Study Designs PDF

1. STUDY DESIGNS Biostatistics-Medicine 2024-2025 Example #1 Risk of hip fracture in women https://bmcmedicine.biomedcentral.com/articles/10.1186/s12916- 022-02468-0 2 / 47 Example #2 Parenteral Vitamin C in Patients with Seve...

1. STUDY DESIGNS Biostatistics-Medicine 2024-2025 Example #1 Risk of hip fracture in women https://bmcmedicine.biomedcentral.com/articles/10.1186/s12916- 022-02468-0 2 / 47 Example #2 Parenteral Vitamin C in Patients with Severe Infection: A Systematic Review https://evidence.nejm.org/doi/full/10.1056/EVIDoa2200105 3 / 47 First steps of a study Define the question of interest Try to define the questions of interest for examples # 1 and 2 Decide which study design will be the most appropriate Think about what is the best design for these studies Determine: Setting and Study population Intervention Variables Main outcome Statistical analysis Limitations 4 / 47 Types of study design 5 / 47 From Grimes and Schulz, Lancet 2002; 359: 57-61. Experimental studies The aim of these studies is to compare two or more treatments or interventions, either therapeutic, preventive or palliative The experimental studies that include human beings are called clinical trials (assaigs clínics) Clinical trials evaluate the efficacy and safety of medications, medical devices or other interventions, by monitoring their effects on large groups of people Researchers control the experimental conditions 6 / 47 Experimental studies (cont.) Participants are randomly distributed in the study groups to ensure comparability in all characteristics related to the outcome of interest The subjects in the intervention group receive the treatment that is being evaluated, while the subjects in the control group receive the standard treatment or a placebo. These studies allow to test cause-effect relationships 7 / 47 Observational studies There is no intervention Subjects self-select to different risk factors or life styles Researchers observe 8 / 47 Observational studies: Main limitation Well-conducted observational studies provide a lower level of evidence than well-conducted experimental studies because of confounding Confounding occurs because risk factors tend to cluster, which makes it difficult to isolate the causative agent A new methodology named Causal Inference explains that under determined conditions an observational study can be treated as if it were an experimental study 9 / 47 Observational studies Can be classified as: Analytic / Descriptive Prospective / Retrospective Longitudinal / Cross-sectional 10 / 47 Descriptive/Analytic According to the objective Descriptive studies: Measure the frequency of diseases, risk factors or other health related variables Analytic studies: Try to identify determinants or causes of diseases 11 / 47 Prospective/Retrospective Depending on the time frame where data are collected (future or past) Prospective studies: Follow subjects up to the future Retrospective studies: Collect data from past events Figure from Levin KA. Evidence-Based Dentistry (2006) 12 / 47 Longitudinal/Cross-sectional Depending on repetition of measurements Longitudinal studies: Involve repeated observations of the same items over time Cross-sectional studies: Involve observations at a defined time Examples Longitudinal: A study on the effect of a diet with measurements at 1, 2, 3, 6, and 12 months of follow-up Cross-sectional: Population health survey, study of pain prevalence in a hospital 13 / 47 Experimental studies 14 / 47 Experimental studies: Comparability of study subjects Essential to demonstrate the effect of an intervention Methods aimed at ensuring the comparability: Random assignment of subjects to the study groups Heads/tails Computer randomization Random blocks Blinding Maintaining comparability during the experiment http://www.random.org/ 15 / 47 Experimental studies: Placebo effect (1) A placebo is pharmacologically inactive treatment managed like an active one The placebo effect is a phenomenon whereby a patient's symptoms may improve because the patient believes that the treatment works http://www.youtube.com/watch?v=yfRVCaA5o18 16 / 47 Experimental studies: Placebo effect (2) All substances that are used with curative or palliative purposes have a double effect: Pharmacological Caused by suggestion Brain neurotransmitters are at work including chemicals that use the same pathways as opium and marijuana Placebos increase dopamine, a chemical that affects emotions and sensations of pleasure In any experimental study the treatment should be compared with the best existing treatment or, otherwise, with a placebo 17 / 47 Experimental studies: Blinding Knowing the treatment may affect the evaluation of its effect: If the patient knows, it may affect the outcome If the doctor knows, it may affect the outcome measurement If possible, treatment should be blinded Single blind: patients don't know whether they belong to the experimental or the control group Double blind: neither the patients nor the researchers know Triple blind: neither the patients nor the researchers nor the data analysts know 18 / 47 Example: Women's Health Initiative Clinical Trial (WHI CT) Objective: To evaluate the benefits and risks of hormone replacement therapy (HRT), dietary changes and supplements of calcium and vitamin D It included thousands of U.S. women 50-79 years, between 1993 and 1998, the planned follow-up was on average 8.5 years Was discontinued due to adverse effects of HRT http://www.ncbi.nlm.nih.gov/pubmed/12117397} 19 / 47 20 / 47 21 / 47 22 / 47 Conclusions of the WHI trial Overall health risks