Module 2 - Case - HIRING PRACTICES AND LEGAL COMPLIANCE

ORIGINAL ARTICLE

Bas Sorgdrager Æ Carel T. J. Hulshof Frank J. H. van Dijk

Evaluation of the effectiveness of pre-employment screening

Received: 16 July 2003 / Accepted: 14 November 2003 / Published online: 28 February 2004 � Springer-Verlag 2004

Abstract Aim: As pre-employment screening and selec- tion is a main function in the protection of susceptible applicants from developing an occupational disease, we need insight into the effectiveness of this intervention under different exposure conditions. The aim of our study was to demonstrate the feasibility and usefulness of three indicators to evaluate the effectiveness of pre- employment examinations. Methods: We used a pre- employment medical evaluation table to gather the data needed for the indicators for effectiveness. The first indicator chosen is the predictive value of a positive test result (PPV) corresponding to the percentage of appli- cants who will develop an occupational disease after a positive test result. The second indicator is the number of pre-employment medical examinations needed to re- duce the number of new cases of an occupational disease by one (number needed to test, NNT). The third is the number of rejections for the job, as the consequence of a positive test result, needed to reduce the number of new cases of an occupational disease by one (number needed to reject, NNR). To illustrate feasibility and usefulness, we used the example of potroom asthma in the primary- aluminium industry. We used data on personal risk factors and on the incidence of potroom asthma from a nested case–control study in the Netherlands. Results: The three indicators for effectiveness could be applied. For high incidence rates, defined as 0.04 (40 cases/1,000 employees per year), the PPV values for personal risk indicators varied from 5% to 27%. The NNT varied from 116 to 667. Finally, the NNR varied from 4 to 20. For low incidence rates, defined as 0.005 (5 cases/1,000 employees per year), the PPV values were low (0.6% to 5%). The NNT were high (1,111 to 5,000). The NNR varied from 23 to 155. Conclusions: The three indicators for effectiveness are applicable under the condition of the availability of relevant empirical data.

The indicators provided useful information for the evaluation of the effectiveness of specific tests, which might be added as selection criteria. The personal risk factors studied were far from effective as selection instruments, especially in situations where a low inci- dence of potroom asthma exists. Personal risk factors at the pre-employment stage should not be added to the standard procedure to select susceptible applicants. Under conditions, they may be taken into account in a workers’ health surveillance programme. As a contri- bution to evidence-based occupational medicine, we recommend the use of the pre-employment medical evaluation table and the three chosen indicators for effectiveness as a standard tool to evaluate the effec- tiveness of pre-employment medical examinations.

Keywords Pre-employment medical examination Æ Evidence-based medicine Æ Evaluation Æ Occupational health services Æ Potroom asthma

Introduction

After a period of serious discussions on the various limitations of pre-employment examinations, the Medi- cal Examinations Act came into force in the Netherlands in 1998. According to this Act, pre-employment exam- inations with the purpose of selection of workers are forbidden unless the job poses very special requirements for the medical suitability of the candidate. The Act does not indicate in detail for which job a pre-employment examination may still be allowed, but a later published decree presents some strict policy regulations in this (Pre-employment Medical Examinations Decree 2001). One of these regulations states that if an employer considers that a pre-employment examination is neces- sary for candidates for a certain post, the employer must consult the Occupational Health Service about the legitimacy and content of the examination. Evaluation of the effects of this Act during the first 3 years showed

B. Sorgdrager (&) Æ C. T. J. Hulshof Æ F. J. H. van Dijk Netherlands Centre of Occupational Diseases, Coronel Institute AmCOGG, Academic Medical Centre, Amsterdam, The Netherlands

Int Arch Occup Environ Health (2004) 77: 271–276 DOI 10.1007/s00420-003-0492-z

that the number of pre-employment examinations per- formed in the Netherlands had decreased drastically, and that the legal status of the candidate for a job, in general, was improved. Finally, a general policy guide- line for occupational health services was published (Hulshof 2000). However, complaints about unjustified examinations still existed, and the need for more detailed guidelines for occupational physicians on advice about legitimacy and content of pre-employment examinations was expressed. An important issue in the decision- making process in this is knowledge about the validity of pre-employment examinations and their preventive effectiveness.

