| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
LETTER TO THE EDITOR |
Boston University, Goldman School of Dental Medicine, Harvard School of Public Health
In the April issue of the Journal of Dental Research, Dr. Moles et al. introduced a method of summarizing research results by cumulative meta-analysis (Moles et al., J Dent Res 84:345–349, 2005). The lessons learned from this report were valuable and informative, but I would like to add a few cautionary words for those performing cumulative meta-analysis.
A cumulative meta-analysis yields results similar to those of regular meta-analysis only if the studies were given equal weights (Berlin et al., 1989). However, when there are substantial irregularities in earlier studies—whether due to ill-conceived study rationale, the adoption of ineffective interventions, or the use of an inappropriate standard deviation—performing cumulative analyses will amplify these biases.
It should be noted that the emergence of significance in a cumulative meta-analysis is a random variable with uncertainties around it (Berkey et al., 1996). Each time cumulative meta-analyses are performed, the effects of the aforementioned irregularities and uncertainties will be replicated. In the presence of less-than-perfect study quality, performing a regular meta-analysis with random-effects modeling may alleviate the effects of irregularities by combining under- and overestimation (Janket et al., 2003).
REFERENCES
Berkey CS, Mosteller F, Lau J, Antman EM (1996). Uncertainty of the time of first significance in random effects cumulative meta-analysis. Controlled Clin Trials 17:357–371.[ISI][Medline]
Berlin JA, Laird NM, Sacks HS, Chalmers TC (1989). A comparison of statistical methods for combining event rates from clinical trials. Stat Med 8:141–151.[ISI][Medline]
Janket S, Baird A, Chuang S, Jones JA (2003). Meta-analysis of periodontal disease and risk of coronary heart disease and stroke. Oral Surg Oral Med Oral Pathol Oral Radiol Endod 95:559–569.[ISI][Medline]
Moles DR, Needleman IG, Niederman R, Lau J (2005). Introduction to cumulative meta-analysis in dentistry: lessons learned from undertaking a cumulative meta-analysis in periodontology. J Dent Res 84:345–349.
To the Editor:
We thank Dr. Janket for the interest in our paper and for raising these issues. They speak to the need for a clear understanding of both meta-analysis and cumulative meta-analysis. We are grateful to the Editor for the opportunity to clarify some misunderstandings.
It is important to realize that a cumulative and a conventional meta-analysis using the same metrics and statistical model (fixed or random effects) will yield identical results. What a cumulative meta-analysis offers in addition is a picture of emerging trends (if any exist). A cumulative meta-analysis does not amplify biases any more than a standard meta-analysis, once it is decided that studies can be combined. With all meta-analyses, cumulative or conventional, the decision regarding whether or not studies are suitable for combination remains crucial. Cumulative meta-analysis of studies ordered chronologically by publication year is a retrospective exercise trying to see when the first instance of statistical significance occurred, to try to learn some lesson. All the factors that affect the outcomes of meta-analyses (quality of the original studies [i.e., likelihood of bias], heterogeneity, lack of available data, etc.) will affect standard meta-analyses and cumulative meta-analyses equally.
Furthermore, studies ordered according to some available covariate may be analyzed by cumulative meta-analysis so that the possible relation of the magnitude of treatment effect with the covariate can be examined. Some of the uses of cumulative meta-analyses have been described by Lau et al.(1995). The impact of heterogeneity and stability of the estimates can be explored by the technique of recursive cumulative meta-analysis (Ioannidis et al., 1999; Ioannidis and Lau, 2001).
REFERENCES
Ioannidis J, Lau J (2001). Evolution of treatment effects over time: empirical insight from recursive cumulative metaanalyses. Proc Natl Acad Sci USA 98:831–836.
Ioannidis JP, Contopoulos-Ioannidis DG, Lau J (1999). Recursive cumulative meta-analysis: a diagnostic for the evolution of total randomized evidence from group and individual patient data. J Clin Epidemiol 52:281–291.[ISI][Medline]
Lau J, Schmid CH, Chalmers TC (1995). Cumulative meta-analysis of clinical trials builds evidence for exemplary medical care. J Clin Epidemiol 48:45–57.[ISI][Medline]
Related articles in JDR:
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| IADR Journals | Advances in Dental Research ® |
| Journal of Dental Research ® | Critical Reviews (1990-2004) |
