Clinical history

A number of epidemiological studies of asthma and allergy have relied on clinical interview and examination to diagnose disease. Clinicians, of course, would regard this as the gold standard for such diagnosis - supplemented by appropriate laboratory testing. However, disagreement between trained observers as to the presence of clinical signs [Loudon & Murphy 1984] and final diagnosis is common, and well documented. Aberg [1989] compared diagnosis of asthma and allergic rhinitis made by himself versus military physicians in 1187 consecutive subjects. The presence of asthma was underdiagnosed by 20% in the routine military examination compared to his own diagnoses.



The British Medical Research Council respiratory questionnaire (MRC 1960) was designed for the diagnosis of chronic bronchitis in epidemiological studies. As a result, it contains relatively few items for the diagnosis of asthma. These ask about wheeze (Q 14a Does your chest ever sound wheezing or whistling: Yes/No; Q 14b Do you get this on most days - or nights: Yes/No) and the diagnosis of asthma (Q17 Have you ever had...bronchial asthma: Yes/No; specify type and any important details: open ended response). As a result, most studies of asthma using the MRC questionnaire supplement these items with ones on severity, frequency and recency of episodes, treatment, age at onset. The MRC core items, however, have now been used in a great number of studies and allow appropriate comparisons between studies.


The American Thoracic Society and the Division of Lung Diseases on the National Heart, Lung, and Blood Institute developed a more extensive respiratory questionnaire based on the MRC questionnaire. The items on asthma and wheezing are only slightly more extensive (Q 12a Does your chest ever sound wheezy or whistling when: you have a cold Yes/No; Occasionally apart from colds Yes/No; Most days or nights Yes/No; Q18a Have you ever had Asthma: Yes - still have it; Yes - no longer have it; No).

Helsing et al [1979] compared the ATS-DLD-77 with the MRC respiratory questionnaire both administered by interviewer (telephone), and self-administered to separate covariate-matched (age, sex, smoking, and years of education) samples of white Maryland residents. The response rate in the telephone samples was 80%, but only 52% (MRC) and 62% (ATS-DLD) in the mailed group, despite a single repeat mailing, telephone followup and home visit to those uncontactable by telephone. The authors state that the latter is partly due to termination of the study cutting off followup of late- enrolled replacements (22% of the overall sample were replacements for subjects who refused to take part).

The important results were that telephone interview and self- administered questionnaire obtained the same prevalences of wheezing, chronic wheeze and asthma; the prevalence of ever wheeze and asthma was 51.9% and 4.8% respectively using the ATS-DLD-77, and 31.5% and 5.6% using the MRC.

The children's version of the ATS-DLD-77 was used as the basis for the WHO childhood respiratory symptoms questionnaire. One month repeatabilities for this are described by Brunekreef et al [1992]. For "ever" wheeze the repeatability (Cohen's ■) was 0.76; wheeze in the last twelve months was 0.78; and for doctor-diagnosed asthma, 0.76. Prevalences on both occasions tend to remain unchanged, suggesting error was at random.


The Tasmanian Asthma Foundation Questionnaire was developed for a survey of all seven year-olds in Tasmania in 1968 [Gibson et al 1969]. This questionnaire contains numerous questions measuring wheeze and cough - recency of last episode, frequency in last two years, lifetime number and average duration of episodes, association with dyspnoea, provocation by exercise or season; respiratory tract infection; other atopic diseases; passive exposure to tobacco smoke, animals; history of infant feeding. The key (switch) wheeze item is "[h]as he/she at any time in his/her life suffered from attacks of asthma or of wheezy breathing? Yes/No (NOTE: Please regard 'asthma' and 'wheezy breathing' as being much the same thing for this survey; we do not ask you to try to tell the difference.)". The key (switch) cough item is "[h]as he/she at any time in his/her life suffered from bronchitis or attacks of cough with sputum (phlegm) in the chest ('loose' or 'rattly' cough)? Yes/No (NOTE: Please regard 'bronchitis' and 'cough with sputum (phlegm)' as being much the same thing for this survey; we do not ask you to try to tell the difference.)". No information about medical diagnosis of asthma was sought. Modifications of the TAFQ have been used in several Australian asthma surveys, most recently a followup of the 1968 Tasmanian cohort and their families - so consistency of questions with this instrument allows comparisons with other Australian studies. Mitchell and Miles [1983] reported the three month repeatability of "ever" wheeze at kappa=0.86, and "ever" cough at 0.81.

