The UK Faculty of Public Health has recently taken ownership of the Health Knowledge resource. This new, advert-free website is still under development and there may be some issues accessing content. Additionally, the content has not been audited or verified by the Faculty of Public Health as part of an ongoing quality assurance process and as such certain material included maybe out of date. If you have any concerns regarding content you should seek to independently verify this.

Alternative study designs in epidemiology

Introduction

Learning objectives:You will learn about nested case control studies, variations of the randomised controlled trial, ethical issues and intention to treat analysis.

This section will discuss some alternative versions of the common study designs and their uses. This assumes prior knowledge of the basic epidemiological study designs. Nested case control studies are described, as are variations of the randomised controlled trial. These include crossover trials, factorial trials and cluster randomised trials. Finally, some ethical and analytical issues around randomised controlled trials are discussed.

Read the resource text below.

Resource text

Nested Case Control Studies

A nested case control study is one where the cases and controls are selected from individuals within an established cohort study. Cases that arise within the cohort during the follow up period are selected, together with a sample of unaffected members of the cohort, which are selected as controls. The advantage of nested case control studies is that certain exposure data will already have been collected for both cases and controls, which limits the potential for recall bias[1]. They are analysed in the same way as a traditional case control study [2].

Strengths of nested case control studies

  • Relatively cheap and easy to conduct since the resources of the existing cohort can be utilised.
  • Data related to exposure and confounding have been already been collected.
  • Can utilize the baseline data on exposure and confounding collected before the onset of disease, which reduces the potential for recall bias and uncertainty regarding the temporal sequence between exposure and disease onset (1).
  • In a traditional case control study it is impossible to verify whether the cases and controls represent random samples from a particular population.
  • Unlike a traditional case control study, they do not depend on an assumption that the disease (outcome) is rare (2).

Example of nested case control study [3],

Dietary Intake, Atopy, and Wheeze at Age 3 Years: A Nested Case-Control Study. C. S. Murray et.al 2003.

In this study, dietary intake of children aged 3-5, who were atopic and had a history of recurrent wheeze (cases), was compared with children who were non-atopic and had never wheezed (controls). A prospective birth cohort in the UK was used from which to select the cases and controls. They were matched for age, sex, parental atopy, indoor allergen exposure and pet ownership. Cases and controls were matched 1:1, and there were 37 in each group. The study found that the atopic children with recurrent wheeze had a significantly higher polyunsaturated fat intake compared with non-atopic, non-wheezy children and that this could be linked to an increased risk of allergic inflammation.

This study was possible due to the selection of cases and controls from a large prospective birth cohort containing substantial information about participants.

A better understanding of this study design can be achieved by reading other studies which show how they are carried out. Examples of these can be found in the related links section.

Variations of the randomized control trial

Crossover trials

A crossover trial is where each subject acts as their own control. That is, each patient receives all treatments under investigation in sequence. For example, each subject receives two or more different treatments during the trial. Random allocation determines the sequence in which each participant receives each treatment. This type of trial may be used when the intervention does not have long term effects.

A particular advantage of a crossover trial is that each subject is acting as their own control, so that groups are as similar as possible.

This study design is suitable in studies such as trials of analgesics for pain relief or therapies for asthma. This is because there is unlikely to be a carry over effect of the first treatment into the period when the second treatment is given. To avoid the carry over effect it is common to have a 'washout period' where no treatment is given between each intervention to avoid the possibility of a residual effect from the previous treatment [5].

The main advantage of these trials is that they account for between participant variability in the outcome, so they may be more efficient than a trial where treatments are randomly allocated to different individuals [5].

Analysis of cross over trials

Analysis of these trials must consider the design by using methods for paired data. If the outcome is numerical then the mean difference between outcomes should be analysed and the standard deviation of the mean differences reported. If the outcome is binary then methods for analysing matched pairs must be used [5].

Factorial trials

A factorial trial is where two or more interventions are evaluated simultaneously compared with a control group in the same trial. This type of RCT is commonly used to evaluate interactions between interventions. It may be used in public health to compare two slightly different interventions with a control.

An example of a factorial trial can be found in the related links section [4].

Community or cluster randomised controlled trials

Cluster randomised trials involve groups of individuals or communities as opposed to individuals. Groups or communities are randomised to receive the intervention or standard/no treatment. This type of study design may be used to evaluate preventative health services such as impregnated bed nets in malaria endemic areas. For example, a community may be randomly allocated to impregnated bed nets or water fluoridation.

Cluster randomised trials are necessary where it is not feasible to randomise individuals, and where individual randomisation would introduce bias.

Examples of cluster randomised trials

  • Evaluation of screening of hypertension among the elderly in the UK, in which the unit of randomisation was the GP practice.
  • Evaluation of the impact on adolescent sexual behaviour of a sex education programme delivered through school, in which the schools were the unit of randomisation.
  • Evaluation of screening of hypertension among the elderly in the UK, in which the unit of randomisation was the GP practice.

