Abstract
A key stage of epidemiologic research is the study design. This is defined to be the process of planning an empirical investigation to assess a conceptual hypothesis about the relationship between one or more exposures and a health outcome. The purpose of the study design is to transform the conceptual hypothesis into an operational hypothesis that can be empirically tested. Since all study designs are potentially flawed, it is therefore important to understand the specific strengths and limitations of each design. Most serious problems or mistakes at this stage cannot be rectified in subsequent stages of the study.
Keywords
These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.
This is a preview of subscription content, log in via an institution.
Buying options
Tax calculation will be finalised at checkout
Purchases are for personal use only
Learn about institutional subscriptionsReferences
References for Efficacy, Effectiveness, and Efficiency
Detsky, A S. Evidence of effectiveness: evaluating its quality. In F. A. Sloan (Ed.), Valuing health care: costs, benefits, and effectiveness of pharmaceuticals and other medical technologies (pp. 15-29). Cambridge, UK: Cambridge University Press, 1995.
Detsky AS, Naglie IG (1990) A Clinician Guide to Cost-Effectiveness Analysis. Ann Intern Medicine 113(2):147–54
Grimes DA, Schulz KF (2002) An overview of clinical research: the lay of the land. Lancet 359(9300):57–61
Kocher MS, Henley MB (2003) It is money that matters: Decision analysis and cost-effectiveness analysis. Clinical Orthopaedics and Related Research 413:106–16
Reference for Diabetes Research Group Study (Clinical Trial)
The Diabetes Control and Complications Trial Research Group (1993) The effect of intensive treatment of diabetes on the development and progression of long-term complications in insulin-dependent diabetes mellitus. N Engl J Med 329(14):977–86
Reference for Framingham Heart Study (Prospective Cohort Study)
Feinleib M, The Framingham study: sample selection, follow-up, and methods of analysis, in National Cancer Institute Monograph, No. 67, Greenwald P (editor), US Department of Health and Human Services, 1985.
Dorgan JF, Brown C, Barrett M et al (1994) Physical activity and risk of breast cancer in the Framingham Heart Study. Am J Epidemiol 139(7):662–9
Margolis JR, Gillum RF, Feinleib M, Brasch RC, Fabsitz RR (1974) Community surveillance for coronary heart disease: the Framingham Cardiovascular Disease Survey. Methods and preliminary results Am J Epidemiol 100(6):425–36
Reference for VDT use and Spontaneous Abortion (Retrospective Cohort Study)
Schnorr TM, Grajewski BA, Hornung RW, Thun MJ, Egeland GM, Murray WE, Conover DL, Halperin WE (1991) Video display terminals and the risk of spontaneous abortion. N Engl J Med 324(11):727–33
Reference for Nonsteroidal Anti-inflammatory Drugs and Alzheimer’s Disease (Prospective Cohort Study)
in t’ Veld BA, Ruitenberg A, Hofman A, Launer LJ, van Duijn CM, Stijnen T, Breteler MM, Stricker BH. Nonsteroidal antiinflammatory drugs and the risk of Alzheimer’ disease. N Engl J Med 2001;345(21):1515-21.
References for Reye’s Syndrome (Case-Control Studies)
Waldman RJ, Hall WN, McGee H, Van Amburg G (1982) Aspirin as a risk factor in Reye’ syndrome. JAMA 247(22):3089–94
Halpin TJ, Holtzhauer FJ, Campbell RJ, Hall LJ, Correa-Villasenor A, Lanese R, Rice J, Hurwitz ES (1982) Reye’ syndrome and medication use. JAMA 248(6):687–91
Daniels SR, Greenberg RS, Ibrahim MA (1983) Scientific uncertainties in the studies of salicylate use and Reye’ syndrome. JAMA 249(10):1311–6
Hurwitz ES, Barrett MJ, Bregman D, Gunn WJ, Pinsky P, Schonberger LB, Drage JS, Kaslow RA, Burlington DB, Quinnan GV et al (1987) Public Health Service study of Reye’ syndrome and medications. Report of the main study JAMA 257(14):1905–11
Forsyth BW, Horwitz RI, Acampora D, Shapiro ED, Viscoli CM, Feinstein AR, Henner R, Holabird NB, Jones BA, Karabelas AD et al (1989) New epidemiologic evidence confirming that bias does not explain the aspirin/Reye’ syndrome association. JAMA 261(17):2517–24
References for Creutzfeldt-Jakob Disease (Case-Control Studies)
van Duijn CM, Delasnerie-Laupretre N, Masullo C, Zerr I, de Silva R, Wientjens DP, Brandel JP, Weber T, Bonavita V, Zeidler M, Alperovitch A, Poser S, Granieri E, Hofman A, Will RG. Case-control study of risk factors of Creutzfeldt-Jakob disease in Europe during 1993-95. European Union (EU) Collaborative Study Group of Creutzfeldt-Jakob disease (CJD). Lancet 1998;351(9109):1081-5.
