Objectives Experimental Design Stat 1040 Chapters 1 and 2 Given the description of conducted research, Distinguish between a controlled experiment and an observational study. Identify the treatment group and the control group. Evaluate the use of a placebo. Evaluate the use of blind and double-blind procedures. Distinguish between longitudinal research and crosssectional research. Identify and describe confounding factors. Evaluate claims of causal relationships. The Last Mile https://podcasts.usu.edu/podcasts/onlinestat_1040_fair bourn/2010-05-10/chapter_1_video-video.mp4 https://www.youtube.com/watch?v=5bfxpdzwxla Controlled Experiments Chapter 1 How do you determine if a drug is effective? 1954 Polio Vaccine Trials Placebo Controls Size Rate Treatment (vaccine) 200,000 28 Control (placebo) 200,000 71 No Consent 350,000 46 Observed Controls (NFIP study) Size Rate Grade 2 (vaccine) 225,000 25 Grades 1 & 3 (control) 725,000 54 Grade 2 (no consent) 125,000 44 Rate is the number of polio cases per 100,000 in each group. 1
Consent Children with parental consent to participate were different from those without consent. No consent group Polio rate was smaller Generally from lower income strata More natural (early) immunity to polio Consent group Higher income Less natural (early) immunity to polio Higher incidence of polio Diagnosis Mild cases of polio resemble the flu. How does this influence a vaccine trial? Placebos and the Placebo Effect Randomization and Lexical Ambiguity What does the word random mean? What random means to a statistician. Why is it important? Image credit: http://xkcd.com/1462/ The Decline of Polio Vaccine Safety and Effectiveness United States 1952 1963 1979 1994 Worldwide 1988 2014 57,628 cases 396 cases Last recorded case US certified polio free 350,000 cases in 125 countries approximately 333 cases in 9 countries For almost all healthy humans, vaccines are very safe. Side effects are minimal. Herd immunity protects those who cannot be vaccinated. No medicine is perfect. Accidents and vaccine failures are tragic the diseases they prevent are MUCH worse. Learn to weigh the risks appropriately. http://www.polioeradication.org/dataandmonitoring/poliothisweek.aspx 2
RCDB Experiments Other types of controls Randomized Observed controls Controlled Historical controls Double-blind Example Scenario 1 Saw Palmetto First experiment: http://www.sciencedaily.com/releases/2006/02/0602 13102419.htm Second experiment: http://www.sciencedaily.com/releases/2011/09/1109 27161811.htm Popular herbal remedy fails to offer benefits Saw palmetto doesn't improve symptoms of enlarged prostate, study finds The Associated Press Updated: 5:34 p.m. ET Feb. 8, 2006 NEW YORK - A popular herbal pill used by millions of men doesn t reduce the frequent urge to go to the bathroom or other annoying symptoms of an enlarged prostate, a rigorous new study concludes. The yearlong research found the plant extract, saw palmetto, was no more effective than dummy capsules in easing symptoms for the 225 men in the study. The results contrast with previous research that showed it helped. This certainly sheds some doubt on whether the product s effective and suggests that it might not work, said the lead researcher, Dr. Stephen Bent, of the San Francisco VA Medical Center. Identify The new study recruited men over 49 with enlarged prostates who had moderate to severe symptoms. They took 160 milligrams of saw palmetto twice a day or similar-looking dummy capsules. At each visit, they filled out a symptoms survey and their urine flow was measured. After a year, there was no significant difference between the groups in symptom changes or other measures, the researchers reported. Bent said the dummy capsule was carefully designed to match the brown color, bitter taste and strong odor of the extract. At the end of the study, 40 percent in the saw palmetto group and 46 percent in the comparison group thought they were getting the extract, showing it was well disguised. Treatment and control group Use of a placebo/blind/double blind procedures Use of randomization 3
Example Scenario 2 http://www.today.com/id/31852963/ns/todaytoday_health/t/stub-your-toe-say-sh-youll-feelbetter/#.vj9iyp8wi http://journals.lww.com/neuroreport/abstract/2009/080 50/Swearing_as_a_response_to_pain.4.aspx Stub your toe? Say *&#! You ll feel better Shouting swear words has a powerful pain-killing effect, study shows British scientists have shown that swear words can have a powerful pain-killing effect, according to a new study published in the journal NeuroReport. The researchers originally thought that swearing would make pain worse by focusing a person s attention on the injury and its implications. To prove their hypothesis, they set up an experiment with 67 college students. The students were asked to plunge their hands into frigid 41-degree Fahrenheit water for as long as they could stand the pain. Half were told to repeat their favorite curse word while their hands were submerged. The other half were asked to repeat a neutral word describing a table, such as solid or brown, while keeping their hands under water. Then the whole experiment was repeated with the two groups switching types of word. Identify To the researchers surprise, the cursing group not only reported lower levels of pain, but also were able to keep their hands in the icy water longer. The men in the study, for example, were able to keep their hands in the water for an average of 190 seconds while swearing, but for only 140 seconds when uttering a neutral word. The difference was even more pronounced in women. While men s pain scores dropped by a point when they cussed, the women s dropped by almost two full points. Treatment and control group Use of a placebo/blind/double blind procedures Use of randomization Definition: Observational Study In an observational study, subjects actions determine whether they are in the treatment group or the control group. Researchers do not determine or influence what subjects do. Observational Studies Chapter 2 Classic Example Smoking and Lung Cancer Because subjects are not assigned to treatment and control using an objective chance procedure, differences can be caused by something other than the treatment. 4