    Essential points to note

    • Where the number of clusters is small, the randomisation may not exclude baseline differences between control and intervention groups. Any factors where differences are apparent should be controlled for in the analysis.
    • Intracluster correlation is the extent to which members of a cluster resemble each other more than they resemble members of other clusters.
    • Intracluster correlation needs to be taken into account in assessing effect size and sample size to ensure an adequate sample size.

    Other issues in randomised controlled trials

    Ethical issues

    The use of RCT raises important ethical issues. For example, there must be sufficient doubt about the particular agent or intervention being tested to allow withholding of it from half the subjects, and at the same time there must be sufficient belief in the agent's potential to justify exposing the remaining half of all willing and eligible participants [6].

    In addition, there must be sufficient belief that the intervention under investigation is safe.

    Informed consent is essential in a RCT. Study subjects must understand that they are participating in an experiment and that in a placebo-controlled trial they may receive an inactive product. In addition, participants must be informed of the aims, methods and potential benefits or hazards of participating in the trial.

    In a controlled trial, careful consideration needs to be given to what intervention is given to the control group. For example, where an effective intervention exists they should not be deprived of receiving that intervention.

    It is essential that study participants do not suffer as the result of a RCT. RCT protocols are inspected by an ethics committee before they may proceed, and most RCTs incorporate an ongoing data monitoring and safety committee who are independent of the investigators.

    Intention to treat analysis

    research.jpgIntention to treat is a method for the analysis of randomised controlled trials that compares patients in the groups to which they were originally randomly assigned [7]. That is, in an intention to treat analysis the outcome in a subject is included in the analysis, regardless of whether they were lost to follow-up and of the treatment they actually received. During the RCT, subjects may refuse to continue to participate, or may stop taking their allocated treatment.

    The aim of the intention to treat analysis approach is to maintain treatment groups that are similar apart from random variation. The strength of randomisation may be lost if analysis is not performed on the groups produced by the randomisation process. In addition, an intention to treat analysis is used to avoid the potential biases which can arise due to variations in the level of participation in the intervention and control groups. Note that subjects who failed to continue to take the allocated treatment may have done so due to adverse side effects, or because the treatment was not satisfactory and they changed treatments or declined to receive the intervention.

    In the statistical analysis, all study subjects should be retained in the group to which they were originally allocated, regardless of whether or not that was the treatment received. It should be noted that this type of analysis may underestimate the true effect of an intervention treatment, since the intervention group will include participants who did not adhere to the treatment for the duration of the study.

    References

    1. Hennekens CH, Buring JE. 1987. Epidemiology in Medicine, Lippincott Williams & Wilkins.
    2. Essebag, V. Genest, J. Suissa, S. Pilote, L. The nested case-control study in cardiology. American Heart Journal. 2003. 146: (4) 581-590.
    3. Murray, C.S. Simpson, B.M. Poletti. G. Woodcock, A. Custovic, A. Dietary intake, atopy, and wheeze at age 3 years: A nested case-control study. Journal of Allergy and Clinical Immunology. 2003. 111 (2) Supplement 2 S203.
    doi:10.1016/S0091-6749(03)80699-1
    4. Kendall JM. Designing a research project: Randomised Controlled trials and their principles. Emerg Med J. 2003; 20(2)164-168.
    5. Kirkwood, B.R. Sterne, J.C. 2003. Essential Medical Statistics pp 263-270. Blackwell Science.
    6. Pocock, S.J. Clinical Trials: A practical approach. Chichester, Wiley, 1984.
    7. Hollis, S. Campbell, F. What is meant by intention to treat analysis? Survey of published randomised controlled trials. BMJ 1999; 319;670-74.

    Further Reading

    Nested case control studies:

    E. Hak, F. Wei, D.E. Grobbee, K.L. Nichol. 2004. Journal of Clinical Epidemiology, 57; ( 9) pp: 875-880.

    Essebag, V. Genest, J. Suissa, S. Pilote, L. The nested case-control study in cardiology. American Heart Journal. 2003.146: (4) 581-590.

    Crossover trials

    Sibbald, B. Roberts, C. Understanding controlled trials. Crossover trials. BMJ 1998. 316 (7146): 1719.

    Factorial trials

    Day, L. Fildes, B. Gordon, I. Fitzharris, M. Flamer, H. Lord, S. Randomised factorial trial of falls prevention among older people living in their own homes. BMJ 2002. 325 (7356): 128.

    General trial design.

    Pocock, S.J. Clinical Trials: A practical approach. Chichester, Wiley, 1984.

    Sibbald, B. Roland, M. Understanding controlled trials: Why are randomised controlled trials important? BMJ 1998, 316:201.

    Altman, D.G. Randomisation. BMJ 1991;302;1481-2.