Will RG, Ironside JW, Zeidler M, Cousens SN, Estibeiro K, Alperovitch A, Poser S, Pocchiari M, Hofman A, Smith PG (1996) A new variant of Creutzfeldt-Jakob disease in the UK. Lancet 347(9006):921–5
General Epidemiologic Design
Checkoway H, Pearce N, Dement JM. Design and conduct of occupational epidemiology studies: II. Analysis of cohort data. Am J Ind Med 1989;(15(4):375-94.
Greenberg RS, Daniels SR, Flanders WD, Eley JW, Boring JR (2001) Medical Epidemiology, 3rd edn. Lange Medical Books, New York
Kleinbaum DG, Kupper LL, Morgenstern H (1982) Epidemiologic Research: Principles and Quantitative Methods. John Wiley and Sons Publishers, New York
Steenland K (ed.). Case studies in occupational epidemiology. Oxford University Press, New York, 1993.
Example of Cross-Sectional Studies
Smith WCS, Woodward M, Tunstall-Pedoe H. Intermittent claudication in Scotland, in Epidemiology of Peripheral Vascular Disease. (ed FGR Fowkes.), Springer-Verlag, Berlin, 1991
Hybrid Designs
Coates RJ, Weiss NS, Daling JR, Rettmer RL, Warnick GR (1989) Cancer risk in relation to serum copper levels. Cancer Res 49(15):4353–6
Linn JT, Wang LY, Wang JT, Wang TH, Yang CS, Chen CJ (1995) A nested case-control study on the association between Helicobacter pylori infection and gastric cancer risk in a cohort of 9775 men in Taiwan. Anticancer Res 15:603–6
Maclure M, Mittleman MA (2000) Should we use a case-crossover design? Annu Rev Public Health 21:193–221
Maclure M (1991) The case-crossover design: a method for studying transient effects on the risk of acute events. Am J Epidemiol 133(2):144–53
Redelmeier DA, Tibshirani RJ (1997) Association between cellular-telephone calls and motor vehicle collisions. N Eng J Med 336(7):453–8
Ecologic
CDC. Summary of notifiable diseases in the United States. MMWR Morb Mortal Wkly Rep 1990;38(54):1-59.
Morgenstern H (1995) Ecologic studies in epidemiology: concepts, principles, and methods. Annu Rev Public Health 16:61–81
Proportional Mortality
Mancuso TF, Stewart A, Kneale G (1977) Radiation exposures of Hanford workers dying from cancer and other causes. Health Phys 33:369–85
Causation and Meta Analysis
Blalock HM, Causal Inferences in Nonexperimental Research, Chapter 1, Norton Publishing, 1964.
Chalmers I, Altman DG (eds.), Systematic Reviews, BMJ Publishing Group, London, 1995.
Chalmers TC (1991) Problems induced by meta-analyses. Stat Med 10(6):971–80
Hill AB, Principles of Medical Statistics, 9th Edition, Chapter 24, Oxford University Press, 1971.
Lipsey MW, Wilson DB. Practical meta-analysis. Applied Social Research Methods Series; Vol. 49. Sage Publications, Inc., Thousand Oaks, CA: 2001.
Mosteller F, Colditz GA (1996) Understanding research synthesis (meta-analysis). Annu Rev Public Health 17:1–23
Petitti DB, Meta-analysis Decision Analysis and Cost-Effectiveness Analysis; Methods for Quantitative Synthesis in Medicine, Oxford University Press, 1994.