Jenkins et al [1996] gave a modified TAFQ to 96 adults who had taken part in the original 1968 survey, who were then interviewed by a respiratory physician, and underwent hypertonic saline challenge. The crude overall agreement between questionnaire and physician was 92% (sensitivity 80%, specificity 97%). Only 39% of those diagnosed as asthmatic by the physician exhibited BHR to saline.


Burney et al [1987, 1989] describe the development and validation of a questionnaire to predict bronchial hyperresponsiveness. These authors argue that BHR, being an objective measurable entity, is a better criterion than the diagnosis of asthma.

The International Union Against Tuberculosis and Lung Disease (IUATLD) questionnaire was completed by 4307 out of 6900 eligible adults aged 18 to 64 years in three English towns. A subset of 1325 respondents (all subjects with a history of wheeze, and a 20% random sample of the remainder) were invited to undergo a histamine inhalation challenge (Yan protocol), of whom 833 actually underwent testing. Using the results from half this group, the questionnaire items that best predicted BHR were determined using logistic regression and discriminant analysis. The performance of these chosen items was then evaluated on the other half-sample. BHR (PD20<7.8 umol) was present in 28% of the training sample and 14% of the evaluation sample.

The items chosen covered: (1) wheeze in the previous twelve months; (2) nocturnal dyspnoea in the previous twelve months; (3) "regular trouble with breathing that gets completely better"; and (4) chest tightness in contact with dust, animals or feathers. The discriminant function based on these four items had a sensitivity of 53% and specificity of 90%. This was only a slight improvement on the single item on wheeze in the last twelve months which had a sensitivity of 47% and specificity of 92%. In addition, the three month repeatability of recent wheeze was higher - Cohen's kappa=0.86 - than that of the discriminant function - kappa=0.58.

Despite the rationale of using BHR as a proxy for asthma, the authors note that BHR was strongly associated with smoking in this population, as asthma is not. The correlation between diagnosed asthma and BHR was not strong (sensitivity of asthma in the previous twelve months as a marker of BHR was 22%).

Abramson et al [1991] administered this questionnaire to 827 Australian subjects, validating against both a modified form of the MRC questionnaire and methacholine inhalation challenge. The two questionnaires agreed fairly well - for the critical item of wheeze, Cohen's kappa=0.72 (casewise concordance 62%). In predicting BHR, wheeze in the last twelve months had a sensitivity of 49% and specificity of 86%, not markedly different from Burney [1987]. Again, the sensitivity of "recent asthma" was 21.5%.


The European Community Respiratory Health Survey uses a screening questionnaire based on the IUATLD questionnaire (and others). The short version (used in the European centres) contained nine yes/no items [Grassi et al 2003]:

ItemTextScore for a yes (HOMALS)
1 Have you had wheezing or whistling in your chest at any time in the last 12 months?1.19
1.1 Have you been at all breathless when wheezing noise was present?2.01
1.2 Have you had this wheezing or whistling when you did not have a cold?1.39
2 Have you been woken up with a feeling of tightness in your chest at any time in the last 12 months?1.02
3 Have you been woken by an attack of shortness of breath at any time in the last months?0.83
4 Have you been woken by an attack of coughing at any time in the last 12 months?0
5 Have you had an attack of asthma in the last 12 months?1.61
6 Are you currently taking any medicine (including inhalers, aerosols, or tablets) for asthma?1.93
7 Do you have any nasal allergies including hayfever?0

Factor analysis and homogeneity analysis suggested a single factor might underlie the items. The sensitivity and specificity of a score of one or more were 75.2% and 80.1% versus asthma diagnosis reached by a panel of three physicians after clinical interview, spirometry and methacholine challenge, serum IgE and skin prick testing.

