Downloaded 4,448 times
Uploaded byAchyut Raj Pandey
Cohort Study
This document provides an overview of cohort studies. It defines a cohort study as an analytical study that observes groups over time to determine the frequency of disease among those with and without an exposure. Key features discussed include cohorts being identified prior to disease appearance and the study proceeding from cause to effect. Prospective, retrospective, and ambidirectional cohort study designs are described. Steps of cohort studies include selection of study subjects, obtaining exposure data, comparing exposed and unexposed groups, follow up, and analysis of incidence rates.
More Related Content
![Redox titrations [compatibility mode]](https://cdn.slidesharecdn.com/ss_thumbnails/redoxtitrationscompatibilitymode-130421132405-phpapp02-thumbnail.jpg?width=640&height=640&fit=bounds)
Cohort Study
- 1. Cohort Study Achyut RajPandey Nepal Health Research Council 1
- 2. Concept of cohort Theword cohort is derived from the Latin “cohorts” meaning an enclosure, company, or crowd. In Roman times a cohort was a body of 300–600 infantry. In epidemiological terms the cohort is a group of people with something in common, usually an exposure or involvement in a defined population group. 2
- 3. Definition • Cohort studyis a type of analytical study which is undertaken to obtain additional evidence to refute or support existence of association between suspected cause and diseases. • Other names of cohort study are Longitudinal study, Incidence study and forward looking study 3
- 4. Features of cohortstudies • Cohorts are identified prior to appearance of disease under investigation • The study groups are observed over a period of time to determine the frequency of disease among them • The study proceeds from cause to effects 4
- 5. Indications for cohortstudy • There is good evidence of an association between exposure and disease, from other studies. • Exposure is rare. • Attrition of study population can be minimized. • Sufficient fund is available. 5
- 6. Framework of cohortstudy 6
- 7. Design of CohortStudy Then (a+b) is called study cohort and (c+d) is called control cohort 7
- 8. Consideration during selectionof Cohort • The cohort must be free from disease under study. • Insofar as the knowledge permits, both the groups should be equally susceptible to disease under study. • Both the groups must be comparable in respect of all variable which influence the occurrence of disease • Diagnostic and eligibility criteria of the disease must be defined beforehand. 8
- 9. Types of cohortstudy • Prospective study • Retrospective cohort study • Ambi-directional cohort study 9
- 10. Prospective cohort study •The common strategy of cohort studies is to start with a reference population (or a representative sample thereof), some of whom have certain characteristics or attributes relevant to the study (exposed group), with others who do not have those characteristics (unexposed group). • Both groups should, at the outset of the study, be free from the condition under consideration. Both groups are then observed over a specified period to find out the risk each group has of developing the condition(s) of interest. 10
- 11. Example of ProspectiveCohort Study Framingham Heart Study Framework • Initiated in 1948 to study the relationship of a variety of factors to the subsequent development of heart disease with 5127 samples( 30 to 59 yrs ) at Framingham. • Study subjects were examined every 2 yrs for 20 years. • Daily Surveillance of hospitalization at Framingham hospital. • Study found that Hypertensive, tobacco smoking, elevated blood cholesterol are associated to CHD • Increased physical activity associated with decreased risk of CHD 11
- 12. Problem of prospectivestudy • Study might take long duration. • Sufficient amount of funding for long period. • Missing of study subjects. 12
- 13. Retrospective Cohort Study •A retrospective cohort study is one in which the outcome have all occurred before the start of investigation. • Investigator goes back to the past to select study group from existing records of the past employment, medical and other records and traces them forward through time from the past date fixed on the records usually to the present. • Known with the name of Historical Cohort and noncurrent cohort 13
- 14. Example of RetrospectiveStudy • Suppose that we began our study on association between smoking habit and lung cancer in 2008 • Now we find that an old roster of elementary schoolchildren from 1988 is available in our community, and that they had been surveyed regarding their smoking habits in 1998. • Using these data resources in 2008, we can begin to determine who in this population has developed lung cancer and who has not. 14
- 15. Ambi-directional cohort Study •Elements of prospective and retrospective cohort are combined. • The Cohort is identified from past records and assesses of date for the outcome. The same cohort is the followed up prospectively into future for the further assessment of outcome 15
- 16. Example of Ambi-directionalcohort study • Curt- Brown and Dolls study on effects of radiation Began in 1955 with 13,352 patients who received large dose of radiation therapy for ankylosing spondylitis between 1934 to1954. • Outcome evaluated was death from Leukemia or aplastic anemia between 1934 to 1954. • A prospective component was added up in 1955 and surviving subjects were followed up to identify deaths in subsequent years 16
- 17. Comparison of retrospectiveand prospective cohort study 17
