Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Atopic Signs and Symptoms: Assessing the ‘Atopy Score’ Concept

Atopic Signs and Symptoms: Assessing the ‘Atopy Score’ Concept IntroductionAtopic dermatitis can often be diagnosed ‘at first sight’, if its appearance is typical and – obviously – if dermatitis is active on examination, which is often not the case in this relapsing disease. In other cases with uncharacteristic findings and an equivocal previous history, this diagnosis may be less straightforward. The diagnosis ‘atopic dermatitis’, however, has important implications not only for the immediate treatment and advice: in view of the well-known propensity of the skin to develop irritant dermatitis it is also important for pre-placement counselling of young adults and for the evaluation of workers who have developed an occupational skin disease. Thus, there are good reasons to strive for an improvement and for standardisation of the potentially difficult diagnosis of atopic dermatitis.As such an instrument, founded on the set of atopic criteria elaborated by Hanifin and Rajka [1], scoring concepts have been suggested by Svensson et al. [2], further elaborated, based on a larger case-control study, by Diepgen et al. [3, 4]and Diepgen and Fartasch [5]and later validated by the same authors [6]. Basically, the strength of association between flexural atopic dermatitis as ‘gold standard’ criterion and any one of the minor criteria has been quantified in these studies, and point values have been assigned to these criteria according to the observed strength of association. The individual sum of these respective point values showed good discriminating properties between cases and controls. However, so far no study has attempted to apply this score to a non-hospital-based sample of the general population not selected for a certain morbidity, which would reflect the proposed broad use for pre-employment screening and advice. Results of such an analysis derived from a large prospective cohort study are presented in this paper.MethodsDuring the initial examination of participants of the prospective cohort study POSH (Prevention of Occupational Skin Disease in Hairdressers) [7, 8, 9], a self-administered questionnaire relating to family and personal history (atopy, intolerances), leisure activities and occupational tasks was first checked by the observer; anamnestic items concerning atopy are listed in table 1 and partly in table 2. The gold standard criterion was anamnestically determined by the following question: ‘Have you had – as a child or adolescent – a rash in the flexures, e.g. popliteal flexure or hollow of the knee (neurodermatitis or endogenous eczema)?’ – for the original German wording see Uter [8]. Concerning the family history, only first-degree relatives were considered. Thereafter, a complete examination of the skin was performed. Due to this temporal order, the respective observer was blind to physical findings when checking the history but not vice versa. Each participant was questioned and examined by one, and only one, of the dermatological observers.Table 1Prevalence (%) of anamnestic atopic (minor) criteria in persons with (previous) flexural eczema: association quantified with the odds ratio (OR) with the respective approximate 95% confidence interval (CI)Table 2Prevalence of clinical signs of atopy in persons with (previous) flexural eczema: association quantified with the odds ratio (OR) with the respective approximate 95% confidence interval (CI)Of the items contained in the atopy score developed by Diepgen et al. [4]both photophobia and laboratory results were omitted: the first, because validity was found poor in a pilot study; the second, because in our field study routine serological tests were not considered cost-efficient and probably unnecessary, as indicated by Diepgen et al. [6]in a later re-analysis. Statistical analyses presented here focus on (1) the distribution of relevant parameters in our population-based sample and (2) the association between atopic minor criteria and (past) flexural dermatitis, considered as manifestation of atopic dermatitis. Observations with values ‘unclear/unknown’ or missing are excluded from calculations in contingency tables and prevalence computations. The association between minor and the major criterion mentioned above is quantified by the odds ratio accompanied by its 95% confidence intervals (CI); its statistical significance is assessed by Fisher’s exact test. Standard measures of diagnostic test performance like sensitivity, specificity and predictive values (PV) accompanied by their 95% CI are calculated to evaluate the properties of the atopy score as a diagnostic tool for atopic dermatitis. For a graphical visualisation of