Reliability of proxy reporting of asthma

In many genetic epidemiological studies of asthma, reports of presence or absence of the condition in relatives by the individual completing the questionnaire are often the only source of family information. Litonjua et al [1998] report on the reliability of such proxy data.

Table 1. Proxy (wife) versus self-reported prevalence of allergic conditions from Litonjua et al [1998].
No 13110
No 12127


The most informative, and often only investigation used in the diagnosis of asthma is the assessment of lung function. The documentation of variable airway obstruction fulfils the ATS criteria for asthma reviewed in the first chapter, and may be spontaneous or provoked by bronchodilator or bronchoconstrictor agents.

Peak flow

The instantaneous maximal forced expiratory flow (PEF) is the largest flow rate recorded during a forced expiratory manoeuvre. This can be measured by spirometer, or peak flow meter. It is highly correlated with FEV1. The mini-peak flow meter has now become a standard tool for use by chronic asthmatics to record the degree of airway obstruction over time. Alterations in PEF from the longterm average for a given subject precede and follow symptomatic asthma exacerbations [Brown 1991]. In addition, indices measuring diurnal variation in PEF are correlated with degree of bronchial responsiveness [Ryan et al 1982; Brand et al 1991].

Conventional spirometry

I refer here to simple volume-time spirometry. The classical measure arising from the forced expiratory manoeuvre volume-time recording used to assess asthma is the forced expiratory volume in one second - FEV1. Acute FEV1 decrements reflect acute airway obstruction - mainly in the airways over 2 mm in diameter [Brown 1991]. In the case where previous values of FEV1 for an individual are not available, an expected value can be calculated using regression equations based on samples of normal subjects. These will usually include height and age, but can be extended to other anthropometric measures such as weight, sitting height and chest diameters. In this case, diminished values of FEV1 relative to the normal population may arise from short or long term airway obstruction, or from restrictive lung disease. The latter alternative is usually excluded by simultaneous assessment of the vital capacity (VC) or forced vital capacity (FVC) - which is diminished in restrictive disease.

A number of measures of small airway obstruction are in common use. FEF25-75 is the forced expiratory flow rate between volumes 25% and 75% of FVC. This measure is correlated with histopathologic small airway disease [Wright et al 1984]. More sophisticated indices include those arising from time- and frequency- domain analysis of the volume-time curve (where mixtures of exponential distributions are fitted to the overall curve), as well as those based on the flow-volume loop. Between asthma episodes, the FEV1 commonly overlaps the range seen in the normal population, but measures such as FEF25-75 are more likely to be reduced.

Spirometry reference values

A number of reference regression equations are in common use to provide norms for the assessment of FEV1 and FVC. In European and some Australian laboratories, those of the European Coal and Steel Community have become accepted as appropriate for Caucasian adults. US laboratories follow the ATS recommendation of selecting reference values that the group thinks is appropriate to their particular patient population. Individuals from other races tend to have smaller lung volumes for the same height and age - a factor adjusted for by the addition of cubital span in the regression equations, or multiplying final values by an adjustment factor, usually 0.9.

The ECCS equations [Quanjer et al 1993] for nonsmoking males are,

     FEV1 = 4.30*(Height) - 0.029*(Age) - 2.49; RSD=0.51
     FVC  = 5.76*(Height) - 0.026*(Age) - 4.34; RSD=0.61

and for nonsmoking women,

     FEV1 = 3.95*(Height) - 0.025*(Age) - 2.60; RSD=0.38
     FVC  = 4.43*(Height) - 0.026*(Age) - 2.89; RSD=0.43

where RSD is the standard deviation of the residual (Observed-Predicted).

Bronchodilator test

The test most commonly used in the diagnosis of asthma in clinical practice is to assess the acute response of the subject's lung function to an inhaled dose of bronchodilator. This demonstrates the reversibility of obstruction, thus fulfilling one of the ATS criteria for asthma. As I noted earlier, the reversibility in asthma may be slow, and assessment of improvement in lung function over several days of therapy is often used clinically.