- 18. Prognostic cohort studies Prognosticcohort studies are a special type of cohort study used to identify factors that might influence the prognosis after a diagnosis or treatment. These follow-up studies have the following features: The cohort consists of cases diagnosed at a fixed time, or cases treated at a fixed time by a medical or surgical treatment, rehabilitation procedure, psychological adjustment. By definition, such cases are not free of a specified disease, as in the case of a conventional cohort The outcome of interest is usually survival, cure, improvement, disability, or repeat episode of the illness, etc. 18
- 19. Steps of CohortStudy 19
- 20. 1. Selection ofstudy subjects The usual procedure is to locate or identify the cohort, which may be a total population in an area or sample thereof. Cohort can be: • community cohort of specific age and sex; • exposure cohort e.g. radiologists, smokers, users of oral contraceptives; • birth cohort e.g. school entrants; • occupational cohort e.g. miners, military personnel; • marriage cohort; • diagnosed or treated cohort, e.g. cases treated with radiotherapy, surgery, hormonal treatment. 20
- 21. Open or dynamiccohort • Open population or dynamic population describe a population in which the person-time experience can accrue from a changing roster of individuals. • For example, in a study, the incidence rates of cancer reported by the Connecticut Cancer Registry come from the experience of an open population. Because the population of residents of Connecticut is always changing, the individuals who contribute to these rates are not a specific set of people who are followed through time. 21
- 22. Fixed and ClosedCohort • Fixed Cohort :When the exposure groups in a cohort study are defined at the start of follow-up, with no movement of individuals between exposure groups during the follow-up, the groups are called fixed cohorts. • If no losses occur from a fixed cohort, the cohort satisfies the definition of a closed population and is often called a closed cohort 22
- 23. 2. Obtaining dataon Exposure • From Cohort Members : Personal interview, mailed questionnaire • Review of Records : Certain kinds of information like dose of radiation, kinds of surgery received can only be obtained from medical records. • Medical examination/ Special tests: In some cases information needs to be obtained from medical examination like in case of blood pressure, serum cholesterol, • Environmental Survey of location where cohort lives 23
- 24. Information should becollected in a manner that allows classification of cohort according to • whether or not they have been exposed to suspected factor • According to level or degree of exposure • Demographic variables which might influence frequency of disease under investigation 24
- 25. 3. Comparison Group InternalComparison Group : Single Cohort enters the study and its members on the basis of information obtained , can be classified into several comparison according to degree of exposure Classification of exposure No. of Deaths Death rate ½ pack 24 95.2 ½ to 1 pack 84 107.82 1-2 pack 90 229.2 + 2 pack 97 264.2 Age Standardized death rate among 100000 men per year according to amount of cigarette smoking 25
- 26. External Comparison Group:when information on degree of exposure is not available. if all workers at the factory had some degree of exposure, we would need to select a comparison group from another population, possibly another type of factory Comparison with general population can also be used as comparison group 26
- 27. 4. Follow UP •The length of follow-up that is needed for some studies to reach a satisfactory end- point, when a large enough proportion of the participants have reached an outcome, may be many years or even decades. • At the start of study, method should be determined depending on the outcome of study to obtain data for assessing outcome. 27
- 28. Procedure may be: •Periodic medical examination of each member of cohort • Reviewing physician and hospital records • Routine surveillance of death records • Mailed questionnaire, telephone calls and periodic home visits 28
- 29. 5. Analysis Data analyzedin terms of • Incidence rate of outcome among exposed and non exposed • Estimation of risk 29
- 30. Incidence rate Choice betweencumulative incidence and Incidence Density is a crucial issue • Cumulative incidence: In cohort studies on acute diseases with short induction periods and a short time of follow-up, like outbreaks, the risk of disease can be estimated directly using the cumulative incidence, given a fixed cohort with fixed period of follow-up and a low fraction of drop-outs. • Incidence Density: In cohort studies on chronic diseases with their long follow-up periods, however, the use of the cumulative incidence is not appropriate because usually disease-free follow-up periods differ strongly among cohort members. In such case incidence density is apposite measure 30
- 31. Death No deathIncidence rate Total Exposed A B A/(A+B) A + B Unexpos ed C D C/(C+D) C + D Total A + C B + D A+B+C+ D Outcome* * Outcome : death/disease ANALYSIS OF COHORT STUDIES 31
- 32. A = Exposedpersons who later develop disease or die B = Exposed persons who do not develop diseases or die C = Unexposed persons who later develop disease or die D = Unexposed persons who do not develop diseases or die The total number of exposed persons = A + B The total number of unexposed persons = C + D Incidence of disease(or death) among exposed= A/A+B Incidence of disease(or death) among non-exposed= C/C+D 32
- 33. Relative Risk (RR) •Estimates the magnitude of an association between exposure and disease • Indicates the likelihood of developing the disease in the exposed group relative to those who are not exposed • Ratio of risk of disease in exposed to the risk of disease in nonexposed Relative Risk RR = Risk in exposed(Incidence in exposed group) Risk in non exposed(Incidence in non exposed group) 33
- 34.