According to the International Labour Office (ILO), the main goals of pre-employment medical examinations are as follows (ILO 1998). First, to prevent occupational diseases in workplaces where preventive measures have reached maximum effectiveness but a health risk for susceptible applicants remains. Secondly, to avoid the health condition of the applicant from becoming a risk for others in the working environment or the commu- nity. A secondary advantage of pre-employment exam- ination is the availability of baseline health data for future health surveillance programmes, as well as the opportunity to inform applicants about risks and con- trol measures in their workplace. However, baseline data and information can be well provided separately from the selection procedure. The first step in the prevention of all occupational diseases should be exposure reduc- tion, which can work quite effectively (Paggiaro et al. 1994). Monitoring of personal exposure levels as a part of occupational health care is an important risk man- agement tool. Pre-employment screening may be added to identify major personal risk factors in applicants, in a strategy to prevent occupational diseases.

In general, evidence of the preventive effectiveness of pre-employment medical examinations is missing, not- withstanding a widespread use of this selection instru- ment all over the world. In our view, a professional decision to consider the implementation of pre-employ- ment examinations to select susceptible applicants should be based on an evaluation that includes several factors (de Kort and van Dijk 1997; Sackett et al. 2000). Most fundamental is knowledge about the benefits of the test, i.e. mostly the prevention of occupational dis- eases, and the harm caused by the tests when they lead to undeserved exclusion. Consequently, we need to know costs and alternatives. Thus, we need to know the number of examinations needed to prevent one case of occupational disease and the advantages and disadvan- tages of alternative prevention strategies. Basic infor- mation includes knowledge about the incidence rate of the occupational disease in the industry studied, asso- ciated with the current level of exposure or demands, and the fate of the occupational disease, medically and socially. The prevalence of relevant personal risk factors in the applicant population should be well known. A low prevalence corresponds, unfortunately, with low pre- dictive values of a positive test result. The relative risk of

developing the occupational disease related to these risk factors should be clear, as a low relative risk can cause problems with effectiveness of selection. For the avoid- ance of problems with sensitivity and specificity, knowledge about the validity of the tests applied is obligatory.

To stimulate evaluation we can develop and apply appropriate instruments. In our opinion three indicators for effectiveness can form the core of a standard evalu- ation instrument: the predictive value of a positive test result, the necessary number of pre-employment medical examinations, and the number of rejections necessary to reduce the number of new cases of an occupational disease by one (de Kort and van Dijk 1997; Sackett et al. 2000). To determine these indicators we need empirical data or reliable estimations.

To illustrate the feasibility and usefulness of this evaluation instrument, we analysed the consequences of a more stringent selection of susceptible applicants in the aluminium industry, as a potential addition to the standard procedure, as is usual in most countries. This practice implies that subjects with current respiratory complaints and asthma and/or bronchial hyper-respon- siveness at the pre-employment stage are excluded. All over the world approximately 20,000 potroom workers in about 100 primary-aluminium smelters are potentially exposed to hazardous agents in the working atmosphere. Potroom asthma, when present, is a serious disease. Even after cessation of exposure, a significant propor- tion of patients with occupational asthma is limited in daily activities because of persistence of asthma symp- toms and bronchial hyper-responsiveness (Paggiaro et al. 1994). Among other researchers of potroom asth- ma, Soyseth et al. (1995) found that 50% of workers with potroom asthma do not recover completely. Pot- room exposure can affect employees, mediated by work methods, personal behaviour and personal risk factors. The occurrence of potroom asthma has been described as being due to exposure to fluorides in gaseous form or as dust; a dose–response relationship has been found between the level of exposure to fluorides and work- related asthma symptoms (Soyseth and Kongerud 1992). Seventy percent of the cases become manifest in the first year, 85% in the first 2 years of exposure (Sorgdrager et al. 1998). In industrialised countries, a minority of exposed workers develops asthma, 0.06 to 4% of those exposed per year (Abramson et al. 1989). In recent years the incidence rate of potroom asthma has decreased below 1%, presumably due to two factors: exposure reduction and screening of susceptible subjects at the pre-employment stage (Sorgdrager et al. 1998). In developing countries, the incidence is probably higher.

