00:04:41.430 --> 00:04:43.980 in those with the disease than those without. 103 00:04:43.980 --> 00:04:46.113 This results in an odds ratio. 104 00:04:47.580 --> 00:04:50.155 If the case control study is well designed 105 00:04:50.155 --> 00:04:51.930 it should give the same answer 106 00:04:51.930 --> 00:04:54.330 as if it were conducted as a cohort study. 107 00:04:54.330 --> 00:04:56.370 But it can be really difficult to design a case 108 00:04:56.370 --> 00:04:58.800 control study very well. 109 00:04:58.800 --> 00:05:00.720 Controls may not be representative 110 00:05:00.720 --> 00:05:02.610 of the general population. 111 00:05:02.610 --> 00:05:05.220 It may be difficult to determine past exposures 112 00:05:05.220 --> 00:05:06.480 with certainty 113 00:05:06.480 --> 00:05:08.970 and cases may remember past exposures differently 114 00:05:08.970 --> 00:05:10.470 than your control group does. 115 00:05:10.470 --> 00:05:12.663 This is referred to as recall bias. 116 00:05:15.210 --> 00:05:17.850 Cross-sectional studies measure exposure 117 00:05:17.850 --> 00:05:20.850 and disease at the same time. 118 00:05:20.850 --> 00:05:24.930 Since temporality can usually not be inferred 119 00:05:24.930 --> 00:05:27.510 this is considered a weaker study design 120 00:05:27.510 --> 00:05:30.003 than cohort or case-control studies. 121 00:05:30.870 --> 00:05:33.300 However, cross-sectional studies are often 122 00:05:33.300 --> 00:05:36.810 the only possible design for surveys. 123 00:05:36.810 --> 00:05:39.030 Findings from cross-sectional studies can then 124 00:05:39.030 --> 00:05:42.153 often indicate when a more rigorous study is warranted. 125 00:05:45.360 --> 00:05:47.490 So here is a graphical depiction 126 00:05:47.490 --> 00:05:51.270 of the different types of epidemiological studies. 127 00:05:51.270 --> 00:05:52.920 There are a number of different ways that you 128 00:05:52.920 --> 00:05:55.276 could depict how the studies are similar or different 129 00:05:55.276 --> 00:05:57.960 but I like this version and it might be helpful 130 00:05:57.960 --> 00:06:00.213 to those of you who are visual learners. 131 00:06:01.680 --> 00:06:06.400 This slide splits studies into observational studies 132 00:06:07.860 --> 00:06:09.603 and experimental studies. 133 00:06:10.860 --> 00:06:12.900 The observational studies are then split further 134 00:06:12.900 --> 00:06:17.370 into population-based studies and individual-based studies. 135 00:06:17.370 --> 00:06:20.040 The individual-based studies are typically 136 00:06:20.040 --> 00:06:21.693 stronger study designs. 137 00:06:23.226 --> 00:06:25.020 The individual based studies are split 138 00:06:25.020 --> 00:06:27.630 into descriptive and analytical studies. 139 00:06:27.630 --> 00:06:31.020 The descriptive studies have no comparison group 140 00:06:31.020 --> 00:06:34.980 and include things like case reports or case series. 141 00:06:34.980 --> 00:06:38.940 The analytical studies must have a comparison group 142 00:06:38.940 --> 00:06:43.650 and include the study design types that we just discussed. 143 00:06:43.650 --> 00:06:47.760 So you have cohort studies where groups are assembled based 144 00:06:47.760 --> 00:06:50.280 on exposure and then followed 145 00:06:50.280 --> 00:06:53.310 over time to determine whether those who are exposed 146 00:06:53.310 --> 00:06:54.720 and those who weren't exposed end 147 00:06:54.720 --> 00:06:56.120 up with the disease outcome. 148 00:06:57.120 --> 00:07:00.420 Case-control studies where cases have the outcome 149 00:07:00.420 --> 00:07:03.390 of interest and controls don't, and we look back 150 00:07:03.390 --> 00:07:06.340 in time to see if their exposures differ 151 00:07:07.260 --> 00:07:09.510 and then cross-sectional studies that look 152 00:07:09.510 --> 00:07:12.213 at exposure and outcome at the same time. 153 00:07:13.710 --> 00:07:15.900 On the other side of the slide, 154 00:07:15.900 --> 00:07:19.260 under experimental studies, we have randomized studies 155 00:07:19.260 --> 00:07:22.080 including randomized control trials which are often referred 156 00:07:22.080 --> 00:07:27.070 to as clinical trials and as well as non-randomized studies. 157 00:07:28.680 --> 00:07:31.920 This includes quasi-experimental, field trials 158 00:07:31.920 --> 00:07:33.510 or community trials 159 00:07:33.510 --> 00:07:36.780 and often we can't truly randomize people 160 00:07:36.780 --> 00:07:41.163 into two groups based on the practicalities of the study. 