Popper KR, The Logic of Scientific Discovery, Harper and Row Publishers, 1968.
Rothman KJ (1976) Causes. Am J Epidemiol 104(6):587–92
Susser M, Causal Thinking in the Health Sciences, Oxford University Press, 1973.
U.S. Department of Health, Education, and Welfare, Smoking and Health, PHS Publ. No. 1103, Government Printing, Washington DC, 1964.
Weiss NS (1981) Inferring causal relationships: elaboration of the criterion of ’dose-response’”. Am J Epidemiol 113(5):487–90
Author information
Authors and Affiliations
Appendices
Homework Exercises
ACE-1. Study Type. State the type of study described by each of the following paragraphs:
-
a.
To investigate the relationship between egg consumption and heart disease, a group of patients admitted to a hospital with myocardial infarction were questioned about their egg consumption. Another group of patients admitted to a fracture clinic and matched on age and sex with the first group were also questioned about their egg consumption using an identical protocol.
-
b.
To investigate the relationship between certain solvents and cancer, all employees at a factory were questioned about their exposure to an industrial solvent, and the amount and length of exposure measured. These subjects were regularly monitored, and after 10 years a copy of the death certificate for all those who died was obtained.
-
c.
A survey was conducted of all nurses employed at a particular hospital. Among other questions, the questionnaire asked about the grade of the nurse and whether or not she was satisfied with her career prospects.
-
d.
To evaluate a new school (i.e., approach) for treating back pain, patients with lower back pain were randomly allocated to either the new school or to conventional occupational therapy. After 3 months, they were questions about their back pain, and observed lifting a weight by independent monitors.
-
e.
A new triage system has been set up at the local Accident and Emergency Unit. To evaluate this new system, the waiting times of patients were measured for 6 months and compared with the waiting times at a comparable nearby period.
-
f.
The Tumor Registry in a certain US state was used to identify all primary cases of bladder cancer in the state during a given period. These cases were compared to a sample of non-cases from the same state that have been matched on age and time of diagnosis. All subjects or their surviving relatives in both groups were interviewed to collect information on saccharin consumption and other known risk factors for bladder cancer.
-
g.
In the Hanford study of nuclear power workers (1977), 3500 certified deaths occurred among plant workers between 1944 and 1972. Among these deaths, a significantly greater proportion of workers exposed to low levels of ionizing radiation than unexposed workers had died of RES (reticuloendothelial system) cancers.
ACE-2. Case-Control vs. Prospective Cohort. Which of the following choices is not an advantage of using a case-control study as opposed to a prospective cohort study? (There may be more than one correct answer here.)
-
a.
Less expensive
-
b.
Can be completed more rapidly
-
c.
More appropriate for the study of rare diseases,
-
d.
More appropriate for the study of diseases that develop slowly
-
e.
More appropriate for the study of several exposures.
-
f.
More appropriate for the study of several diseases.
-
g.
Allows more accurate assessment of exposure.
ACE-3. Randomization. A randomized clinical trial was designed to compare two different treatment approaches for irritable–bowel syndrome. The purpose of randomization in this study was to:
-
a.
increase patient compliance with treatment
-
b.
obtain comparison groups that are similar on other variables that may influence the disease.
-
c.
obtain comparison groups that are similar on any other variables measured in the study.
-
d.
Increase the likelihood of finding a significant effect of treatment
-
e.
obtain a representative sample in the study.
ACE-4. Clinical Trial. In a randomized clinical trial designed to compare two treatments for asthma, the clinicians knew which treatment the patients received, but the patients themselves did not know which treatment they received. This is an example of:
-
a.
compliance
-
b.
intention-to-treat
-
c.
double-blinding
-
d.
placebo effect
-
e.
none of the above
ACE-5. Case-Control Study: TB. (Primarily for medical students/clinicians) The following questions apply to the article ” Variations in the NRAMPI gene and susceptibility to tuberculosis in West Africans.” (Bellamy R, Rowende t al., New Eng J of Med, 38 (10), pp. 640-643, March 1998).