The aluminium industry has some questions con- cerning the prevention of potroom asthma. In spite of technological measures, significant exposure reduction and improved work organisation, potroom asthma still occurs (Sorgdrager et al. 1998). The implementation of an intensive health programme for workers in potrooms has been suggested (Fisher 1989, Sorgdrager et al. 2001).

272

Fisher, from Australia, has proposed more stringent pre- placement criteria for persons in contact with airborne potroom products (Fisher 1989). Consequently, in the aluminium industries of Australia and New Zealand, not only significant chronic respiratory illness, but also a difference greater than 10% (as asthma characteristic) between pre- and post-bronchodilator volume in the first second of a forced expiration (FEV1) and asymptomatic low lung function [FEV1 and/or forced vital capacity (FVC) <80% of predicted], may give rise to exclusion. However, evidence supporting the need for the in- troduction of a more stringent selection is lacking.

To evaluate the potential addition of new tests as a more stringent selection procedure, we studied the fol- lowing questions concerning low lung function and allergic predisposition as personal risk factors related to the incidence of potroom asthma:

1. What is the predictive value of a positive test result at the pre-employment stage?

2. What is the number of pre-employment medical examinations needed to reduce the number of new cases of potroom asthma by one?

3. What is the number of rejections for the job needed to reduce the number of new cases of potroom asthma by one?

Material and methods

Indicators for effectiveness

We used the diagnostic model in 2·2-table format, modified after Sackett et al. (2000), as the basis for a ‘pre-employment medical evaluation table’. In this calculation model we presume the practice to reject all applicants with a positive test result in order to prevent a case of occupational disease. A positive or negative test result, respectively, refers to the presumed presence or absence of a per- sonal risk factor. We have calculated indicators for effectiveness related to two time periods, one with a high incidence rate of potroom asthma and one with a lower incidence rate.

Indicators for effectiveness of a pre-employment test on a per- sonal risk factor are:

– The positive predictive value (PPV), i.e. the proportion of those people with a positive test result that develops the disease within a certain period: a/(a+b).

– The number of tests needed to prevent one case (number needed to test, NNT): (a+b+c+d)/a.

– The necessary number of rejections needed to prevent one case (number needed to reject, NNR): (a+b)/a

To apply the pre-employment medical evaluation table as an evaluation tool in concrete cases, we need to have access to data, often from cohort or case–control studies where relevant tests have

been applied at the pre-employment stage without selection as a consequence. We need to know the following data or reliable estimations of these data:

– The distribution of positive and negative tests, or the distribu- tion of the personal risk factors in the applicant population ((a+b)/a+b+c+d and (c+d)/a+b+c+d, respectively).

– The incidence rate of potroom asthma in the working popula- tion during an appropriate follow-up period ((a+c)/ a+b+c+d).

– The proportion of workers with a positive or negative test result, respectively, who developed potroom asthma during an appro- priate follow-up period (a/a+b or c/c+d).

Material

For the evaluation table, we have chosen to present values for an imaginary population of 10,000 exposed workers. The distribution of personal risk data and the proportion of test positive/negative workers developing potroom asthma have been derived from a nested case–control study (182 cases, 182 controls) in two Dutch aluminium producing plants (Sorgdrager et al. 1995). Cases were workers unable to work because of work-related respiratory dis- ease, who met the criteria for potroom asthma. The selected controls were matched for age, year of starting employment, and working conditions. In recent years, technological and organisa- tional measures significantly reduced the level of fluorides (F) and the risk of potroom asthma in most industrialised countries. In this case–control study, a significantly lower fluoride-in-urine level was found for the period 1976–1982 (mean level 3.57 mg F/l, SD 0.74) than for the period 1982–1990 (mean level 2.67 mg F/l ur- ine, SD 0.69) (Sorgdrager et al. 1998). Since there is a dose– response relationship between exposure to fluorides and occur- rence of asthma, we decided to study the potential effectiveness separately in two periods, one with a high incidence of potroom asthma and one with a low incidence (Soyseth and Kongerud 1992). The values for the incidence rates for potroom asthma have been derived from the literature in order for the most reliable data to be presented (Fisher 1989). For the table we used, as a high incidence rate of potroom asthma, the value 0.04 or 40 cases/ 1,000 employees per year; as a lower incidence rate we used the value 0.005 or 5 cases/1,000 employees per year. Risk factors such as low FEV1 and high blood eosinophil count were equally di- vided in these two situations. The follow-up period was 2 years, a period during which the majority of cases is expected to develop the disease (Sorgdrager et al. 1998).