- 35. Children (<12 yrs) 1000 Family smoker 500 children Exposed Familynon-smoker 500 children Not exposed Diseased 300 Not diseased 200 Diseased 120 Not diseased 380 OutcomeStart 35
- 36. Rate: Incidence rate •Incidenceof Resp. Infection among exposed children: 300 500 = 60% •Incidence of Resp. Infect. Among non exposed children: 120 500 = 24% 36
- 37. Cohort Study (cont.)RelativeRisk: Incidence rate among exposed Risk Ratio Incidence rate in non exposed. 60 24 = 2.5 Exposed individuals are 2.5 times more likely to develop disease than non exposed individuals. 37
- 38. Difference Measures • Attributablerisk – No. of cases among the exposed that could be eliminated if the exposure were removed = Incidence in exposed - Incidence in unexposed • Population Attributable Risk percentage: PAR expressed as a percentage of total risk in population 100x I I-I PAR% population unexposedpopulation = 38
- 39. Attributable Risk Incidence Exposed Unexposed Iexposed– Iunexposed I = Incidence 39
- 40. Yes No IncidenceRD Yes 100 1900 2000 0.05 No 80 7920 8000 0.01 180 9820 10000 AR: Smoking and Lung cancer Smoking 0.04 Lung Cancer Attributable risk = Incidence in exposed - Incidence in unexposed =0.5-0.1 =0.4 40
- 41. • Excess riskof disease in total population attributable to exposure • Reduction in risk which would be achieved if population entirely unexposed • Helps determining which exposures relevant to public health in community Population Attributable Risk (PAR) unexposedpopulation I-IPAR = 41
- 42. Population Attributable Risk Risk PopulationUnexposed unexposedpopulation I-I Ipopln– Iunexposed 42
- 43. Yes No Risk Yes100 1900 2000 Incidence in exposed= 0.050 No 80 7920 8000 Incidence in unexposed=0.010 180 9820 10000 Incidence in population=0.018 PAR: Smoking 44%100x 0.018 0.010-0.018 PAR% == 0.0080.010-0.018PAR == Smoking Lung Cancer 43
- 44. Conclusion: 44% of lungcancer in the population could be prevented if use of smoking were eliminated 44
- 45. But calculations are not thatsimple in real Cohort studies 45
- 46. British Doctors Study •In 1951, a prospective cohort study was set up among British doctors to investigate the relationship between smoking and mortality, particularly the association between smoking and lung cancer • In 1951, a questionnaire on smoking habits was sent to 49,913 male and 10,323 female doctors , 34,440 male doctors and 6194 female doctors gave sufficient information to classify their smoking status. • The causes of death of 10,072 male and 1094 female doctors who had died during this period were ascertained from death certificates. • The rate of death from lung cancer among smokers was compared to that among non-smokers. 46
- 47.