the interdependence of sensitivity and specificity of the dichotomised atopy score for different cutpoints a receiver operating characteristic curve was used. For data management and analyses, the statistical software package SAS (version 6.12, SAS Institute Inc., Cary, N.C., USA) was used.ResultsIn the 3 years of recruitment, altogether 2,352 hairdressing apprentices were examined, on average 8 weeks after the start of training. The mean age was 17.3 years, with a standard deviation of 2.7. The proportion of females was 93.8%. Altogether 7.5% of participants had suffered from flexural atopic eczema earlier, including 1.3% who were found to have flexural eczema – of usually mild degree – on examination.The prevalence of single atopic features in the whole cohort is shown in table 1 for anamnestic items and table 2 for clinical signs (which are partly also anamnestic), together with prevalences found in the groups with flexural dermatitis. For some variables more than 5% of observations had to be discarded due to missing data and/or values unclear/unknown, namely family history of flexural dermatitis, cradle cap, wool intolerance, xerosis, keratosis pilaris, hyperlinear palms, periorbital darkening, itch after shower.A significant association (p < 0.05) between ‘flexural dermatitis’ and most single atopic features can be noted (tables 1, 2). The item ‘itch after shower’, which had yet not been included in the lists of atopic criteria considered here [1, 2, 3, 4], is more strongly associated with either type of dermatitis than many other, especially clinical, features.The distribution of the atopy score in the group with and without flexural dermatitis is significantly different between the two subsets (p < 0.0001, Wilcoxon-Mann-Whitney test). The distribution of point values of the atopy score in the groups with (without) flexural dermatitis is depicted in figure 1, in a fashion identical to Diepgen et al. [5]; half point values have been rounded up for this graph. The diagnostic performance of the score is illustrated by calculating its respective sensitivity and specificity, using inclusive cutpoints between 11 and 7 points of the atopy score (table 3), and by a receiver operating characteristic curve (fig. 2) depicting graphically the behaviour of these parameters over the range of possible cut-off values.Table 3Measures of the diagnostic performance for the atopy score (dichotomised at various cutpoints) to detect atopic flexural dermatitis: sensitivity, specificity and PVs of a ‘high’ atopy scoreFig. 1Distribution (% in class) of the point values of the atopy score classified as suggested in Diepgen and Fartasch [5] – not considering photophobia and laboratory results – in 176 persons with versus 2,176 persons without (previous) flexural dermatitis.Fig. 2Receiver operating characteristic curve for different cut-off values of the atopy score.DiscussionTo our knowledge this is the first study which undertook to validate the atopy score concept not in a clinical setting, but in a population selected solely for their occupation. Additionally, relevant data were collected not primarily for this purpose, but for prospectively evaluating the prognostic significance of the various atopic features. Thus, the evaluation of the diagnostic significance of these features, i.e. their association with atopic dermatitis, was not the focus of the study, which made the setting more similar to a clinical situation, where such diagnostic signs are applied, and not assessed, which may influence bias in this respect (see below).While for most atopic features a significant association with the diagnostic gold standard ‘(previous) flexural dermatitis’ is found in this analysis, which is in accordance with the studies quoted here, this association is by far not as strong as found in the case-control studies of Svensson et al. [2], Diepgen et al. [3, 4, 6]and Diepgen and Fartasch [5], respectively. Accordingly, the discriminating power of the atopy score [3]is much poorer than in the original studies (fig. 1). Possible reasons for this poor diagnostic performance could be:(1) In groups of patients treated for atopic eczema (current [2]or current or previous [5, 6]) in a hospital facility, the average disease severity may have been greater than in our young, unselected cohort, which included relatively few persons with only slight current dermatitis. Thus, in clinical cases there may have been more and probably also more clear-cut atopic features. Such a difference in the disease spectrum of study populations leading to discrepancies in the performance of diagnostic tests has been repeatedly described in other contexts [10]and the special term ‘spectrum bias’ has been coined for this phenomenon [11], which has already been highlighted in relation to differences between hospital and community validation of diagnostic criteria of atopic dermatitis [12].