During both periods, as in all industrialised countries, appli- cants with current respiratory complaints, including asthma, were considered unsuitable for employment in the potrooms, so they were rejected. In addition, applicants with bronchial hyper- responsiveness without symptoms have been rejected since 1982, a second argument to study the period after 1982 separately. In both periods the presence of low lung function without symptoms, or allergy without current symptoms, were not used as a selection criterion. All personal risk factors were assessed in the pre- employment medical examination in a standardised way. Atopic history and blood eosinophil count were used as a marker of allergy (Sorgdrager et al. 1995).

Lung function

In occupational health practice and in occupational epidemiology, the lung function parameter mostly used is the volume in the first second of a forced expiration (FEV1) (Quanjer et al. 1993). FEV1 is assessed by spirometry, which is reproducible and mostly expressed as percentage predicted, corrected for gender, age and height. A limit of 80% of predicted value (FEV1<80%) has been used in most studies as the criterion to distinguish low and normal lung function (Fisher 1989; Sorgdrager et al. 1995; Quanjer et al. 1993;

Disease occurred Disease absent

Test result positive a b a+b Test result negative c d c+d

a+c b+d a+b+c+d

273

Post et al. 1996). Therefore, in the case–control study, a test result of FEV1<80% has been used to refer to low lung function as a potential personal risk factor.

Atopic history

Atopy can be defined as one’s capacity to develop immediate hypersensitivity after exposure to common environmental aller- gens, as demonstrated by positive skin test results or elevated specific IgE levels (Mensinga et al.1990). In occupational health care, atopy is mostly assessed by questions about its history, as more intense methods are thought to be inappropriate in the con- text of pre-employment medical examination (Kremer et al. 1995). A screening questionnaire is regarded as an efficient tool for dif- ferential inclusion of subjects with or without atopy in epidemio- logical studies (Lakwijk et al. 1998). Moreover, allergic predisposition assessed by questions might be a good predictor for occupational allergy in the future (de Zotti and Bovenzi 2000). In the case–control study atopic history was defined as one or two positive answers out of two questions asked at the pre-employment examination concerning allergic respiratory diseases in childhood or in the family (Sorgdrager et al. 1995).

Blood eosinophil count

Blood eosinophil count is elevated in allergic and respiratory dis- eases but also in many potroom asthma patients. A limit of 275 cells per cubic millimetre has been used for clinical eosinophilia (Mensinga et al. 1990). A cut-off point of 220 cells in the eosinophil count for pre-employment has been found, which resulted in a significant odds ratio of 6.0 in cases and controls related to the risk for potroom asthma (Sorgdrager et al. 1995). This cut-off level has been used to distinguish applicants with and without a personal risk factor.

Results

In the nested case–control study (182 cases, 182 con- trols) prevalence data of the risk indicators in the applicant population were assessed. Next, the propor- tion of potroom asthma cases were analysed in the cat- egory with a positive and negative test result for the personal risk indicator for both time periods. All data were extrapolated to the evaluation tables for lung function, atopic history and high blood eosinophil cell count (Tables 1, 2 and 3, respectively).

Lung function

The prevalence of low lung function in occupational populations in general is below 5%. In the case–control population (Sorgdrager et al.1995) the prevalence of low lung function (<80% predicted) was 3.1% in the whole applicant population and was 3.8% in the cases. We have extrapolated these data to the imaginary popula- tion of 10,000 applicants in Table 1.