Change in FEV1 is the usual outcome measure, and a number of indices for diagnosis of significant reversibility have been developed based on dFEV1 normed using predicted FEV1, and observed or predicted FVC. The most used measures are dFEV1/pre-bronchodilator FEV1 (dbase), and dFEV1/predicted FEV1 (dpred). Previously, a diagnosis of reversibility has been commonly based on a +15% dbase [American College of Chest Physicians 1974]. Since the upper 95% confidence limit for dbase in general population samples is 8-9%, the American Thoracic Society diagnostic threshold is +12% dbase [ATS 1991]. The ECCS criteria [Quanjer et al 1993] require a dpred=+12% that exceeds 200 ml. Miller et al [1987] presents a graph of dbase versus predicted FEV1 to categorise response allowing for the fact that large proportional changes in FEV1 on a small pre-bronchodilator volume may in fact represent trivial absolute volumes.

Bronchoprovocative testing

The assessment of bronchial responsiveness is performed by inhalation challenge using specific and nonspecific bronchoconstrictors. Although in use since the 1940s, its role in the diagnosis (and management) of asthma, as opposed to research, is still limited. This is because most asthma is still satisfactorily diagnosed using clinical criteria. In addition, the discovery of a proportion of the population with asymptomatic bronchial hyper-responsiveness has blurred the interpretation of this finding in individuals with questionable symptoms. Bronchial responsiveness to different agents may not be correlated in the same individual - though particular classes of challenge are intercorrelated.

Agents can be divided by prevalence of response and the related specificity for asthma [Cockcroft & Hopp 1987]. A higher prevalence of responsiveness is seen to the nonspecific agents histamine and methacholine. Lower prevalence and higher specificity for asthma is seen for exercise and cold air challenge, hypertonic and hypotonic fluids (hypertonic saline and ultrasonically nebulised distilled water). Within-group correlations tend to be higher than those between groups, but are not perfect, so that individuals may respond to histamine, but not methacholine for example.

Indices of bronchial responsiveness

Response to these agents is usually measured as deltaFEV1/pre-challenge FEV1 (dbase). Several alternatives in use are dFEF25-75, dSGaw (specific airway conductance), dFVC, dPEF and dTRR (total respiratory resistance). The dose-response relationship is usually nonlinear. In nonresponsive individuals, there may be no alteration or an increase in response to any dose given. Alternatively, increasing dose will decrease response to a small degree up to a point, after which further increase in dose sees no further fall - a plateau, as earlier discussed. In hyperresponsive individuals, the response function falls away more or less steeply from a inflection point.

Traditionally, the response is plotted versus log-transformed dose, where the response function appears sigmoid or parabolic in nature (Figure 4.1). Use of untransformed dose brings the relationship back to approximate linearity [eg Cockcroft & Berscheid 1983]. The summary indices of these curves are (1) dose required to cause a specified response; (2) slope of the descending portion of the curve ("hyperreactivity"); (3) location of the first inflection point ("hypersensitivity"); and (4) location of the plateau, if present. This last index is not measurable in many asthmatics, and tends to be correlated with the first three - Lougheed et al [1993] found it correlated 0.95 with PD10 FEV1 (dose provoking a 10% fall in FEV1 from baseline).

By far the commonest measure used is PD20 or PC20, the provocative dose or concentration of the agent required to cause a 20% fall in FEV1. A fall in FEV1 of this magnitude is regarded as evidence of significant bronchial hyperresponsiveness. Since a number of challenge protocols terminate when a delta-base of -20% is reached, this is equivalent to the averaged slope of the dose-response curve. PD20 is usually interpolated from the log dose-response plot. Extrapolation beyond the final cumulative dose to one additional doubling has been performed, although the appropriate curve to be fitted is problematical.