-
a.
Who were the patients in this study?
-
b.
Who were the controls?
-
c.
Did the controls and cases differ in any major aspect other than disease status?
-
d.
What was the design of this study?
-
e.
State the “null” hypothesis, either symbolically or in words.
-
f.
What do the investigators conclude about the null hypothesis?
ACE-6. Prospective Cohort Study. In a famous prospective cohort investigation, the population to be studied encompassed all physicians listed in the British Medical Register and resident in England and Wales as of October 1951. Information about present and past smoking habits was obtained by questionnaire. Information about lung cancer came from death certificates and other mortality data recorded during ensuing years.
-
a.
What makes this study prospective? List two advantages and two disadvantages of this approach.
-
b.
What advantages and disadvantages come with selecting physicians as a cohort for follow-up?
ACE-7. Intention to Treat. A clinical trial was conducted to compare a new hypertension therapy to the standard therapy. At the end of the follow-up period, the investigators performed two separate analyses. For the first analysis, they followed the “intention to treat” rule. For the second analysis, they included only those patients known to have taken the prescribed therapy throughout the study period. The results of the two analyses differed substantially. The most likely explanation for the discrepancy is: [Choose one best answer]
-
a.
The randomization was unsuccessful
-
b.
The new therapy was not effective
-
c.
There was a significant degree of recall bias
-
d.
There was a problem with patient compliance
ACE-8. Blinding. A clinical trial was conducted to compare the performance of two treatments. Describe a situation in which it would NOT be feasible for the trial to be blinded. [Be sure that your answer indicates an understanding of what it means for a trial to be blinded.]
ACE-9. Randomization: Clinical Trials. What is the purpose of randomization in a clinical trial? [Choose one best answer.]
-
a.
To make the diseased and non-diseased as similar as possible with respect to all variables except the exposure of interest
-
b.
To reduce the number of subjects who are lost to follow-up
-
c.
To isolate the effect of the exposure of interest
-
d.
To encourage compliance with the assigned treatment regimen
ACE-10. Study Design: Adiposity and CHD. (Primarily for medical students/clinicians) .The following questions apply to the article: “Abdominal adiposity and coronary heart disease in women.” (Rexrode KM, Carey VJ et al., JAMA, 280 (21), pp 1643-1646, December 1998).
-
a.
What is the design of this study?
-
b.
What are the two principal null hypotheses for this study?
-
c.
How many women were included in the final analysis? What percentage of the entire cohort does this represent?
-
d.
The study used self-reported weight. How assured are you that this information is accurate?
-
e.
Given the results in Table 2, which of the two variables, waist-hip ratio or waist circumference, seems to be the better predictor of CHD risk? Why?
ACE-11. Density Sampling. Which of the following is NOT true of “density sampling” of controls in a case-control study? [Choose one best answer.]
-
a.
A subject identified as a control may later be identified as a case.
-
b.
A subject identified as a case may later be identified as a control.
-
c.
The odds ratio calculated from such a study is likely to be a good estimate of an incidence measure of association.
-
d.
An individual subject may serve as a control for more than one case.
ACE-12. Control Group. What is the purpose of the control group in a case-control study? [Choose one best answer.]
-
a.
To provide an estimate of the background risk or rate of disease.
-
b.
To provide an estimate of the exposure frequency among the population that produced the cases.
-
c.
To provide an estimate of the magnitude of the placebo effect.
-
d.
To provide an estimate of the expected number of cases among the unexposed.
ACE-13. Study Design: Breast Cancer. A paper entitled “Electric Blanket Use and Breast Cancer Risk among Younger Women” appeared in a recent issue of the American Journal of Epidemiology. The methods section included the following information:
Cases were women newly diagnosed with in situ or invasive breast cancer between May 1, 1990, and December 31, 1992, who were residents of three U.S. geographic areas. Controls were women identified by random digit dialing and frequency-matched to cases by 5-year age group and geographic area. All women were asked about whether they had ever regularly used electric blankets, electric mattress pads, or heated waterbeds. A positive response referred to the aggregate use of any or all of the devices at any time in the respondent’s life prior to enrollment in the study.
Which one of the following best describes the design of this study?