Atopic history

The prevalence of atopy in the general population varies between 15–30%, depending on the definition of atopy (Kremer et al. 1995). In the chosen population with high incidence, the prevalence of atopic history at the pre- employment examination was 3.6%, in cases 18%, in controls 3% (Sorgdrager et al. 1995). At lower incidence, the prevalence of atopic history again was 3.6%, and was present in 50% of the cases (Sorgdrager et al. 1998). We have extrapolated the data in Table 2.

Blood eosinophil count

In the case–control study, a high eosinophil count (more than 220 eosinophil cells per cubic millimetre blood) was present in 21.4% of the cases. In the whole workers’ population, 3.3% had a high blood eosinophil count in the pre-employment examination (Sorgdrager et al.

Table 1 Pre-employment evaluation table. Normal lung function and low lung function level (<80% predicted) at pre-employment examination in cases and controls, in periods of, respectively, high and low incidence of potroom asthma

Parameter Asthma occurred

Asthma absent

Total

In periods of high incidence of potroom asthma FEV1 % predicted <80% 15 285 310 Normal lung function 385 9,216 9,690 Total 400 9,600 10,000

In periods of low incidence of potroom asthma FEV1 % predicted <80% 2 308 310 Normal lung function 48 9,642 9,690 Total 50 9,950 10,000

Table 2 Pre-employment evaluation table; atopy history at pre- employment examination in cases and controls, in periods of, respectively, high and low incidence of potroom asthma

Parameter Asthma occurred Asthma absent Total In periods of high incidence of potroom asthma In periods of low incidence of potroom asthma

Atopic history 72 288 360 Non-atopic history 328 9,312 9,640 Total 400 9,600 10,000 Atopic history 25 335 360 Non-atopic history 25 9,615 9,640 Total 50 9,950 10,000

Table 3 Pre-employment evaluation table; high blood eosinophil count (>220 cells/mm

3 peripheral blood) and low count at pre-

employment examination in cases and controls, in periods of high respectively low incidence of potroom asthma

Parameter Asthma occurred Asthma absent Total

In periods of high incidence of potroom asthma High count 86 317 403 Low count 314 9,283 9,597 Total 400 9,600 10,000

In periods of low incidence of potroom asthma High count 11 392 403 Low count 39 9,558 9,597 Total 50 9,950 10,000

274

1995). These data were extrapolated in Table 3. The results in periods of high and low incidence of potroom asthma have been summarised in Table 4.

A combination of personal risk indicators may con- stitute a more effective selection instrument. For exam- ple, allergy as risk indicator can be analysed, combining high blood eosinophil count and atopic history. In the chosen study population (n=364), 26 cases (14.2%) were positive on both risk indicators. At the pre- employment stage, the prevalence of allergy measured in this way was 6.3%. In the period of high incidence, 11 rejections should have been needed to prevent one case. Based on these data, there is no advantage in the com- bination of these two personal risk indicators.

Discussion

Evaluation of effectiveness of interventions is an essen- tial but often neglected task of occupational health care (Hulshof et al. 1999). Using data from a case–control study we could provided three indicators for effective- ness, as was the aim of the study. A basic problem is the lack of published cohort and case–control studies, with an appropriate follow-up period and standardised baseline measurements of potential personal risk factors without selection as a consequence. We need these studies to be able to calculate the indicators for effec- tiveness.

For the aluminium industry taken as example, the positive predictive value of pre-employment medical examinations was low, especially in situations with low incidences of occupational respiratory disease. Numbers needed to reject, to prevent one case, were 14, 37, and 155 for personal risk factors such as atopic history, high blood eosinophil count and low lung function, respec- tively. At high incidence, the positive predictive value having a substantial value was for atopic history and blood eosinophil count, 20% and 21%, respectively. However, no fewer than 116 examinations are needed to prevent one case, and for each prevented case, four rejections are needed. Nevertheless, selection of appli- cants with an atopic history or high blood eosinophil