Chinn et al [1987] have championed fitting a power function to the dose-response curve for extrapolative purpose. The form of the function they describe is:

   log(change in FEV1) = a + b*log(dose)

which is equivalent to,

   change in FEV1 = exp(a) * (dose)^b

The former equation is fitted via ordinary least squares, implying exponential error on the untransformed scale of the latter equation. Townley and coworkers fit a parabola (setting b=2 in the latter equation), but summarise responsiveness as the integrated FEV1 over the dose given, rather than dFEV1.

Measures of dose-response slope have an advantage over PD20 in that while the majority of the general population do not attain a ■base of -20%, a slope is always calculable. O'Connor et al [1987] suggested the use of a simplified dose-response slope dbase divided by total dose of histamine given. Peat et al [1991] compared the simplified dose-response slope to PD20 as a discriminator between symptom groups in the Villawood sample of 1217 schoolchildren. In this sample only 15% had exhibited a 20% fall in FEV1 (with thus an interpolatable PD20), but only 43% exhibited no fall at all. Simplified dose-response slope (DRS) was distributed as a unimodal peaked symmetrical distribution on the log scale in children who had never experienced wheezing, but was more skewed in the sample who had experienced wheezing in the previous twelve months. A linear relationship was present between log[DRS] and frequency of symptoms.

In a set of 339 subjects from the Vlagtwedde/Vlaardingen study who underwent methacholine challenge, Rijcken et al [1989] compared several different indices - extrapolated PD10 and PD20, and the linear regression and simplified DRS. All indices were log normally distributed, and all discriminated equally well among symptom groups.

Chinn et al [1993] obtained similar data from a sample of 793 British adults undergoing histamine challenge. Only 200 had an interpolatable PD20. In 104 individuals retested, the intraclass correlation was 0.85 for log[PD20], using methods for censored data, 0.84 for the reciprocal simplified dose-response slope, and 0.89 for the least-squares estimated reciprocal dose-response slope. The correlation between log[PD20] and simplified dose-response slope was 0.95 for uncensored values of PD20, and all measures discriminated equally well between symptom groups. This group rejected log transformation of dose response slope as not producing a normal or homoscedastic distribution in this sample. They also noted that the repeatability of the simplified dose-response slope was worst for subjects with a PD20>7.8 umol of histamine (only 30% of their selected repeat sample), where this measure would be most useful. The repeatability of the least-squares estimate of dose-response slope was superior in this regard. As a result of these considerations, they felt that dose-response slope indices were not superior to PD20.

Protocols for bronchoprocative testing

A multitude of protocols are in current use for inhalation challenge. Broadly, pharmacologically active agents are prepared in buffered saline or glycerine solutions, and administered by nebulisers to the subject either intermittently - timed to inhalation, or continuously. The inhaled dose of agent is systematically increased, so a dose-response relationship can be established, and the dosing schedule can either be cumulative, where effects of the last dose have not remitted by the time of the next dose, or staggered. Several authors have attempted to use single dose challenges with nonspecific agents. Baltsch and Gonslor [1990] for example, found that a single (mean) dose of 0.5 mg histamine had a sensitivity and specificity of 85% for the diagnosis of hyperresponsiveness. I do not believe these to offer much in epidemiology.

Several authors have examined the optimum number of spirograms to be performed after each dose [Scott & Kong 1993; Prieto & Marin 1993]. Deep inspiration leads to bronchodilation, especially in nonasthmatics, so that response to a given dose improves with repeated measurements. Another recent paper [Malmberg et al 1993] noted that deep inspiration immediately prior to a dose of methacholine led to a smaller drop in FEV1 in response, an effect of the same magnitude as that arising from repeating the forced expiratory manoeuvre for a spirogram. Taking the best of two or three tracings therefore will increase specificity for asthma (which may or may not be desirable).

