The objective appears to have been to evaluate different diagnostic techniques for differentiating between moles and melanomas.

MEDLINE was searched from 1966 to 1996 using the MeSH terms 'melanoma' and 'skin neoplasms' combined with 'physical examination', 'sensitivity', 'specificity', 'observer variation', 'mass screening' and 'self examination'. An additional MEDLINE search was undertaken, but no details were given. Current Contents was reviewed using the terms 'melanoma', 'skin cancer' and 'mass screening'. The references of identified articles were also examined.

Study designs of evaluations included in the review

The authors did not report the study designs included. The only inclusion criterion was that studies were included if the level of evidence was graded as C or above, according to the ABCD(E) checklist for detecting cutaneous melanoma.

Specific interventions included in the review

No inclusion criteria were specified. The included studies were of historical feature assessment, physical examination, ABCD(E) checklist (US) for detecting cutaneous melanoma, and a revised 7-point checklist(UK). The ABCD(E) checklist considered asymmetry (A), border irregularity (B), colour variegation (C), diameter (D) and elevation(E). The revised 7-point checklist considered 3 major criteria (change in size, shape and colour) and 4 minor criteria (inflammation, crusting or bleeding, sensory change, diameter). The thresholds defining a positive test, in terms of the number of criteria met, varied between the included studies.

Reference standard test against which the new test was compared

No inclusion criteria were reported. The authors stated that the 'gold' standard for comparison is histopathological evaluation of excised tissue. The reference standard used in the included studies was unclear.

Participants included in the review

No inclusion criteria were specified. The authors did not provide any details of the participants.

Outcomes assessed in the review

No inclusion criteria were specified. The outcome measures reported in the review were the sensitivity, specificity, and likelihood ratios for positive and negative test results.

How were decisions on the relevance of primary studies made?

The authors did not state how the papers were selected for the review, or how many reviewers performed the selection.

The quality of published articles was assessed using the Sackett-Goldsmith grading scheme (see Other Publications of Related Interest) as follows.

Grade A: independent, blind comparison of sign or symptom with a 'gold' standard of diagnosis among a large number of consecutive patients suspected of having the target condition.

Grade B: independent, blind comparison of sign or symptom with a 'gold' standard of diagnosis among a small number of consecutive patients suspected of having the target condition.

Grade C: independent, blind comparison of sign or symptom with a 'gold' standard of diagnosis among nonconsecutive patients suspected of having the target condition; or nonindependent comparison of sign or symptom with a 'gold' standard of diagnosis among samples of patients who obviously have the target condition plus, perhaps, normal individuals; or nonindependent comparison of sign or symptom with a standard of uncertain validity.

The authors did not state how the papers were assessed for quality, or how many reviewers performed the quality assessment.

The authors did not state how the data were extracted for the review, or how many reviewers performed the data extraction.

How were the studies combined?

The 95% confidence intervals (CIs) for test performance characteristics were reported for each included study; those not reported in the original articles were calculated from the available data, where possible. The studies were summarised in a narrative review.

How were differences between studies investigated?

Differences in the scope of the included studies were discussed in the review.

Twelve studies were included. Two studies evaluated examiners' precision for specific features of benign pigmented lesions (number of lesions and patients were not given); 4 studies assessed the accuracy of the checklists (557 lesions); 12 studies evaluated the accuracy for detecting the presence or absence of melanoma (minimum of 10,936 lesions).

All 12 studies were of grade C quality.

Precision estimates among physicians for four features of the ABCD(E) checklist found that benign pigmented lesions are considered fair-to-moderate. Precision estimates for global assessments of the presence or absence of melanoma were not available.

One study found the sensitivity for the ABCD checklist to be 92% (95% CI: 82, 96); the specificity was not reported. The other found a sensitivity of 100% (95% CI: 54, 100) and a specificity of 98% (95% CI: 95, 99).

The 7-point checklist has been reported to have a sensitivity ranging from 79% (95% CI: 70, 85) for an unreported specificity to 100% (95% CI: 94, 100) for a specificity of 37% (95% CI: 28, 46). One further study reported a specificity alone of 30% (95% CI: 21, 39).

Physicians' global assessments for detecting the presence or absence of melanoma were estimated to have a specificity of 96 to 99%, and a sensitivity of 50 to 97%.

Nondermatologists' examinations appear to be less sensitive than examinations performed by dermatologists.

The usefulness of the ABCD checklist or revised 7-point checklist to distinguish melanoma from benign lesions has not been fully established. Both checklists appeared to be sensitive diagnostic aids when evaluating individual lesions and they rarely classified a melanoma as a benign lesion. However, the revised 7-point checklist lacked specificity, resulting in benign lesions being classified as potentially malignant. The specificity of the ABCD checklist was less well described. Better study designs are necessary to define the operating characteristics of physicians' examination for detecting the presence or absence of melanoma.

There was evidence of a search, inclusion criteria and quality assessment, but since the review lacked methodological detail it would be difficult to replicate. No attempt was made to identify unpublished literature and it was unclear whether non-English language publications were included; both of these issues raise the possibility of publication bias. In addition, search terms specific to diagnostic accuracy studies were used; these may limit the sensitivity of the searches. There were insufficient details of the study design and size of the primary studies, and tabulated details were not provided for all of the studies referenced in the review.

Given the limitations it is difficult to evaluate the strength of the authors' conclusions, but they appear to follow from the results presented.

Practice: The authors did not state any implications for practice.

Research: The authors stated that better study designs are necessaryto determine the operating characteristics of physicians' examinations for detecting the presence or absence of melanoma.

Holleman R, Simel DL. Does the clinical examination predict airflow limitation? JAMA 1995;273:313-9.

This is a critical abstract of a systematic review that meets the criteria for inclusion on DARE. Each critical abstract contains a brief summary of the review methods, results and conclusions followed by a detailed critical assessment on the reliability of the review and the conclusions drawn.