count under the condition of high exposure levels might be considered. On the other hand, exposure reduction to relatively low levels contributes to a lower incidence of occupational asthma, which supports the first step in prevention, namely exposure reduction (Sorgdrager et al. 1998). In that study, another factor related to lower incidence was the presence of asymptomatic bronchial hyper-responsiveness. Bronchial hyper-responsiveness has been introduced as an extra selection criterion, which might be overkill. However, Soyseth et al. (1994) have mentioned that subjects with a positive test result develop asthma symptoms after brief exposure to the potroom environment, which supports the suggestion that exclusion of applicants with bronchial hyper- responsiveness may be justified. In most countries applicants with respiratory symptoms are rejected. As the presence of symptoms shows a special vulnerability, we do not recommend changing this practice. Never- theless, it is worthwhile to search for evidence, produc- ing concrete values for benefits and harm.

The indicators do not support selection of applicants with low lung function or with high eosinophil count without complaints, especially at low exposure levels. These findings are in accordance with other data (de Kort and van Dijk 1997). On the other hand, when selection did not take place, the attributable risk of low lung function, for example, can be estimated as 20% when, in accordance with Table 1, relative risk is approximately 1.25. In periods of high turnover of per- sonnel, selection on personal risk factors appears to be a good preventive measure for employers. However, when the criteria proposed by Fisher (1989) are implemented, especially the proposed selection of asymptomatic applicants with a low lung function, this should imply that 19 out of 20 applicants will be selected out unjus- tifiably in periods of high exposure. In periods of low exposure, 154 in 155 rejected candidates will be rejected unjustifiably.

Instead of using real data from the nested case–con- trol study, we have extrapolated the data to make an imaginary population of 10,000 potroom workers. This simplifies the calculations. The prevalence of personal risk factors found in the applicant population study was sometimes low in comparison with general population data. One explanation might be healthy workers’ effect due to self-selection by workers. Some problems arose in the finding or estimation of empirical data. The limited follow-up period of 2 years in the case–control study may have led to a (limited) underestimation of the incidence of potroom asthma and, thus, to a lower indicator for effectiveness. The example of the alumin- ium industry demonstrates the need for practice evalu- ation looking for evidence and the need for cohort and case–control studies to establish a more evidence-based occupational health practice. Critical evaluation of the established practice is necessary, with special attention to the context of low exposure or low demand levels. The presented pre-employment medical evaluation table and the three indicators for effectiveness can be used as

Table 4 The preventive effectiveness of low lung function, atopic history, and high blood eosinophil count evaluated in terms of positive predictive value (PPV), number of tests needed (NNT) and number of rejections needed (NNR) to prevent one case, in situa- tions of high (0.04) and low (0.005) incidence rate of potroom asthma

Parameter PPV (%) NNT NNR

Incidence rate 0.04 Low lung function 5 667 20 Atopic history 20 138 5 High eosinophil count 21 116 5

Incidence rate 0.005 Low lung function 0.6 5,000 155 Atopic history 7 400 14 High eosinophil count 2.7 909 37

275

an evaluation instrument, under the condition of the presence of data, mostly from cohort or case–control studies. When data are lacking one might consider estimating some values or a range of values in order to produce best estimated guesses. This solution has to be preferable to a personal clinical judgement, which often implies no transparency or evidence at all. Especially because pre-employment examinations can also have negative effects such as undeserved rejections causing harm instead of benefit, occupational health care has to assign a high priority to the evaluation of this practice. Cohort and case–control studies have to be set up, offering more valid data to evaluate this policy.

With respect to limitations of our data, we may conclude that there is sufficient support, so far, for the view that applicants currently suffering from asthmatic and other respiratory symptoms, including bronchial hyper-responsiveness, have to be excluded (Sorgdrager et al. 1998). A more stringent policy, including selection by other personal risk indicators such as an asymp- tomatic, low FEV1 level in a pre-employment examina- tion, atopic history and high blood eosinophil counts at pre-employment, cannot be supported by the indicators for effectiveness. One exemption might be the selection of applicants with an atopic history or high blood eosinophil count in periods of high potroom incidence. However, the conclusion, unfitness for the job without consequences for those other than the applicants them- selves, might be discussed. We recommend that all three risk factors be taken into account in workers’ health surveillance programmes during engagement, including medical examinations, information and education.