Cockcroft/Hargreave:: A recognised continuous inhalation protocol is that of Cockcroft et al [1977], which is an adaption of that of deVries et al [1962]. A Wright nebuliser driven by 7-8 l per minute of air nebulises the challenge solution which is inhaled via tidal breathing over two minutes such that 0.135 ml/minute of solution is delivered. The subject wears a loose fitting face mask. Spirometry is performed 1.5 to 3 minutes after that dose. The concentration of histamine solution is increased two fold on each occasion a positive test does not occur from 0.03 to a final dose of 16 mg/ml. An initial concentration of 1 to 2 mg/ml is used for subjects not expected to be responsive. A positive response is a delta-base FEV1 of -20%. The protocol ceases when a positive response is recorded, or when the final dose of agent is given. Responsiveness is expressed as PC20 (provocative concentration).

Chai:: Chai et al [1975] described a standardised protocol for administration of antigen, histamine or methacholine by Rosenthal-French dosimeter. A compressed air driven DeVilbiss nebuliser (originally a No. 42, but later a No. 646) nebulises the challenge solution, and is connected to the dosimeter, which inspiration triggers to open for 0.6 seconds. Five breaths are taken at each concentration (slow from FRC to IC). Spirometry is performed 1.5 minutes after the dose, and repeated at 3 minutes to document a sustained drop. The concentrations of methacholine given are 0.075, 0.15, 0.31, 0.62, 1.25, 2.5, 5.0, 10.0 and 25.0 mg/ml. Dose delivered is expressed as a breath unit - one inhalation of 1 mg/ml solution, so that the total dose that can be given is 225 breath units. Histamine is measured in the same fashion, though the concentrations used are 0.03, 0.06, 0.12, 0.25, 1.0, 2.5, 5.0, and 10.0 mg/ml, so that the total cumulative dose is 97.3 breath units. Approximately the same dose of agent is given via five inhalations via dosimeter as is given by two minutes of tidal breathing [Ryan et al 1981], though the droplet size produced by the Wright nebuliser is smaller than that of the DeVilbiss nebuliser so that deposition is slightly different. Responsiveness is expressed as PD20 in breath units.

Yan:: Yan and coworkers [1984] described a short protocol for use in large scale surveys. It has gained wide acceptance and has been used in several large Australian and European epidemiological studies. Histamine and methacholine have both been administered by this method, the results of which correlate well with those obtained by the protocols just described.

Histamine or methacholine saline solution is administered by hand operated DeVilbiss nebulisers (No. 45 or 47), using essentially the same open- mouth technique as that used for metered aerosol inhalation. Usually four doses (increasing fourfold) are given one to two minutes apart, thus effects are cumulative. Various numbers of inhalations of the same concentration solutions are used. FEV1 is assessed one minute after each dose. The total dose given is 7.8 umol histamine in the original protocol, though that now used by Ann Woolcock's group stops at 3.9 umol [Peat et al 1992].


Histamine has been used for bronchoprovocation since 1910, and acts to cause bronchoconstriction by direct stimulation of bronchial smooth muscle, and by vagal reflex. By inhalation it causes upper and lower respiratory tract symptoms, as well as systemic symptoms. These include watering eyes, dry irritated throat, cough, flushing and headache. Hoarseness and temporary (up to 24 hour) loss of voice can occur, and is more marked in the case of intercurrent or recent upper respiratory tract infection. Its bronchoconstrictive effects wear off within half an hour - the upper respiratory tract symptoms commonly seem to take longer.


This parasympathomimetic agent directly stimulates bronchial smooth muscle muscarinic receptors. Its period of action is similar to that of histamine, though it does not cause upper respiratory tract symptoms. It is not formally approved for administration to humans in Australia.


Bronchoconstriction can be induced by inhaled (or oral) propranolol in asthmatics via the blockade of presynaptic beta-receptors in the parasympathetic system. This agent is more specific to the diagnosis of asthma, but more likely to precipitate severe bronchoconstriction than methacholine.


If an individual with asthma or allergic rhinitis demonstrates a positive skin (allergic) reaction to testing with a particular allergen, they commonly will respond to inhalation of that allergen by bronchoconstriction. The dilution of allergen that produces a 5 mm wheal on skin testing is the initial dose usually used for these challenges. Spirometry is performed 10 minutes after inhalation by one of the methods described earlier. The late reaction (at 3-8 hours) is also assessed. This is another agent that can cause severe bronchoconstriction.