- 48. Since mortality dependson age and the distribution of subjects by age group is different between the smokers and non-smokers, the effect of age on mortality has to be adjusted for when making comparison on lung cancer mortality between these two groups. A commonly used method to adjust for the age is direct standardization 48
- 49. It would notbe rational to categorize individual smoking one cigarette per day and more than 25 cigarette in same category with equal emphasis So Its better we opt for stratification 49
- 50. Again its notonly the dose of exposure that determines the frequency of disease, there are some other factors like duration of exposure and age at initiation of exposure that can influence occurrence of disease. We need to make adjustment for that too 50
- 51. The relative riskof lung cancer death increased with the level of smoking in both males and females. The relative risk in the men smoking 1–14 and 15– 24 cigarettes per day is much higher than in the women; in the group smoking 25 or more cigarettes per day, the relative risk in men is marginally less than that in women. Does this mean that the effect of low levels of smoking is higher among men than among women? 51
- 52. The proportion ofmen inhaling smoke is higher than women in all three levels of smoking. Men seemed to have started to smoke at an earlier age than women. Since these features of smoking may modify the effect of smoking on lung cancer, their effects have to be adjusted for when comparing the association between smoking and lung cancer in men and women. 52
- 53. …….. too complicated???? But Problem does not end here…. 53
- 54. What if, asubject is followed up from age 23 but has been exposed from age 19 on, he|she is exposed until age 27 followed by an unexposed 5 year period. He|she is again exposed until age 39 at which time his|her person-time at risk ceases either because of disease diagnosis or because of end of follow-up. 54
- 55. For analyzing suchdata we use Poisson models and Cox Proportional Hazards Specialized software packages exist to perform these computations such as Stata (Version 7 or later and Epicure 55
- 56. Advantage of CohortStudies • Temporality can be established • Incidence ca be calculated. • Several possible outcome related to exposure can be studied simultaneously. • Provide direct estimate of risk. • Since comparison groups are formed before disease develops certain forms of bias can be minimized like misclassification bias. • Allows the conclusion of cause effect relationship 56
- 57. Disadvantage of CohortStudies • Large population is needed • Not suitable for rare diseases. • It is time consuming and expensive • Certain administrative problems like loss of staff, loss of funding and extensive record keeping are common. • Problem of attrition of initial cohort is common • Study itself may alter people’s behavior 57
- 58. Ethics in CohortStudy • Classic example issues on research ethics is Tuskegee study on natural history of syphilis in which US Public health service recruited 399 poor black sharecroppers in Macon County as cohort. • Study was lasted from 1932 to 1972. • They were denied of treatment of syphilis although effective treatment was available. Government deceived by saying that they were being treated. 58
- 59. Ethics in CohortStudy • On July 26, 1972, The New York Times described the study as “the longest non therapeutic experiment on human beings in medical history.” The disclosure of this study by the press was a major scandal in the United States. • Led to The Belmont Report: Ethical Principles and Guidelines for the Protection of Human Subjects in Research 59
- 60. Ethics in CohortStudy • These problems can be encountered in cohort study designed to study natural history of disease. • What if treatment becomes available in the middle of research, should we continue research with treatment denial of abort research? • Should we communicate the research finding to individuals are controversial issues. 60
- 61. Biases in cohortstudy Differential loss of follow up Differential follow-up between compared groups may be a major problem. Losses to follow-up, whether due to study withdrawals, unmeasured outcomes, or unknown reasons, are always a concern. This is particularly true when more outcome data is missing in one group than another, as there is no way to be certain that the factor being studied is not somehow related to this observation. 61
- 62. Contamination Subjects initially unexposedto the risk factor of interest may become exposed at a later date. Such “ contamination ” tends to reduce the observed effect of the risk factor. 62
- 63. Selection Bias Perhaps thelargest threat to the internal validity of a cohort studies is selection bias, also called case-mix bias . Select participants into exposed and not exposed groups based on some characteristics that may affect the outcome Information bias− Collect different quality and extent of information from exposed and not exposed groups 63
- 64. Misclassification Bias Differential misclassification Nondifferential misclassification 64
- 65. • Differential misclassification– Errors in measurement are one way only – Example: Measurement bias – instrumentation may be inaccurate, same cut off level of weight for male and female to determine malnourishment 65
- 66. Misclassification Bias (cont.) 250100150 1005050Nonexposed 15050100Exposed TotalDisease-Disease+ RR = a/(a+b)/c/(c+d) = 1.3 True Classification 250100150 905040Nonexposed 16050110Exposed TotalDisease -Disease + RR = a/(a+b)/c/(c+d) = 1.6 Differential misclassification - Overestimate exposure for 10 cases, inflate rates 66
- 67. • Nondifferential (random)misclassification – errors in assignment of group happens in more than one direction – This will dilute the study findings - BIAS TOWARD THE NULL 67
- 68. Misclassification Bias (cont.) Disease+ Disease - Total Exposed 100 50 150 Nonexposed 50 50 100 150 100 250 RR = a/(a+b)/c/(c+d) = 1.3 True Classification Disease + Disease - Total Exposed 110 60 170 Nonexposed 40 40 80 150 100 250 RR = a/(a+b)/c/(c+d) = 1.3 Nondifferential misclassification - Overestimate exposure in 10 cases, 10 controls – bias towards null 68
- 69. Control of Bias •Restriction • Stratification • Mathematical Modeling -Poisson regression model -Cox proportional hazard 69
- 70. When Is aCohort Study Warranted? • When the (alleged) exposure is known • When exposure is rare and incidence of disease among exposed is high (even if the exposure is rare, determined investigators will identify exposed individuals) • When the time between exposure and disease is relatively short • When adequate funding is available • When the investigator has a long life expectancy 70
- 71. Reference • Park.k.,(2007), Park’stextbook of preventive and social medicine(19th edition) • Gordis.L.,(2008), Epidemiology.,(4th edition) • Framimgham Heart study retrived from www.framingham.com/heart/timeline.htm,on 29th Jan,2011. 71
- 72.