References

Abramson MJ, Wlodarczyk JH, Saunders NA, Hensley MJ (1989) Does aluminium smelting cause lung disease? State of the art. Am Rev Respir Dis 139:1042–1057

de Kort WLAM, van Dijk FJH (1997) Preventive effectiveness of pre-employment medical assessments. Occup Environ Med 54:1–6

de Zotti R, Bovenzi M (2000) Prospective study of work related respiratory symptoms in trainee bakers. Occup Environ Med 57:58–61

Fisher J (1989) Screening, surveillance and training. In: Winder C (ed) Occupational asthma in the aluminium industry. J Occup Health Saf Aust N Z 5:421–424

Hulshof CTJ (2000) Een algemene richtlijn voor aanstellingskeur- ingen. TBV 8:46–51

International Labour Office (1998) Technical and ethical guidelines for workers’ health surveillance. Occupational safety and health series 72. ILO, Geneva

Hulshof CTJ, Verbeek JH, van Dijk FJH, van der Weide WE, Braam IT (1999) Evaluation research in occupational health services: general principles and a systematic review of empirical studies. Occup Environ Med 56:361–377

Kremer A, Pal TM, Schouten JP, Rijcken B (1995) Airway hy- perresponsivenss in workers exposed to low levels of irritants. Eur Respir J 8:53–61

Lakwijk N, van Strien RT, Doekes G, Brunekreef B, Gerritsen J (1998) Validation of a screening questionnaire for atopy with serum IgE tests in a population of pregnant Dutch women. Clin Exp Allergy 28:454–458

Mensinga TT, Schouten JP, Rijcken B, Weiss ST, Speizer FE, van der Lende R (1990) The relationship of eosinophilia and posi- tive skin test reactivity to respiratory symptom prevalence in a community-based population study. J Allergy Clin Immunol 86:99–107

Paggiaro PL, Vagaggini B, Bacci E, et al (1994) Prognosis of occupational asthma. Eur Respir J 7:761–767

Post WK, Steyerberg E, Burdorf A, Heederik D, Kromhout D (1996) Choosing optimal values of FEV1 and FEV1/FVC for surveillance for respiratory disorders in occupational popula- tions. J Occup Environ Med 38:673–680

Quanjer PH, Tammeling GJ, Cotes E, Pederson OF, Peslin R, Yernault J-C (1993) Standardization of lung function testing. Eur Respir J 6 [Suppl 16]:5–40

Sackett DL, Strauss SE Richardson WS, et al (2000) Evidence- based medicine. How to practise and teach EBM, 2nd edn. Churchill Livingstone, Edinburgh London

Sorgdrager B, Pal TM, de Looff AJA, Dubois AEJ, de Monchy JGR (1995) Occupational asthma in aluminium potroom workers related to pre-employment eosinophil count. Eur Re- spir J 8:1520–1524

Sorgdrager B, de Looff AJA, de Monchy JGR, Pal TM, Dubois AEJ, Rijcken B (1998) The occurrence of occupational asthma in aluminium potroom workers in relation to preventive mea- sures. Int Arch Occup Environ Health 71:53–59

Sorgdrager B, de Looff AJA, Pal TM, van Dijk FJH, de Monchy JGR (2001) Factors affecting FEV1 in workers with potroom asthma five years after removal from exposure. Int Arch Occup Environ Health 74:55–58

Soyseth V, Kongerud J (1992) Prevalence of respiratory disorders among aluminium potroom workers in relation to exposure to fluorides. Br J Ind Med 49:125–130

Soyseth V, Kongerud J, Kjuus H, Boe J (1994) Bronchial respon- siveness and decline in FEV1 in aluminium potroom workers. Eur Respir J 7:888–894

Soyseth V, Kongerud J, Aalen O, Botten G, Boe J (1995) Bronchial responsiveness decreases in relocated aluminium potroom workers compared with workers who continue their potroom exposure. Int Arch Occup Environ Health 67:53–57

276

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.