Family Practice Vol. 16, No. 6, 619-620
© Oxford University Press 1999
The diagnostic accuracy of Danish GPs in the diagnosis of pigmented skin lesions
Division of Dermatology, Dept of Medicine, Roskilde Hospital, Roskilde, Denmark.
Dr Gregor Jemec, Prss. Alexandrines Alle 18, DK-2920 Charlottenlund, Denmark.
Jemec GBE. The diagnostic accuracy of Danish GPs in the diagnosis of pigmented skin lesions. Family Practice 1999; 16: 619–620.
Received 17 March 1999; Revised 14 May 1999; Accepted 25 June 1999.
Abstract
Background. The GP often has a primary function in assessing pigmented skin lesions in Denmark. No data are available on the diagnostic accuracy of this process.
Objective. We aimed to study the sensitivity, specificity and positive prognostic value of the diagnosis made by 27 trained or trainee GPs.
Method. We tested the diagnostic accuracy of the viewing of colour slides of pigmented skin lesions under standardized conditions at a seminar on skin cancer. Diagnostic accuracy was determined only for the clinically relevant diagnosis of benign or malignant.
Results. The median diagnostic accuracy (sensitivity) for the group as a whole was 0.75 (95% CI 0.65–0.80), the specificity was 0.70 (95% CI 0.68–0.79) and the positive predictive value 0.70 (95% CI 0.62–0.77).
Conclusion. These values are comparable with previously published figures for trainee dermatologists, and it is therefore concluded that ongoing interest rather than basic training is the major determinant for clinical acumen.
Keywords. Clinical competence, diagnostic errors, melanoma, nevi..
Introduction
The treatment of pigmented skin lesions (PSL) clinically suspected of being malignant consists of surgical removal of the lesions. This is usually done either by a dermatologist or by a surgeon. The clinically significant diagnostic choice is therefore between malignant and benign, rather than between specific differential diagnoses of PSL.
In Denmark, the GP has a gate-keeper function with regard to specialist referrals. In the case of PSL, the diagnosis of the GP is therefore also restricted to the simple choice between malignant and benign. Testing of more complex diagnostic skills, e.g. dysplastic naevus versus compund naevus, is thus not immediately relevant for actual practice. The level of clinical accuracy for this PSL-triage is not known.
Materials and methods
A study was conducted among 27 doctors working as whole or part-time GPs (13 whole-time GPs (4 men, 9 women) and 14 part-time (trainee) GPs (7 men, 7 women)) at a seminar on skin cancer and skin surgery to gauge the level of diagnostic accuracy. The study group was therefore biased towards interested physicians. None of the participants in the survey had any formal dermatological experience, and none had had lesions removed themselves. The assessment was done prior to any teaching. To describe immediate recognition, 20 slides of PSL were shown for 20 seconds each, and the audience asked to assess if the lesion was benign or malignant on a form provided. This was thought to be a realistic level of decision to be made by most GPs, as patients suspected of malignancy are generally referred for treatment elsewhere. The PSLs were chosen to represent a range of tumours and morphologies and included 10 malignancies (7 malignant melanoma (3 typical/4 atypical), 3 pigmented basal cell carcinomas (2 typical/1atypical)) and 10 benign lesions (8 pigmented naevi (5 typical/3 atypical), 1 skin tag, 1 seborheic keratosis).
The diagnostic accuracy was assessed by calculating sensitivity (= true positive/true positive + false negative), specificity (= true negative/true negative + false positive) and positive predictive value (= true positive/true positive + false positive) for each GP, and presenting the median values with 95% confidence intervals— see Table 1
. The values were compared the Bonferroni Multiple comparisons test, and linear regression for analysis of correlation between age and accuracy.
|
Results
The median diagnostic accuracy (sensitivity) for the group as a whole was 0.75 (95% CI 0.65–0.80), the specificity was 0.70 (95% CI 0.68–0.79) and the positive predictive value 0.70 (95% CI 0.62–0.77). Correlations between postgraduate age and sensitivity, specificity and positive predictive value were not significant. See Table 1
for results.
Discussion
Diagnostic accuracy for PSL had not been described previously in Denmark, but was found to be comparable to previously published studies from dermatologists or plastic surgeons in other countries. Sample size and selection may have influenced the result, as only GPs attending a skin cancer seminar, i.e. suggesting a preexisting interest in PSL, were studied. The results may therefore be interpreted to represent a better than average result. Full-time GPs were older (P < 0.001) and had a greater postgraduate age (P < 0.001) than part-timers (trainees), but no difference in sensitivity, specificity or positive predictive value was found. Early studies in the UK suggested that the overall accuracy as described by the sensitivity was only about 50% for PSL, although a more complex question was asked than in this study, as the physicians in question were asked to give a specific diagnosis e.g. dysplastic naevi, rather than only benign/ malignant.1 Benign/malignant is thought to be a more realistic end-point, as suspicion of malignancy leads to onward referral and therefore no decision has to be made regarding the ultimate treatment. Different levels of accuracy may be expected from different specialities, with plastic surgeons and dermatologists having higher levels than other professions, reaching up to 90% accuracy for both benign and malignant lesions.2–4 The level of accuracy among junior dermatologists has been given as 56–62%.4 In contrast, the clinical diagnostic accuracy for malignant melanoma among GPs in the UK has been reported to be as low as 17%.5
This may be explained partly by the frequency with which the GPs are exposed to diagnostic decisions regarding PSL. The contribution of benign tumours to the case-mix of dermatological departments suggest that they constitute a large proportion in the UK while they constitute a negligible proportion in Denmark,6,7 and it is therefore hypothesized that simple exposure to clinical decisions regarding PSL among GPs is lower in the UK.
Common sense and a previous study imply that diagnostic accuracy increases with experience.4 Increasing postgraduate age did not, however, correlate with the diagnostic accuracy in the sample studied here, which may be explained by individual differences overshadowing the small sample size used.
In conclusion, this study of diagnostic accuracy for PSL among Danish GPs and trainees interested in skin cancer suggest that it is comparable to previously published figures for trainee dermatologists, and that ongoing interest rather than basic training are determinant for clinical acumen.
References
1 Curley RK, Cook MG, Fallowfield ME et al. Accuracy in clinically evaluating pigmented lesions. Br Med J 1989; 299: 16–18.
2 DeCoste SD, Stern RS. Diagnosis and treatment of nevomelanocytic lesions of the skin. A community-based study. Arch Dermatol 1993; 129: 57–62.[Abstract]
3 Hallock GG, Lutz DA. Prospective study of the accuracy of the surgeons' diagnosis in 2000 excised skin tumours. Plast Reconstr Surg 1998; 101: 1255–1261.[Medline]
4 Morton CA, Mackie RM. Clinical accuracy of the diagnosis of cutaneous malignant melanoma. Br J Dermatol 1998; 138: 283–287.[ISI][Medline]
5 Khorshid SM, Pinney E, Bishop JA. Melanoma excision by general practitioners in north-east Thames Region, England. Br J Dermatol 1998; 138: 412–417.[Medline]
6 Harris DWS, Benton EC, Hunter JAA. The changing face of dermatology out-patient referrals in the south-east of Scotland. Br J Dermatol 1990; 123: 745–750.[Medline]
7 Jemec GBE, Thorsteinsdottir H, Wulf HC. The changing referral pattern in Danish dermatology—Rigshospitalet, Copenhagen, 1986–1995. Int J Dermatol 1997; 36: 453–456.[Medline]
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||