THE RELIABILITY AND PREDICTIVE VALUE OF DIFFERENT CONCEPTS OF ATOPY
INTRODUCTION
It is o f prim e interest to o c c u p a tio n a l physicians to b e a w a re o f a n y m eans of identifying groups of p e o p le a t special risk o f d e v e lo p in g on o c c u p a tio n a l disease. Most directly, this is because an o p p o rtu n ity for primary prevention of disease b e co m e s or m ay b e c o m e availa b le . A dditionally, in legal terms, special risk implies special c a re so th a t there is a d u ty to explore th e likely use of discriminants predictively to ovoid disorders. In scientific a n d th e ra p e u tic terms, a d d itio n a l b e n e fit m ay a c c ru e from investigation by th e insights w hich m ay be o ffe re d on th e nature o f th e disease in its relation to possible pred ictive factors.
Perhaps it is for these sorts o f reasons th a t a to p y has assumed a ce n tra l position os a discrim inant for o c c u p a tio n a l allergic disease. W hat is rem arkable is th a t this position has bee n re a c h e d in the a b se n ce o f any science or scientific literature base to support it. In contrast, th e relationship b e tw e e n a to p y and oncogenesis has b e e n w ell explored(<^^>.
G enerally, studies of o c c u p a tio n a l allergic disease h a ve co n fin e d themselves only to considering the ossociational c o n n e c tio n o f a to p y w ith th e condition under study(33.34,35,40.49,65-71 )_ These studies hove usually used a to p y os d e fin e d by
skin prick tests os th e discriminant of ch o ice . However, in clinical p ra ctice , it is still m uch m ore c o m m o n to use family or personal history os indicators o f a to p y . This harks b a c k to th e original m eaning o f th e term in tro d u ce d by C o c o a n d C ooke in 1923^^^^ w h e re a to p y was described as "a syndrom e o f c o m m o n allergic disease including eczem a, urticaria, hay fever a n d asthm a a n d w hich had a g e n e tic basis".
The p re c e d in g text indicates th a t there exist a t least three different co n c e p ts of a to p y d e fin e d by personal history, fam ily history a n d skin prick tests. It seem ed useful a n d im p o rta n t to discover w h e th e r these different c o n c e p ts w ere synonymous a n d , if not, how g re a t w ere th e differences b e tw e e n the populations thus identified. This study appears to be the first to consider such on analysis. Also, to assess the true w orth o f a to p y os a discriminant for o c c u p a tio n a l disease, in this case LAA, it is necessary to consider form ally the sensitivity, specificity a n d predictive value of the c o n c e p t. The work described b e lo w addresses these matters.
OBJECTIVES
1. To e xplore th e c o n c o rd a n c e b e tw e e n different c o n c e p ts o f a to p y nam ely, personal history, fam ily history a n d skin prick tests.
2. To consider the sensitivity, specificity a n d pred ictive va lu e o f a to p y as a discrim inant in LAA.
METHODS
Analyses in this c h a p te r derived from m aterial c o lle c te d during th e course o f the studies described in C h a p te r 2 a n d C h a p te r 4. The questionnaire used o b ta in e d e v id e n c e o f fam ily history (parents, grandparents, siblings) a n d personal history o f allergy. Family or personal history o f allergy w e re taken as positive if such hod bee n d ia g n o se d a t any tim e by a clinician. Such definition naturally lacks precision but b o th fam ily a n d personal history ore essentially subjective criteria a n d th e definitions w e re in te n d e d to re fle ct th e co m m o n a lity o f eve ryd a y clinical p ra c tic e . Tighter definitions w ou ld ho ve o ffe re d som ew hat spurious precision a n d w o u ld ho ve underestim ated true rates.
Standard skin prick tests to gross mixture, house dust a n d Aspergillus (all Bencard) w e re used, a to p y being d e fin e d os th e presence o f a 3mm w h e a l (or greater) to any o n e o f th e three tests a fte r 10 minutes. Skin tests for c o t, d o g a n d horse hair w e re also pe rfo rm e d but not in clu d e d os criteria for a to p y because o f the possibility o f cross-reactivity w ith LAA antigens.
Subjects w e re divid e d into three a c c o rd in g to their symptoms. The first group consisted o f those w h o hod LAA asthma, the second com prised p e o p le with LAA rhinitis (w ithout asthma) a n d th e third grou p com prised th e rest o f the study p o p u la tio n , having no e v id e n c e o f labo ra to ry anim al allergy.
RESULTS
Table 5.1 shows th e num erical d a ta w hich w e re used to derive th e sensitivities, specificities a n d pred ictive values shown in Table 5.2. A to p y d e fin e d by skin prick tests w ith c o m m o n allergens was a sensitive (80%) a n d quite specific (82%) test for LAA asthm a. A topy d e fin e d by personal or fam ily history was less w ell discriminating. For LAA rhinitis, all three definitions o f a to p y w e re insensitive
(24-39%) a n d rather non-specific (65-77%). Predictive va lu e o f a to p y for disease was low e ve n for skin prick criteria (LAA asthm a 34%, LAA rhinitis 23%).