(2) Many of the (clinical) features used as building blocks of the score are intrinsically ‘soft’ [13], and probably even the use of operational definitions (as employed here) or, e.g., colour photographs as reference will achieve standardisation only to a certain, limited extent. This will obviously limit the diagnostic value of the respective single features and of the atopy score based on these features.(3) Facing this potentially poor reproducibility, a certain type of ‘investigator bias’ has to be faced, which may easily occur if the observer cannot be blinded to the gold standard criterion: during physical examination for the presence of clinical atopic features, the observer will inevitably note flexural dermatitis too. In view of the common dermatological notion of an association between atopic features and atopic dermatitis, an observer may in this case be more inclined to regard a finding as positive, which would otherwise have been diagnosed as negative or equivocal. Being non-blind towards the previous history of atopic eczema – as in our study – may introduce bias in a similar way. While the direction of this bias is probably the same in all studies, its extent may have been higher in the clinical studies aiming at the assessment of this very association than in our study. This assumption is supported by a comparison between odds ratios found in our study (tables 1, 2) and those reported in Diepgen and Fartasch [5]revealing a considerably greater difference for clinical than for anamnestic features.(4) The omission of three items contained in the original atopy score, namely photophobia, total serum IgE and SX-1, does in our opinion not contribute to impaired validity, because (i) the inclusion of an item with poor validity cannot be expected to contribute to increased overall validity and (ii) laboratory results can apparently be omitted without weakening the score properties [6]. Probably the threshold values as defined by Diepgen et al. (more than 10 points – not considering laboratory results: ‘atopic skin predisposition’, 8–9 points: ‘... unclear’, below 8 points: ‘... unlikely/not present’ [4]) cannot be transferred to the score values as calculated here as a consequence of these omissions. However, the discriminating power is poor also with lowered threshold point values.In conclusion, the atopy score is useful to summarise the multitude of atopic features. However, as its diagnostic performance is poor (i.e. low sensitivity and specificity with any chosen cutpoint) the relevance of certain point values found individually should not be overestimated. Furthermore, a certain proportion of patients with atopic dermatitis has no flexural involvement but dermatitis of other sites. These persons will ‘dilute’ the association between atopic features and atopic dermatitis defined solely as flexural dermatitis; however, for reasons of comparability, we used the same definition of the gold standard as in the preceding score studies.In addition, the usefulness of any diagnostic tool in practice does not only depend on its sensitivity and specificity, but also on the prevalence of the disease or condition in the population to which it is applied: the positive PV is dependent on the prevalence of the disease in the population studied and will thus be low if the disease is rare, while the negative PV will be impaired if a disease is very common (see also Popescu et al. [14]). For example, assuming a fairly low prevalence of 7.5% as found in this sample of the population [which may reflect (self)selection processes as a consequence of the high irritant profile found in this occupation] and a high sensitivity and specificity of 93.2 and 96.0%, respectively, as found in the original study (cutpoint 10 points [5]), the positive PV, i.e. the posttest probability of actually suffering from the disease, would be 65.4% in our population setting. In the setting of a dermatitis clinic with an assumed prevalence of atopic dermatitis of 75% among its patients, the positive PV would be 98.6%, under identical assumptions concerning sensitivity and specificity. With the much lower values for sensitivity and specificity found in the present validation study, the positive PV is even much lower in the population setting (table 3); for instance, a positive PV of 20.9% for a score of 11 and above means that for every truly diseased person 4 others with such score values are ‘false-positives’. – Note that such a subset of the population represents one of the target groups for the application of the atopy score. http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Dermatology Karger