161 00:07:43.860 --> 00:07:44.850 So now that we've walked 162 00:07:44.850 --> 00:07:48.870 through the types of epidemiological studies 163 00:07:48.870 --> 00:07:53.340 let's do a quick self-check to see if we have 164 00:07:53.340 --> 00:07:56.823 fully understood the differences between these studies. 165 00:07:57.840 --> 00:07:59.670 This is the first self-check. 166 00:07:59.670 --> 00:08:01.770 Feel free to pause the video 167 00:08:01.770 --> 00:08:05.834 as you think about your answers for these questions. 168 00:08:05.834 --> 00:08:08.197 The first self-check reads: 169 00:08:08.197 --> 00:08:11.760 "A study in which children are randomly assigned to receive 170 00:08:11.760 --> 00:08:14.580 either a newly formulated vaccine 171 00:08:14.580 --> 00:08:17.730 or the currently available vaccine, and then 172 00:08:17.730 --> 00:08:20.760 are followed to monitor for side effects and effectiveness 173 00:08:20.760 --> 00:08:24.420 of each vaccine is an example of which type of study? 174 00:08:24.420 --> 00:08:26.547 Select all the answers that apply." 175 00:08:33.990 --> 00:08:36.063 So now I'm gonna show you the answers. 176 00:08:37.710 --> 00:08:40.923 A. experimental and E. clinical trial. 177 00:08:41.970 --> 00:08:44.130 So a study in which subjects are randomized 178 00:08:44.130 --> 00:08:46.680 into two intervention groups and then monitored to 179 00:08:46.680 --> 00:08:49.380 identify health outcomes is a clinical trial 180 00:08:49.380 --> 00:08:52.410 which is a type of experimental study. 181 00:08:52.410 --> 00:08:54.030 It's not a cohort study 182 00:08:54.030 --> 00:08:58.113 because that term is limited to observational studies. 183 00:09:01.280 --> 00:09:04.567 Okay, let's try another one. 184 00:09:04.567 --> 00:09:06.840 "The Iowa Women's Health study 185 00:09:06.840 --> 00:09:10.830 in which researchers enrolled 41,837 186 00:09:10.830 --> 00:09:15.780 women in 1986 and collected exposure and lifestyle 187 00:09:15.780 --> 00:09:18.240 information to assess the relationship 188 00:09:18.240 --> 00:09:21.210 between these factors and the subsequent occurrence 189 00:09:21.210 --> 00:09:24.930 of cancer is an example of which types of study? 190 00:09:24.930 --> 00:09:27.537 Select all the answers that apply." 191 00:09:33.750 --> 00:09:35.460 Okay, I'm gonna show you the answers now. 192 00:09:35.460 --> 00:09:37.503 So pause it if you aren't ready. 193 00:09:40.290 --> 00:09:43.830 So B. observational and C. cohort. 194 00:09:43.830 --> 00:09:45.540 So a study that assesses 195 00:09:45.540 --> 00:09:49.710 but doesn't dictate exposure and follows 196 00:09:49.710 --> 00:09:52.320 up with the participants to document their 197 00:09:52.320 --> 00:09:55.770 subsequent occurrence of disease is an observational 198 00:09:55.770 --> 00:09:57.150 cohort study. 199 00:09:57.150 --> 00:10:01.110 Remember in the last example where exposure was dictated 200 00:10:01.110 --> 00:10:02.820 because children were randomized 201 00:10:02.820 --> 00:10:06.630 into two separate groups to receive one vaccine or another. 202 00:10:06.630 --> 00:10:08.010 That would be an experimental 203 00:10:08.010 --> 00:10:11.790 but because this study assesses exposure 204 00:10:11.790 --> 00:10:13.440 and it doesn't dictate exposure 205 00:10:13.440 --> 00:10:15.303 that makes it an observational study. 206 00:10:18.690 --> 00:10:20.733 Alright, let's try another one. 207 00:10:21.637 --> 00:10:25.080 "British investigators conducted a study to compare measles 208 00:10:25.080 --> 00:10:28.980 mumps, rubella or MMR vaccine history 209 00:10:28.980 --> 00:10:32.280 among 1,294 children 210 00:10:32.280 --> 00:10:35.430 with pervasive development disorder." 211 00:10:35.430 --> 00:10:37.170 So that would include things like autism 212 00:10:37.170 --> 00:10:40.263 or what was once referred to as Asperger syndrome. 213 00:10:42.197 --> 00:10:43.860 "and 4,469 children 214 00:10:43.860 --> 00:10:45.697 without such disorders. 215 00:10:45.697 --> 00:10:48.960 "They found no association," just to be clear. 216 00:10:48.960 --> 00:10:50.100 But "this is an example 217 00:10:50.100 --> 00:10:54.537 of which types of study select all answers that apply." 218 00:11:00.630 --> 00:11:02.680 Okay, I'm gonna show you the answers now. 219 00:11:04.260 --> 00:11:07.770 B. observational and D. case-control. 