Michael Hardy (talk | contribs) orphan |
m Date maintenance tags and general fixes, removed orphan tag |
||
| Line 1: | Line 1: | ||
{{orphan}} |
|||
The '''''n'' = 1 fallacy''' is a common error in the analysis of [[epidemiology|epidemiologic]] studies, first defined in 2003 <ref>von Seidlein L, Greenwood BM. Mass administrations of antimalarial drugs. ''Trends Parasitol 2003;19(10):452–60''.</ref>. |
The '''''n'' = 1 fallacy''' is a common error in the analysis of [[epidemiology|epidemiologic]] studies, first defined in 2003 <ref>von Seidlein L, Greenwood BM. Mass administrations of antimalarial drugs. ''Trends Parasitol 2003;19(10):452–60''.</ref>. |
||
=== Background === |
=== Background === |
||
The unit at which a [[public health intervention]] (e.g. a [[vaccination]] campaign, a mass drug administration) is applied can be an individual or a group of individuals such as a household cluster or a village. If the effect of the intervention can only be observed at a village level, for example an intervention to block the transmission of [[malaria]], a [[cluster randomized trial]] is needed. Depending upon the degree of similarity between villages and the size of the villages, this may require a much larger sample than an individually randomised trial <ref>Jaffar S, Leach A, Hall AJ, Obaro S, McAdam KP, Smith PG, et al. Preparation for a pneumococcal vaccine trial in The Gambia: individual or community randomisation? Vaccine 1999;18(7–8):633–40.</ref>,<ref>Smith PG, Morrow R. Field Trials of Health Interventions in Developing Countries: a Toolbox. London: Macmillan, 1996.</ref>. The additional sample in a cluster randomised study compared to an individually randomised trial is sometimes called the design effect<ref>Donner A, Klar N. Design and Analysis of Cluster Randomization Trials in Health Research. London: Arnold, 2000.</ref>. |
The unit at which a [[public health intervention]] (e.g. a [[vaccination]] campaign, a mass drug administration) is applied can be an individual or a group of individuals such as a household cluster or a village. If the effect of the intervention can only be observed at a village level, for example an intervention to block the transmission of [[malaria]], a [[cluster randomized trial]] is needed. Depending upon the degree of similarity between villages and the size of the villages, this may require a much larger sample than an individually randomised trial <ref>Jaffar S, Leach A, Hall AJ, Obaro S, McAdam KP, Smith PG, et al. Preparation for a pneumococcal vaccine trial in The Gambia: individual or community randomisation? Vaccine 1999;18(7–8):633–40.</ref>,<ref>Smith PG, Morrow R. Field Trials of Health Interventions in Developing Countries: a Toolbox. London: Macmillan, 1996.</ref>. The additional sample in a cluster randomised study compared to an individually randomised trial is sometimes called the design effect<ref>Donner A, Klar N. Design and Analysis of Cluster Randomization Trials in Health Research. London: Arnold, 2000.</ref>. |
||
| Line 15: | Line 13: | ||
{{reflist}} |
{{reflist}} |
||
[[Category: |
[[Category:Sciences]] |
||
[[Category: |
[[Category:Statistics| ]] |
||
[[Category: |
[[Category:Epidemiology]] |
||
[[Category: |
[[Category:Randomised trials]] |
||
[[Category: |
[[Category:Cluster randomisation]] |
||
Revision as of 21:42, 13 June 2009
The n = 1 fallacy is a common error in the analysis of epidemiologic studies, first defined in 2003 [1].
Background
The unit at which a public health intervention (e.g. a vaccination campaign, a mass drug administration) is applied can be an individual or a group of individuals such as a household cluster or a village. If the effect of the intervention can only be observed at a village level, for example an intervention to block the transmission of malaria, a cluster randomized trial is needed. Depending upon the degree of similarity between villages and the size of the villages, this may require a much larger sample than an individually randomised trial [2],[3]. The additional sample in a cluster randomised study compared to an individually randomised trial is sometimes called the design effect[4].
Description
The n = 1 fallacy occurs when the intervention is allocated at the village or cluster level but when the sample size is calculated on the basis of an individual analysis and when results are presented on this basis.
Example
If 30 cases of malaria occur in village A, which has a population of 300, where a mass drug administration has been undertaken, but 150 cases occur in village B, which did not receive the intervention and which is of the same size as village A, then there is an intuitive tendency to calculate a chi square test comparing 30/300 with 150/300 giving a highly statistically significant result. However, the correct comparison is between an incidence of 10% in village A and of 50% in village B, a result which does not allow any statistically meaningful interpretation. Problems arising from one-to-one comparison of this kind were eloquently described 25 years ago, but continue to bedevil the evaluation of public health interventions[5].
References
- ^ von Seidlein L, Greenwood BM. Mass administrations of antimalarial drugs. Trends Parasitol 2003;19(10):452–60.
- ^ Jaffar S, Leach A, Hall AJ, Obaro S, McAdam KP, Smith PG, et al. Preparation for a pneumococcal vaccine trial in The Gambia: individual or community randomisation? Vaccine 1999;18(7–8):633–40.
- ^ Smith PG, Morrow R. Field Trials of Health Interventions in Developing Countries: a Toolbox. London: Macmillan, 1996.
- ^ Donner A, Klar N. Design and Analysis of Cluster Randomization Trials in Health Research. London: Arnold, 2000.
- ^ Blum D, Feachem R. Measuring the impact of water supply and sanitation investments on diarrheal diseases: problems of methodolgy. Int J Epidemiol 1983;12(3):357–65.