Atopic Signs and Symptoms: Assessing the ‘Atopy Score’ Concept

Dermatology , Volume 202 (1): 5 – Jan 1, 2001

Loading next page...
 
/lp/karger/atopic-signs-and-symptoms-assessing-the-atopy-score-concept-0xJwDkw01s

References

References for this paper are not available at this time. We will be adding them shortly, thank you for your patience.

Publisher
Karger
Copyright
© 2001 S. Karger AG, Basel
ISSN
1018-8665
eISSN
1421-9832
DOI
10.1159/000051576
Publisher site
See Article on Publisher Site

Abstract

IntroductionAtopic dermatitis can often be diagnosed ‘at first sight’, if its appearance is typical and – obviously – if dermatitis is active on examination, which is often not the case in this relapsing disease. In other cases with uncharacteristic findings and an equivocal previous history, this diagnosis may be less straightforward. The diagnosis ‘atopic dermatitis’, however, has important implications not only for the immediate treatment and advice: in view of the well-known propensity of the skin to develop irritant dermatitis it is also important for pre-placement counselling of young adults and for the evaluation of workers who have developed an occupational skin disease. Thus, there are good reasons to strive for an improvement and for standardisation of the potentially difficult diagnosis of atopic dermatitis.As such an instrument, founded on the set of atopic criteria elaborated by Hanifin and Rajka [1], scoring concepts have been suggested by Svensson et al. [2], further elaborated, based on a larger case-control study, by Diepgen et al. [3, 4]and Diepgen and Fartasch [5]and later validated by the same authors [6]. Basically, the strength of association between flexural atopic dermatitis as ‘gold standard’ criterion and any one of the minor criteria has been quantified in these studies, and point values have been assigned to these criteria according to the observed strength of association. The individual sum of these respective point values showed good discriminating properties between cases and controls. However, so far no study has attempted to apply this score to a non-hospital-based sample of the general population not selected for a certain morbidity, which would reflect the proposed broad use for pre-employment screening and advice. Results of such an analysis derived from a large prospective cohort study are presented in this paper.MethodsDuring the initial examination of participants of the prospective cohort study POSH (Prevention of Occupational Skin Disease in Hairdressers) [7, 8, 9], a self-administered questionnaire relating to family and personal history (atopy, intolerances), leisure activities and occupational tasks was first checked by the observer; anamnestic items concerning atopy are listed in table 1 and partly in table 2. The gold standard criterion was anamnestically determined by the following question: ‘Have you had – as a child or adolescent – a rash in the flexures, e.g. popliteal flexure or hollow of the knee (neurodermatitis or endogenous eczema)?’ – for the original German wording see Uter [8]. Concerning the family history, only first-degree relatives were considered. Thereafter, a complete examination of the skin was performed. Due to this temporal order, the respective observer was blind to physical findings when checking the history but not vice versa. Each participant was questioned and examined by one, and only one, of the dermatological observers.Table 1Prevalence (%) of anamnestic atopic (minor) criteria in persons with (previous) flexural eczema: association quantified with the odds ratio (OR) with the respective approximate 95% confidence interval (CI)Table 2Prevalence of clinical signs of atopy in persons with (previous) flexural eczema: association quantified with the odds ratio (OR) with the respective approximate 95% confidence interval (CI)Of the items contained in the atopy score developed by Diepgen et al. [4]both photophobia and laboratory results were omitted: the first, because validity was found poor in a pilot study; the second, because in our field study routine serological tests were not considered cost-efficient and probably unnecessary, as indicated by Diepgen et al. [6]in a later re-analysis. Statistical analyses presented here focus on (1) the distribution of relevant parameters in our population-based sample and (2) the association between atopic minor criteria and (past) flexural dermatitis, considered as manifestation of atopic dermatitis. Observations with values ‘unclear/unknown’ or missing are excluded from calculations in contingency tables and prevalence computations. The association between minor and the major criterion mentioned above is quantified by the odds ratio accompanied by its 95% confidence intervals (CI); its statistical significance is assessed by Fisher’s exact test. Standard measures of diagnostic test performance like sensitivity, specificity