- 73. Classic example ofCohort study : Study on London Cholera Outbreak • The classical study on the London cholera epidemic of 1849 conducted by John Snow is an example of a cohort study on infectious diseases . • Two different water companies (the Lambeth and the Southwark & Vauxhall) supplied households within various regions of London 73
- 74. Classic example ofCohort study : Study on London Cholera Outbreak • The companies differed in one important feature, the location of the water intake. The Lambeth had moved their water intake upstream from the sewage discharge point in 1849; whereas, the Southwark & Vauxhall continued to obtain water downstream of the sewage discharge point. • Dr. Snow classified households according to their exposure to the two water sources and showed a substantial difference in cholera mortality, 315 versus 37 cholera deaths per 10,000 households served by the Lambeth and Southwark & Vauxhall companies, respectively. 74
- 75.
Editor's Notes
- #19 By definition, such cases are not free of a specified disease, as in the case of a conventional cohort study (but are free of the ‘outcome of interest’).
- #22 If the follow-up of fixed cohorts suffers from losses to follow-up incidence rates can still be measured directly and used to estimate average risks and incidence times. Accrue : periodic accumulated over time
- #27 if all workers at the factory had some degree of exposure, we would need to select a comparison group from another population, possibly another type of factory, to ensure that the comparison group only differed in terms of their exposure and not in terms of other factors.
- #31 In this case, outcome events are preferably described by rates, that represent the number of outcome events divided by the cumulated duration of event-free follow-up periods of all cohort members at risk.
- #33 Relative Risk(RR) = Incidence of disease(or death) among exposed/Incidence of disease(or death) among non-exposed. Relative Risk(RR)= { A/(A+B)}/ { C/(C+D)}
- #34 Defined as the ratio of the incidence of disease in the exposed group (expressed as Ie) divided by the corresponding incidence of disease in the non exposed group (Io)
- #47 This is an important study and follow-up of participants is still under way. Results for 50 years of follow-up were published recently in the British Medical Journal (Doll et al. 2004).
- #51 the estimation of level of exposure to tobacco by counting the number of cigarettes smoked per day may not be appropriate. For instance, the duration of smoking, inhalation practices and age when started to smoke might have been different Since mortality depends on age and the distribution of subjects by age group is different between the smokers and non-smokers, the effect of age on mortality has to be adjusted for when making comparison on lung cancer mortality between these two groups. A commonly used method to adjust for the age is direct standardization between the men and women. Unless the effects of these features of smoking are taken into account, one cannot conclude that low to moderate smoking has a higher effect in men than in women.