-
a.
Cross-sectional
-
b.
Cohort
-
c.
Nested Case-Control
-
d.
Population-based Case-Control
-
e.
Descriptive
Answers to Study Questions and Quizzes
2.1 Q3.1
-
1.
Experimental
-
2.
Observational
-
3.
Observational
-
4.
Experimental
-
5.
Experimental
-
6.
Observational
2.2 Q3.2
-
1.
Therapeutic
-
2.
Preventive
-
1.
a
-
2.
b
-
3.
b
-
4.
a
-
5.
c
-
6.
c
-
7.
a
2.3 Q3.3
-
1.
For VDT users the percentage is (54/366) x 100 = 14.8 % whereas for VDT non-users the percentage is (82/516) x 100 = 15.9 %. The two percentages differ by only 1 %.
2.4 Q3.4
-
1.
In a cohort study, the directionality is always forward.
-
2.
The timing is prospective, since the health outcome, in this case Alzheimer’s disease, occurs after the onset of the study.
2.5 Q3.5
-
1.
b
-
2.
a
-
3.
b
-
4.
c
-
5.
b
-
6.
Retrospective
-
7.
Retrospective
-
8.
Prospective
-
9.
Both
-
10.
Neither
-
11.
Both
-
12.
Retrospective
2.6 Q3.6
-
1.
b
-
2.
a
-
3.
a
-
4.
c
-
5.
b
-
6.
case-control
-
7.
case-control
-
8.
prospective cohort
-
9.
prospective cohort
-
10.
case-control
-
11.
prospective cohort
-
12.
case-control
-
13.
prospective cohort
-
14.
T – If controls are chosen from a different population from which the cases came, there may be selection bias.
-
15.
F – Hospital controls have an illness; such controls are typically not representative of the community from which the cases came.
-
16.
F – Population-based controls can be obtained from random dialing of telephone numbers in the community from which the cases are derived. There is no guarantee that neighbors of cases will be chosen.
2.7 Q3.7
-
1.
b
-
2.
a
-
3.
c
-
4.
b
-
5.
T
-
6.
F
-
7.
T – A cross-sectional study includes only cases that survive long enough to be available for study. This could lead to a misleading conclusion about an exposure-disease relationship since non-survivors are excluded.
-
8.
T
2.8 Q3.8
-
1.
Backward
-
2.
Prospective
2.9 Q3.9
-
1.
Both
-
2.
Both
-
3.
Neither
-
4.
Neither
-
5.
Both
-
6.
Nested case-control study
-
7.
Both
-
8.
Nested case-control study
-
9.
Neither
-
10.
Case-cohort study
2.10 Q3.10
-
1.
F – The unit of analysis in an ecologic study is a group (e.g., census tract, state, country) and data on both exposure and disease is not simultaneously obtained on individuals
-
2.
T – A proportional mortality study includes observations on deaths without information about the candidate population (i.e., denominators).
-
3.
T – Both ecologic and proportional mortality studies use “incomplete” designs.
-
4.
F – Proportional mortality studies use information on deaths about individuals.
-
5.
T
-
6.
T
Author Queries
AQ1: Please provide citations for all references under the following headings if applicable.
“Reference for Diabetes Research Group Study” ,
“Reference for Framingham Heart Study”,
“Reference for VDT use and Spontaneous Abortion”,
“Reference for Nonsteroidal Anti-inflammatory Drugs and Alzheimer’s Disease”,
“References for Reye’s ………..”. etc.
Rights and permissions
Copyright information
© 2013 Springer Science+Business Media New York
About this chapter
Cite this chapter
Kleinbaum, D.G., Sullivan, K.M., Barker, N.D. (2013). Epidemiologic Study Designs. In: ActivEpi Companion Textbook. Springer, New York, NY. https://doi.org/10.1007/978-1-4614-5428-1_3
Download citation
DOI: https://doi.org/10.1007/978-1-4614-5428-1_3
Published:
Publisher Name: Springer, New York, NY
Print ISBN: 978-1-4614-5427-4
Online ISBN: 978-1-4614-5428-1
eBook Packages: Mathematics and StatisticsMathematics and Statistics (R0)