and predictive values (PV) accompanied by their 95% CI are calculated to evaluate the properties of the atopy score as a diagnostic tool for atopic dermatitis. For a graphical visualisation of the interdependence of sensitivity and specificity of the dichotomised atopy score for different cutpoints a receiver operating characteristic curve was used. For data management and analyses, the statistical software package SAS (version 6.12, SAS Institute Inc., Cary, N.C., USA) was used.ResultsIn the 3 years of recruitment, altogether 2,352 hairdressing apprentices were examined, on average 8 weeks after the start of training. The mean age was 17.3 years, with a standard deviation of 2.7. The proportion of females was 93.8%. Altogether 7.5% of participants had suffered from flexural atopic eczema earlier, including 1.3% who were found to have flexural eczema – of usually mild degree – on examination.The prevalence of single atopic features in the whole cohort is shown in table 1 for anamnestic items and table 2 for clinical signs (which are partly also anamnestic), together with prevalences found in the groups with flexural dermatitis. For some variables more than 5% of observations had to be discarded due to missing data and/or values unclear/unknown, namely family history of flexural dermatitis, cradle cap, wool intolerance, xerosis, keratosis pilaris, hyperlinear palms, periorbital darkening, itch after shower.A significant association (p < 0.05) between ‘flexural dermatitis’ and most single atopic features can be noted (tables 1, 2). The item ‘itch after shower’, which had yet not been included in the lists of atopic criteria considered here [1, 2, 3, 4], is more strongly associated with either type of dermatitis than many other, especially clinical, features.The distribution of the atopy score in the group with and without flexural dermatitis is significantly different between the two subsets (p < 0.0001, Wilcoxon-Mann-Whitney test). The distribution of point values of the atopy score in the groups with (without) flexural dermatitis is depicted in figure 1, in a fashion identical to Diepgen et al. [5]; half point values have been rounded up for this graph. The diagnostic performance of the score is illustrated by calculating its respective sensitivity and specificity, using inclusive cutpoints between 11 and 7 points of the atopy score (table 3), and by a receiver operating characteristic curve (fig. 2) depicting graphically the behaviour of these parameters over the range of possible cut-off values.Table 3Measures of the diagnostic performance for the atopy score (dichotomised at various cutpoints) to detect atopic flexural dermatitis: sensitivity, specificity and PVs of a ‘high’ atopy scoreFig. 1Distribution (% in class) of the point values of the atopy score classified as suggested in Diepgen and Fartasch [5] – not considering photophobia and laboratory results – in 176 persons with versus 2,176 persons without (previous) flexural dermatitis.Fig. 2Receiver operating characteristic curve for different cut-off values of the atopy score.DiscussionTo our knowledge this is the first study which undertook to validate the atopy score concept not in a clinical setting, but in a population selected solely for their occupation. Additionally, relevant data were collected not primarily for this purpose, but for prospectively evaluating the prognostic significance of the various atopic features. Thus, the evaluation of the diagnostic significance of these features, i.e. their association with atopic dermatitis, was not the focus of the study, which made the setting more similar to a clinical situation, where such diagnostic signs are applied, and not assessed, which may influence bias in this respect (see below).While for most atopic features a significant association with the diagnostic gold standard ‘(previous) flexural dermatitis’ is found in this analysis, which is in accordance with the studies quoted here, this association is by far not as strong as found in the case-control studies of Svensson et al. [2], Diepgen et al. [3, 4, 6]and Diepgen and Fartasch [5], respectively. Accordingly, the discriminating power of the atopy score [3]is much poorer than in the original studies (fig. 1). Possible reasons for this poor diagnostic performance could be:(1) In groups of patients treated for atopic eczema (current [2]or current or previous [5, 6]) in a hospital facility, the average disease severity may have been greater than in our young, unselected cohort, which included relatively few persons with only slight current dermatitis. Thus, in clinical cases there may have been more and probably also more clear-cut atopic features. Such a difference in the disease spectrum of study populations leading to discrepancies in the performance of diagnostic tests has been repeatedly described in other contexts [10]and the special term ‘spectrum bias’ has been coined for this phenomenon [11], which has already been highlighted in relation to differences between hospital and community validation of diagnostic criteria of atopic dermatitis [12].