Confounding factor When the treatment group and control group differ in some respect other than the treatment, this factor can be a confounding factor. Ultrasound and Low Birth weight Researchers demonstrated using a controlled experiment, that performing an ultrasound on a pregnant animal caused low birth weight. Variable A Confounding Factor Variable B An observational study in the 1980s found an association between ultrasound exposure and low birth weight in humans. A randomized experiment conducted near the same time found that babies in the ultrasound (treatment) group had slightly higher birth weights. What explains this difference? Confounding factor Cervical Cancer and Religion? Early 20th century researchers noted that cervical cancer was rare in Jewish and Muslim women. It was also rare in nuns. Ultrasound Exposure Low Birth Weight Religion Low cervical cancer rates in women Cervical Cancer and Sex Cervical Cancer and Circumcision Further research demonstrated that: Cervical cancer was common in female sex workers. It was rare in nuns, except for those who had been sexually active before entering the convent. It was more common in the second wives of men whose first wives had died from cervical cancer. Circumcised sex partners Low cervical cancer rates in women 5
Cervical Cancer and HPV Researchers deduced that cervical cancer could be caused by a sexually transmitted agent. In 1935, researchers found a relationship between Rabbit Papillomavirus and skin cancer in rabbits. Eventually demonstrated that the cancer was caused by Human Papillomavirus. In 2007 the FDA approved the first vaccine Gardasil. Control More lexical ambiguity A control (noun) is a subject in a study who did not get treatment. A controlled experiment (adj.) is research where investigators determine who will be in the control group and the treatment group. Investigators control for (verb) confounding factors in an observational study. Look at subgroups in the study and compare treatment and control groups Common Confounding Factors Age Gender Socio-economic status Education Sexual history Genetics Occupation Religion and so many more! Alzheimer s Disease Quick web search shows the following associations: Anesthesia Diabetes Genes Activation of immune cells in the brain Low calorie, low fat diet Busy brains, exercise, learning, sex, social activities good Worrying bad Nicotine? Yes, No, Maybe? Coffee/Caffeine Which are causes and which are confounding factors? Controlled Experiment Researchers decide which subjects are in treatment and control groups. Researchers impose some change in behavior on the participants. Confounding factors are controlled for using randomization. If done well and are replicated, can allow us to (cautiously) infer causation. Observational Studies Subjects assign themselves to the different groups, usually by their own choices or behavior. Establish associations between variables. Confounding factors are controlled for after data is collected. Do not imply causation unless multiple studies over a long period of time all give the same results. (Even then, be careful!) Longitudinal vs. Cross-Sectional Studies Longitudinal One group of subjects is followed over time. Subjects are compared to themselves at different points in time. Subjects must be measured/interviewed more than once. Cross-sectional Subjects are only measured once. Subjects are compared to each other, or to subjects from another study. Do not use to draw conclusions about what happens as individuals age, etc. 6
Example scenario 3 Breast-feeding enhances kids eyesight http://dukeandthedoctor.com/2010/01/breast-feeding-enhances-kids-eyesight/ SOURCE: American Journal of Clinical Nutrition, January 2007. Breast-fed children are significantly more likely to do well in measures of stereoscopic vision than are those who received formula during infancy, according to UK researchers. Our study, Dr. Atul Singhal of the Institute of Child Health, London, told Reuters Health, adds to the growing evidence that breast-feeding has long-term benefits for visual development. A higher concentration of the fatty acid DHA (docosahexaenoic acid) in breast milk than in formula has been proposed as one explanation for this effect, Singhal and colleagues note in the American Journal of Clinical Nutrition, and this has been a rationale for adding DHA to infant formula. To further investigate, the team studied 78 previously breast-fed and 184 previously formula-fed children ages 4 to 6 years who were followed prospectively from birth. In the first 6 months of life, the formula-fed group was randomly assigned to formula with or without supplementation with DHA or arachidonic acid. When they were tested, breast-fed children were significantly more likely to have greater stereo-acuity than did children in either formula-fed group. There were no significant differences between children who did or did not receive formula containing DHA. The researchers conclude that these findings support the hypothesis that breast-feeding benefits long-term stereoscopic development. Questions Does this study describe an observational study or a controlled experiment? What is the treatment group? What is the control group? Based on the information you have, is it a longitudinal study or a cross-sectional study? Carefully explain how socio-economic status could be a confounding factor in this research. 7