Exercise and conditioned air

Exercise is a well known "natural" precipitator of episodes of asthma in susceptible individuals. Usually six to eight minutes of vigorous exercise is required - most effectively, running. In respiratory laboratories, the exercise is usually performed on a treadmill or bicycle ergometer at an energy output provoking a heart rate acceleration to 80% of predicted maximum [Miller et al 1987], or 40-60% predicted MVV (maximum voluntary ventilation) [Sterk et al 1993]. In epidemiological field work in children, free running for five minutes has been used successfully [Burr et al 1990]. Isocapnoeic hyperventilation is an equally effective challenge.

Since exercise induced asthma is completely prevented by provision of fully humidified air at 37C [Strauss et al 1978], challenge with dry subfreezing air has been explored as a bronchoconstrictor. This method gives comparable results. Clinical exercise protocols have combined these stimuli [Sterk et al 1993]. Hypertonic saline and distilled water are other physical challenges that are similar in effect to exercise (and blocked by similar agents). Makker and Holgate [1993] for example found a (Spearman) correlation of 0.5 between hypertonic saline PD20 and an exercise induced asthma score, but only 0.05 with PD20 methacholine, and Amirav et al [1993] have shown prechallenge exercise to (in susceptibles) not increase PD20 histamine.

Tests of allergy

Skin reactivity

Exposure of the dermis to an allergen in sensitised individuals leads to the development of an allergic inflammatory reaction at that site. This occurs naturally via skin lacerations or punctures, often by the plant that produces that allergen. In clinical practice, the skin is pricked or scratched in a standardised manner to introduce a known dose of allergen solution. The resulting reaction is quantified by measuring the area of either the resulting erythema or wheal.

The procedure most commonly used is that described originally by Freeman [1930]. This involves skin puncture with a dedicated skin prick lancetter (though it was originally described using "any sharp needle"). Skin prick lancetters have been recently reviewed [Demoly et al 1991], and those giving the most repeatable results included the Osterballe device (the skin prick lancetter sold by Merck, Sharp and Dohme, and used by our group).

Drops of allergen extract are placed on the skin three centimetres or more apart, usually the volar forearm or back, and the lancetter is used to prick the skin under the drop. After 10 or 15 minutes, the diameters or area of the response is measured. The wheal has the advantage of being the most well delimited area. Turkeltaub and Gergen [1989] report only seven adverse responses to skin prick testing by this method out of 15794 subjects examined, of which six were syncopal episodes (and so likely to be vaso-vagal in nature), and one malaise. Probably only one fatality occurred in the United States due to skin prick (as opposed to intradermal) testing from 1973 to date [Lockey et al 1987; Reid et al 1993].

Serum IgE determination

Levels of sIgE are much lower in molar terms than other immunoglobulins, and so historically it was the last immunoglobulin to be identified.


Total serum IgE is estimated by a wide variety of techniques. Those most in use are the radio-immunosorbent assays, and immunoturbidimetric or nephelometric assays. Total sIgE is log normally distributed in the population [Klink et al 1990]. The reference range cited for all adults is usually 0-100 IU/ml, but levels decrease with age, so the 90% central interval for the Tucson epidemiologic study of obstructive lung disease sample of nonasthmatic skin-test-negative adults varied from 2-306 IU/ml for males aged 15-34 years, down to 1-104 IU/ml for females aged over 75 years [Klink et al 1990]. Use of the upper cutoff for this interval would detect 48% of skin-test-positive current asthmatics from the Tucson sample.


Allergen-specific circulating IgE levels are most commonly assayed by radio-immunosorbent assays as above. They correlate with skin test reactivity to that allergen, but are known to exhibit greater between-occasion variability, especially in or out of the (appropriate) pollen season. I do not believe they offer much additional information over the use of appropriately standardised allergen skin testing. For example, Brand et al [1993] found that SPTs were more sensitive and specific for the diagnosis of allergy in patients with lung diseases (Youden's index 0.45 versus 0.36).