exceeded benefits from use of combined estrogen plus progestin for an average 5.2-year follow-up among healthy postmenopausal US women. All-cause mortality was not affected during the trial. The risk-benefit profile found in this trial is not consistent with the requirements for a viable intervention for primary prevention of chronic diseases, and the results indicate that this regimen should not be initiated or continued for primary prevention of CHD. 23 / 47 Limitations of experimental studies Ethical Difficulty in generalizing the results when the criteria for inclusion of individuals in an experiment are restrictive Difficulty in getting people to take a certain treatment or a diet for a long time High cost 24 / 47 Phases of clinical trials Phase Goal Participants Testing of drug in non-human subjects (efficacy, Preclinical toxicity and pharmacokinetic) Phase 0 Pharmacokinetics and half-life of the drug ≈ 10 Phase 1 Testing of drug on healthy volunteers ≈ 20-100 Phase 2 Testing of drug on patients (efficacy and safety) ≈ 100-300 Phase 3 Testing of drug on patients (efficacy, effectiveness and safety) ≈300-3000 Post-approval studies (surveillance, long term Phase 4 effects) 25 / 47 Observational studies 26 / 47 Observational cohort studies Cohort: group of people with a common characteristic Exposure to a risk factor or variable of interest is assessed Participants are followed to assess the incidence of disease Incidence of the disease is compared between categories of exposure 27 / 47 Advantages of cohort studies Exposure data are collected before the event of interest occurs Prospectively collected information, in general, has higher quality Are the preferred, among observational studies 28 / 47 Disadvantages of cohort studies Confounding Long time Expensive Losses Exposure to risk factors can change over time Not suitable for the study of rare diseases 29 / 47 Observational case-control studies Individuals with the disease of interest are identified Cases Individuals without the disease, but similar to the cases, are identified Controls Exposure to risk factors among cases and controls is compared Is not possible to obtain the relative risk, but it can be approximated with a measure of association called odds ratio 30 / 47 Advantages of case-control studies Although randomized studies are the best designs, they can not be used to assess the effect of toxic substances. It would be unethical. Case-control studies are a good alternative Faster and cheaper than cohort studies Very appropriate when studying factors associated with an infrequent disease More information on observational studies: https://www.youtube.com/watch? v=yaE7ll1qYO8&t=452s&ab_channel=LlampecsdeCi%C3%A8ncia 31 / 47 Limitations of case-control studies Difficulty in selecting an appropriate control group Confounding Recall bias Data quality problems Case-control studies are an important research tool. Allow to obtain results quickly at low cost. However, the results should be interpreted with caution until more robust studies (experimental or cohort) confirm the results 32 / 47 Observational cross-sectional studies All information is collected at the same time Most often are descriptive Surveys are an example of cross-sectional studies 33 / 47 Advantages of cross-sectional studies Good for descriptive analyses and for generating hypotheses Relatively quick and easy to conduct (no long periods of follow-up) Multiple outcomes and exposures can be studied Good for assessing the burden of disease in a specified population and in planning and allocating health resources In general, are cheaper and simpler than cohort or case control studies 34 / 47 Limitations of cross-sectional studies Confounding Difficulty of obtaining representative samples of the study population Response rate related to individual characteristics (this also happens in other observational designs) Cause or effect? (e.g. poor health and unemployment) 35 / 47 How to choose a design? If possible (ethical and logistical), design an experiment Comparison of treatments should be done with an experimental study If a experiment is not feasible, then try a prospective cohort study If a cohort study is not feasible, then try a case-control study Cross-sectional studies are useful to generate hypotheses 36 / 47 Example: Salk's study on the polio vaccine Study group Number in Paralytic polio group Number of cases Rate per 100000 Randomized control Vaccinated 2nd grade 200 745 33 16 Control 2nd grade 201 229 115 57 Not inoculated 2nd grade 338 778 121 36 Observed control Vaccinated 2nd grade 221 998 38 17 Control 1st and 3rd grade 725 173 330 46 Unvaccinated 2nd grade 123 605 43 35 37 / 47 Frequency measures in experimental and cohort studies Incidence rate: is the frequency of new cases of a disease in a population, in a given period of time. Usually is expressed as number of cases per \(10^n\) person-time at risk. Cumulative incidence: is the proportion of healthy individuals who get a specific disease during a period of time. The cumulative incidence provides an estimate of the probability that a healthy individual turns sick in a period of time. 38 / 47 Association measures in experimental and cohort studies The relative risk (RR) is a measure of association between a risk factor and a disease. It is calculated by dividing the incidence among the exposed and the non-exposed groups. RR=1 indicates that there is no association between risk factor and the disease, if RR \(>1\) indicates that there is a positive association between the risk factor and the onset of the disease. Finally, if RR \(