- #56 In epidemiological cohort studies the standard model for analyzing such data is the Poisson model which is a statistical model of the disease rates. Basically the Poisson model assumes that the number of events di in each category i (combination of age category j and the kth combination of exposure variables) follows a Poisson distribution with parameter niλi. The standard (multiplicative) model would then assume that ln(λi) = αj + βk A comprehensive account of Poisson modeling is given by Breslow and Day (1987, Chap. 4). An alternative way of analysing event history data (another denomination of cohort data focussed on events), is by using Cox’ proportional hazard model. This model acknowledges that the categorisation of continuous data always implies a loss of information and therefore a loss in statistical power.Moreover, there is no need to explicitly estimate the effects of nuisance parameters if it can be avoided. The first step in proportional hazard model is the choice of one of the time variables considered. This basic time variable can either be age as was implicit at the beginning of this chapter, but in some settings, this variable can be the calendar time or even the time since the beginning of follow-up. Once this special time variable has been fixed, its effects are estimated nonparametrically. The key idea of Cox’s regression is that no information is lost when considering only the time points ti at which an event of interest occurs. At each such time point a “risk set” is set up including all members of the cohort contributing person-time (at risk and under observation) at this time point. If one wants to use a Cox model, the first step is thus to identify all risk sets. Then, one must obtain the value at
- #57 Once investigators assemble the study cohorts, they are free to study more than one outcome, provided that the study subjects are free of each outcome of interest when the study begins. For the smoking and kidney transplant failure example, once cohorts of smokers and non-smokers are identified, investigators could also study whether smoking was associated with the development of kidney or bladder cancer, provided they exclude individuals who have evidence of these cancers at the beginning of the study. One of the largest cohort studies even conducted was the Nurses Health Study, which recruited 127,000 nurses between the ages of 30 and 55. The nurses completed questionnaires every 2 years that queried medical conditions, prescription and nonprescription medication use, social habits, dietary patterns, and physical activity. Creation of multiple cohort studies that evaluated risk factors for a wide range of diseases, including cancer, heart disease, and fractures. The British Doctors Study cohort was established in 1951 and followed for more than 50 years, although many of the original 40,701 participants are now dead. It has been of enormous value, particularly in relation to identifying the manifold health consequences of smoking. This is despite the fact that, compared with studies today, only limited exposure data were collected on a very short postal questionnaire mailed to the doctors at 10-year intervals since 1951 (Doll and Hill, 1964). The US Nurses’ Health Study started in 1976 with 121,964 female nurses aged 30–55, and 5 years of funding. Since then its focus has widened enormously from the oral contraceptive–breast cancer links for which it was first funded (Stampfer et al., 1988) to cover many exposures (including diet) and a multitude of outcomes. It has now accumulated more than 30 years of follow-up and is still going strong. It is very expensive to run, but the scientific and public health yield has been exceptional. The Nurses’ Health Study II began in 1989 with 117,000 nurses aged 25–42 (Rockhill et al., 1998) and recruitment has recently started for a Nurses’ Health Study III!
- #61 It is now generally accepted that studies on humans should be carried out with informed consent. This principle, originally developed in relation to controlled clinical trials, has generally now been extended to observational epidemiology studies, including cohort studies. In the past, if a cohort was recruited that involved the subjects participation in providing data, their agreement to supply the data (e.g. respond to a questionnaire) was generally regarded as implied consent. However, now, in addition to providing information on questionnaires, for many cohorts, biological specimens (e.g. blood, buccal cells) are requested, and then it becomes mandatory that the respondent provide consent for the future use of such specimens for research purposes. However, at the time the specimens are provided, it is impossible to know the precise use the investigators may wish to apply to this material. An example relates to the fact that the majority of participants in the sub-cohorts of the European Prospective Investigation of Diet and Cancer (EPIC; Riboli and Kaaks 1997) provided blood specimens in the early 1990s; a few without signing a consent form, the majority did so. However, now that genetic studies are commonplace on such specimens, it has become apparent that some of the consent forms did not specifically mention genetic analyses as potential research usages. This has led to difficulties in obtaining approval for such sub-studies from human experimentation committees, some of which wanted new consent forms to be signed, specific to the genetically-associated sub-study planned. Obtaining new consent, however, will become increasingly difficult as time goes on, and a number of subjects with the endpoint of interest may have died. In the United States, potential restrictions upon studies such as these have caused difficulties. In Europe, especially Scandinavia, there has been a more relaxed view of the ethical acceptability of studies on stored specimens, many such collections having been originally made without a formal informed consent process, but for which studies conducted with full preservation of confidentiality have been deemed to be ethically acceptable.
- #63 Intention to treat analysis : The conventional approach is to analyze the data in a intention-to treat fashion; that is, subjects are analyzed in the group to which they were originally assigned, regardless of what occurred after that point. Given that it reduces observed risk, however, some researchers choose to additionally analyze in a “ treatment-as-received ” manner. A number of statistical methods, including Cox proportional hazard modeling and Poisson regression, are suitable for this approach. In the latter technique, rather than allowing a single subject to remain in only one group for analysis, the amount of “ person time ” each subject represents within a study is considered. If the groups being compared consist of a given treatment versus no treatment or an alternate therapy, it is possible that some controls will start treatment or subjects may switch therapies.
- #66 , such as using only one size blood pressure cuff to take measurements on both adults and children
- #74 Previous reports from the Registrar General had drawn attention to the possibility that differences in water supply were associated with differences in cholera rates across sections of London.
- #75 Previous reports from the Registrar General had drawn attention to the possibility that differences in water supply were associated with differences in cholera rates across sections of London.