LAA asthm a
LAA rhinitis No LAA Total A to p y d e fin e d A topy + 8* 9 18 by personal history A to p y - 8 24 79 A to p y d e fin e d A to p y + 7 13 32 by fam ily history A to p y - 9 20 65 A to p y d e fin e d A to p y + 12* 8 15 by skin prick tests w ith c o m m o n allergens A to p y - 3 25 82 4 - 1 * * 146 * Results re ach in g statistical significance a t p £ 0.05
** O ne subject refused skin prick tests.
TABLE 5.1
Concepts of atopy, basic numerical d ata
The c o n c o rd a n c e b e tw e e n different criteria of a to p y are shown in Table 5.3. Of those p e o p le classified as a to p ic by personal history, 65% w o u ld ha ve bee n so classified by fam ily history. Similarly for a to p y classified by personal history os against skin prick test a n d a to p y classified by family history versus skin prick test, th e p e rc e n ta g e s w e re 72% a n d 64%.
C om parison o f those responding positively to the different a to p y criteria in 1978/9 a n d 1985/6 is o f interest a n d th e d a ta are shown in Table 5.4. The correlation b e tw e e n skin prick results during the tw o different study periods is g o o d (95%). Correlations b e tw e e n a positive personal a n d or a positive fam ily history o f a to p y in 1978/9 a n d 1985/6 are less sure (both 60%). These figures w e re c a lc u la te d a fte r allow ing for a n d excluding those coses w here the
personal history o f allergic disease was th a t a ttrib u te d to LAA a n d w hich had or m ight h o ve d e v e lo p e d during the period b e tw e e n th e tw o studies.
a to p y predictive value predictive value predictive value
Personal tiistory 50 79 23 27 77 26 71
Family history 44 65 13 39 65 25 69
Skin prick tests w ith co m m o n allergens
80 82 34 24 76 23 75
Table 5.2
Concepts of atopy sensitivity and positive predictive value (expressed as percentages) of atopy for LAA asttima and LAA rtiinitis, without asthma. Negative predictive value is for absence of atopy as indicator for ab sen ce of LAA.
Population C o n c o rd a n c e (%)
Personal history vs fam ily history 65
Personal history vs skin prick test 72
Skin prick test vs fam ily history 64
Table 5.3
C o ncordance betw een difference criteria of atopy (original study)
Population C hanges (%)
Skin prick 5
Family history 40
Personal history 40
Table 5.4
Percentage ctiang e in different criteria of atopy betw een 1978/9 and 1985/6 in sam e study population
Am ong o th e r considerations of note was the possibility th a t a to p ic status m ight have a lte re d b e tw e e n the start of e m p lo ym e n t a n d th e start of th e first study period in 1978/9. This was tested by tracing p re -e m p lo ym e n t skin prick reactivity in m e dical records. In 32% of individuals such a record was found a n d these d a ta suggested th a t the a to p ic status o f individuals h a d not bee n c h a n g e d as a co n se q u e n ce of their work in exposure to a lle rg e n ic species (m ean p re -e m ploym e n t - study d o te period 2.6 years, ra n g e 1-6 years). There was no reason to believe th a t the 32% w ere a g rou p in any w a y unrepresentative o f the full survey population.
The d a ta core, os stated in the m ethods section, was derived from cross-sectional studies a n d thus represented a survivor p o p u la tio n . It was thus necessary to consider the possible preferential loss o f vuln e ra b le sub-groups from the study, especially atopies with LAA. However, exam ination o f the d a ta suggests this to be unlikely since the p e rc e n ta g e of atopies in th e study was similar overall to th a t in the g eneral p o p u la tio n a n d there was an excess o f atopies a m o n g those w ith asthma.
Six p e o p le w e re skin test positive to c o t fur, three to d o g a n d o n e to horse hair. O f these, only one, w h o responded to c a t, d o g or horse was not also responsive to gross or house dust mite. The exclusion o f c o t fur in th e skin prick definition of a to p y th e re fo re hod on insignificant e ffe c t on th e n u m ber d e fin e d by the criterion.
Curves o f different sensitivity a n d specificity m ay b e derived from th e d a ta a n d d raw n o u t in relation to p re va le n ce o f a co n d itio n thus dem onstrating p red ictive va lu e graphically. This is d o n e in Fig. 5.1. The m id d le curve is th a t for th e best sensitivity a n d specificity a ch ie v e d in th e study (80 a n d 82%) a n d the low er curve is for the worst (39 a n d 65%). For illustrative purposes th e 95% sensitivity a n d specificity curve is also displayed. Intercepts at th e p re va le n ce rates in the cross-sectional study (C hapter 2) are shown.
100 P o s itiv e p r e d i c t i v e v a l u e (% ) 40 0 10 20 30 40 50 60 70 80 90 100 Fig 5.1 - Prevalence %
Predictive value o f a to p y a t worst and best levels of sensitivity a n d specificity fo u n d in this study
b o tto m = sensitivity 39%: specificity 65%; m id d le = sensitivity 80%; specificity 82%; to p = sensitivity 95%; specificity 95%; lost drow n for illustrative purposes only;
Summary
The three d o m in a n t c o n c e p ts o f a to p y, those d e fin e d by personal history, fam ily history a n d skin prick tests, are im precise discriminants for th e pred ictio n o f LAA disease. O f th e three, skin-prick testing was th e most sensitive a n d specific. A m arked lack o f c o n c o rd a n c e was shown b e tw e e n populations identified using d iffe re n t a to p y criteria. For tw o ou t o f three o f these criteria, there was also n o ta b le tim e va ria n c e in positivity w hich further underm ined p red ictive value in th e p ra c tic a l a n d collo q u ia l senses.