(2) Many of the (clinical) features used as building blocks of the score are intrinsically ‘soft’ [13], and probably even the use of operational definitions (as employed here) or, e.g., colour photographs as reference will achieve standardisation only to a certain, limited extent. This will obviously limit the diagnostic value of the respective single features and of the atopy score based on these features.(3) Facing this potentially poor reproducibility, a certain type of ‘investigator bias’ has to be faced, which may easily occur if the observer cannot be blinded to the gold standard criterion: during physical examination for the presence of clinical atopic features, the observer will inevitably note flexural dermatitis too. In view of the common dermatological notion of an association between atopic features and atopic dermatitis, an observer may in this case be more inclined to regard a finding as positive, which would otherwise have been diagnosed as negative or equivocal. Being non-blind towards the previous history of atopic eczema – as in our study – may introduce bias in a similar way. While the direction of this bias is probably the same in all studies, its extent may have been higher in the clinical studies aiming at the assessment of this very association than in our study. This assumption is supported by a comparison between odds ratios found in our study (tables 1, 2) and those reported in Diepgen and Fartasch [5]revealing a considerably greater difference for clinical than for anamnestic features.(4) The omission of three items contained in the original atopy score, namely photophobia, total serum IgE and SX-1, does in our opinion not contribute to impaired validity, because (i) the inclusion of an item with poor validity cannot be expected to contribute to increased overall validity and (ii) laboratory results can apparently be omitted without weakening the score properties [6]. Probably the threshold values as defined by Diepgen et al. (more than 10 points – not considering laboratory results: ‘atopic skin predisposition’, 8–9 points: ‘... unclear’, below 8 points: ‘... unlikely/not present’ [4]) cannot be transferred to the score values as calculated here as a consequence of these omissions. However, the discriminating power is poor also with lowered threshold point values.In conclusion, the atopy score is useful to summarise the multitude of atopic features. However, as its diagnostic performance is poor (i.e. low sensitivity and specificity with any chosen cutpoint) the relevance of certain point values found individually should not be overestimated. Furthermore, a certain proportion of patients with atopic dermatitis has no flexural involvement but dermatitis of other sites. These persons will ‘dilute’ the association between atopic features and atopic dermatitis defined solely as flexural dermatitis; however, for reasons of comparability, we used the same definition of the gold standard as in the preceding score studies.In addition, the usefulness of any diagnostic tool in practice does not only depend on its sensitivity and specificity, but also on the prevalence of the disease or condition in the population to which it is applied: the positive PV is dependent on the prevalence of the disease in the population studied and will thus be low if the disease is rare, while the negative PV will be impaired if a disease is very common (see also Popescu et al. [14]). For example, assuming a fairly low prevalence of 7.5% as found in this sample of the population [which may reflect (self)selection processes as a consequence of the high irritant profile found in this occupation] and a high sensitivity and specificity of 93.2 and 96.0%, respectively, as found in the original study (cutpoint 10 points [5]), the positive PV, i.e. the posttest probability of actually suffering from the disease, would be 65.4% in our population setting. In the setting of a dermatitis clinic with an assumed prevalence of atopic dermatitis of 75% among its patients, the positive PV would be 98.6%, under identical assumptions concerning sensitivity and specificity. With the much lower values for sensitivity and specificity found in the present validation study, the positive PV is even much lower in the population setting (table 3); for instance, a positive PV of 20.9% for a score of 11 and above means that for every truly diseased person 4 others with such score values are ‘false-positives’. – Note that such a subset of the population represents one of the target groups for the application of the atopy score.

Journal

DermatologyKarger

Published: Jan 1, 2001

Keywords: Atopic dermatitis; Epidemiology; Diagnosis; Atopy

There are no references for this article.