220 00:11:07.770 --> 00:11:10.980 So this is a study in which the subjects were enrolled 221 00:11:10.980 --> 00:11:15.540 on the basis of having or not having the health outcome. 222 00:11:15.540 --> 00:11:18.570 So that makes it an observational case-control study. 223 00:11:18.570 --> 00:11:22.440 Remember, for case control studies, people are split 224 00:11:22.440 --> 00:11:27.090 up based on having the outcome, whereas in a cohort study 225 00:11:27.090 --> 00:11:29.193 the groups are split based on exposure. 226 00:11:34.020 --> 00:11:35.913 Alright, let's try another one. 227 00:11:36.847 --> 00:11:40.500 "A cohort study differs from a case-control study 228 00:11:40.500 --> 00:11:44.700 in that" A. subjects are enrolled or categorized 229 00:11:44.700 --> 00:11:48.090 on the basis of their exposure status in a cohort study 230 00:11:48.090 --> 00:11:50.163 but not in a case-control study. 231 00:11:51.300 --> 00:11:54.630 B. subjects are about their exposure status 232 00:11:54.630 --> 00:11:58.323 in a cohort study, but not in a case-control study. 233 00:11:59.190 --> 00:12:02.460 C. cohort studies require many years to conduct 234 00:12:02.460 --> 00:12:04.503 but case-control studies do not, 235 00:12:06.240 --> 00:12:09.210 or D. cohort studies are conducted to investigate 236 00:12:09.210 --> 00:12:10.590 chronic diseases. 237 00:12:10.590 --> 00:12:13.563 Case-control studies are used for infectious diseases. 238 00:12:19.020 --> 00:12:20.283 Here are the answers. 239 00:12:22.092 --> 00:12:25.530 A. So the key difference between a cohort 240 00:12:25.530 --> 00:12:28.170 and a case-control study is that in a cohort study 241 00:12:28.170 --> 00:12:31.140 subjects are enrolled on the basis of their exposure 242 00:12:31.140 --> 00:12:32.550 whereas in a case-control study 243 00:12:32.550 --> 00:12:34.140 subjects are enrolled on the basis of 244 00:12:34.140 --> 00:12:36.600 whether they have the disease of interest or not. 245 00:12:36.600 --> 00:12:38.700 So we just covered that in the last slide as well 246 00:12:38.700 --> 00:12:40.999 so hopefully we've really hammered it home. 247 00:12:40.999 --> 00:12:45.762 Both types of studies assess exposure and disease status. 248 00:12:45.762 --> 00:12:47.970 While some cohort studies have been conducted 249 00:12:47.970 --> 00:12:50.610 over several years, others 250 00:12:50.610 --> 00:12:52.860 especially when it's related to an outbreak 251 00:12:52.860 --> 00:12:57.090 can be conducted really rapidly even in a matter of days. 252 00:12:57.090 --> 00:13:00.900 Either type of study can be used to study all sorts 253 00:13:00.900 --> 00:13:03.840 of different health problems, including both infectious 254 00:13:03.840 --> 00:13:07.833 and chronic diseases or non-communicable diseases. 255 00:13:12.960 --> 00:13:16.837 Okay, here's the final self-check question. 256 00:13:16.837 --> 00:13:21.780 "A key feature of a cross-sectional study is that:" 257 00:13:21.780 --> 00:13:24.630 A. it usually provides information 258 00:13:24.630 --> 00:13:27.810 on prevalence rather than incidents. 259 00:13:27.810 --> 00:13:30.780 B. it is limited to health exposures 260 00:13:30.780 --> 00:13:33.633 and behaviors rather than health outcomes. 261 00:13:34.680 --> 00:13:37.920 C. it is more useful for descriptive epidemiology 262 00:13:37.920 --> 00:13:40.390 than it is for analytic epidemiology 263 00:13:41.580 --> 00:13:44.793 and D. it is synonymous with survey. 264 00:13:51.420 --> 00:13:52.773 So here are the answers. 265 00:13:53.850 --> 00:13:57.150 A, C and D all hold true. 266 00:13:57.150 --> 00:14:01.470 So a cross-sectional study or a survey provides a snapshot 267 00:14:01.470 --> 00:14:05.310 of the health of a population in one slice of time 268 00:14:05.310 --> 00:14:08.670 so it assesses prevalence rather than incidents. 269 00:14:08.670 --> 00:14:11.670 As a result, it's not as useful as a cohort study 270 00:14:11.670 --> 00:14:15.000 or a case-control study for analytic epidemiology 271 00:14:15.000 --> 00:14:17.790 but a cross-sectional study can still easily 272 00:14:17.790 --> 00:14:20.733 measure the prevalence of exposures and outcomes. 273 00:14:23.100 --> 00:14:24.780 So hopefully this has helped you get up 274 00:14:24.780 --> 00:14:28.353 to speed on the different types of epidemiological studies. 275 00:14:29.430 --> 00:14:31.440 Feel free to go back through these questions 276 00:14:31.440 --> 00:14:33.993 and test yourself again if you find it helpful.