Increased Incidence of Choroidal Malignant Melanoma and the Use of Inappropriate Statistical Methods

Increased Incidence of Choroidal Malignant Melanoma and the Use of Inappropriate Statistical Methods

272 AMERICAN JOURNAL OF OPHTHALMOLOGY Certain factors are of utmost importance in obtaining excellent and sustained mydriasis. Perhaps our comments ...

259KB Sizes 0 Downloads 9 Views

272

AMERICAN JOURNAL OF OPHTHALMOLOGY

Certain factors are of utmost importance in obtaining excellent and sustained mydriasis. Perhaps our comments on these factors will explain Dr. Brenner's incomplete response with tropicamide. First, drugs combined in solution achieve higher intraocular concentrations after instillation of a single eyedrop than when instilled as drops of separate solutions within five minutes because of dilution occurring with multiple drops. Second, prior instillation of a drop of proparacaine 0.5% (Ophthaine, Ophthetic, Alcaine) improves absorption of the administered mydriatics. We advise waiting a minimum of one to two minutes after instilling proparacaine before instilling the mydriatic combination solution. Also we advise gently wiping the closed eyelids to remove the excess proparacaine solution that may be retained in the cul-de-sac before instilling the mydriatic solution, to avoid dilution or overflow. Instillation of the drops by the "pouch method" introduced by Frederick Fraunfelder, and described in detail in our article, assures retention in the cul-de-sac of a maximum amount of fluid and increases the contact time between the solution and the cornea. We cannot overemphasize the importance of the proper method of eyedrop instillation in obtaining wide and sustained pupillary dilation. If the ophthalmologist does not instill the eyedrops himself, he should explicitly teach the proper method to his office personnel. We believe that Dr. Brenner also can achieve excellent and sustained mydriasis by administering the phenylephrinetropicamide combination solution in the manner we described. The use of tropicamide rather than cyclopentolate in combination with phenylephrine permits accommodation and mydriasis to return to normal faster. LEONARD APT, M.D. AND

M.D. Los Angeles, California

ANDREW HENRICK,

AUGUST, 1980

Increased Incidence of Choroidal Malignant Melanoma and the Use of Inappropriate Statistical Methods Editor: The report of a cluster of five choroidal melanomas among the present or former employees of a chemical plant by Dr. Albert and his colleagues, "Increased incidence of choroidal malignant, melanoma occurring in a single population of chemical workers" (Am. J. Ophthalmol. 89:323, 1980), requires that a decision, with far-reaching ethical and scientific consequences, be made that this was a chance aggregation, or that the cases resulted from the action of an unrecognized occupational carcinogen. In the absence of any way of predicting for a specific human tumor the carcinogenicity of chemical compound, a necessary element in this decision is an estimate of the probability that this cluster of cases occurred by change. The authors address the problem, and state that "We have determined that this represents a statistically significantly greater than expected occurrence of choroidal melanomas in the plant white male population." They derived expected numbers of cases from two standard sets of incidence data and obtained estimates of the probability for a plant of this appropriate size of 1/39 and 1/77. Their method was based on the assumption of a single trial of the data; that the particular plant had been specified in advance of knowledge of the actual occurrence of choroidal melanomas among its workers. Such a situation might have occurred if they had been led to the particular plant in response to animal studies or consideration of the actual chemicals handled. Indeed, the identification of the plant rested on an indefinite number of occasions where people looked through plant records, found nothing of note, and took no futther steps. Each such negative trial adds to the probability fraction. Thus the

VOL. 90, NO. 2

CORRESPONDENCE

authors' finding of a single plant in the United States with this number of cases when the single trial expectation is as stated is not surprising. The statistical way out of this difficulty of covert multiple testing is to take the complete set of cases for a large population, and test the randomness of their distribution by standard methods. This has been done for endemic disease, for example leukemia, but can clearly miss small point sources that provide only a small proportion of the total cases. Nevertheless it is mandatory to deal with the effect of multiple trials, whether overt or covert, in statistical analyses where the positive findings have been arrived at by other means than a systematic survey. The alternative is to seek for more data. A continuation of the occurrence of cases in the plant would clinch the matter, as would discovery of similar groups of cases in comparable plants. Without confirmatory data, or the conduct of appropriate statistical tests, the proposition that the current findings are of statistical significance is an error. JOHN A. H. LEE, M.D. Seattle, Washington Reply Editor: Dr. Lee makes the point that out of all the industrial plants in the United States, it is likely that at least one of them might fortuitously have had five cases of choroidal malignant melanoma and that our study may have been but a description of such a fortuitously occurring cluster, identified not as a result of a systematic survey whose hypotheses had been formulated a priori, but rather as a result of multiple overt and covert examinations of different data bases by different investigators. The implication would be that we have studied, after its discovery, the

273

cluster of choroidal malignant melanoma cases that could have been expected to have occurred by chance alone. Dr. Lee additionally notes that the alternative to dealing with the effect of these multiple overt and covert data examinations, on the calculation of the probability fraction upon which is based the inference that a statistically significant excess number of choroidal malignant melanomas has occurred, is " to seek more data. A continuation of the occurrence of cases in the plant would clinch the matter ... " Although our account of the chronology of events that led to our data collection may not have made it sufficiently clear, this is precisely what we have observed and what we have attempted to describe. Our efforts began with the identification in late 1976 of two cases of choroidal malignant melanoma from a list of purported ocular tumors occurring at the plant. We submitted to the National Institute for Occupational Safety and Health a contract proposal, dated January 27, 1977 (quoted from our proposal), to undertake "a screening examination of the Belle chemical workers" to address the question, "Is there an increased incidence of malignant melanomas of the choroid, ciliary body or iris in this population?" About this same time, we reported the two cases in a letter to the editor of the New England Journal of Medicine, the implicit statement being that we would hope that additional cases of choroidal malignant melanoma discovered at other chemical plants might be reported if the medical community were made aware of our observations. This resulted in a letter, dated April 7, 1977, from physicians with the National Institutes of Health reporting to us Case 3. Later in 1977, review of company medical records uncovered Case 4. Case 5 was discovered in April 1978 during our ophthalmologic survey. We believe that these last three cases, discovered after our initiation of