Chapter  129:  Diagnosis and Management of Work-Related Asthma

What is Occupational/Work-Related Asthma?

Occupational asthma (OA) is a heterogeneous clinical syndrome characterized by work-related symptoms, airway inflammation (partially or completely reversible), bronchoconstriction, and hyper-responsiveness induced by workplace exposures (asthmagens). In contrast to non-occupational asthma, the differentiating feature of occupational asthma is the causal asthmagen is a substance in the occupational environment.¹ The symptoms of OA are similar to non-occupational asthma and include wheezing, cough, dyspnea, and impaired quality of work and non-work life. More than 250 asthmagens have been implicated and identified as causative agents in the development of OA (e.g., di-isocyanates, western red cedar, enzymes, snow crab, latex, flour, and laboratory animals). However, there are likely a variety of workplace asthmagens that have yet to be identified. An attempt has been made to classify the majority of the known asthmagens into categories based on their molecular weight: high molecular weight (HMW; ≥ 5000 Daltons) versus low molecular weight (LMW; <5000 Daltons) compounds, as this is probably important in their mechanism of action.² Due to the nature and complexity of workplace exposures, this is not always possible. Examples of HMW asthmagens include flour and latex; common LMW asthmagens are di-isocyanates, metals, and dyes.¹

OA can be broadly classified into two categories: OA with latency and OA without latency. Latency periods are observed in all instances of immunologically mediated asthma, even though the immunological mechanism may not yet have been clearly identified.² The latency period, which represents the time between the first exposure and onset of first symptoms, can range from weeks to years. When a worker has OA with latency, a specific inhalation test with the causal agent will usually be positive and often an immunological response for HMW and some LMW asthmagens is identified with specific skin prick tests (SPT) and/or serum specific IgE antibody testing. Generally, serum specific IgE antibodies, biomarkers for sensitization to the specific asthmagen, have been identified for HMW compounds; serum specific IgE antibodies associated with LMW OA have yet to be fully characterized, as the antibodies have yet to be discovered or occur in only a small portion of workers.² The need to produce appropriate hapten-protein conjugates is probably a common reason for the failure to detect antibodies to LMW agents. OA with latency for which serum specific IgE antibodies have been identified is referred to as IgE associated.

OA can also exist without a latency period and can occur after a single large exposure to irritant gases, fumes, or chemicals, such as nitrogen oxide, ammonia, and chlorine.³ This type of exposure results in reactive airways dysfunction syndrome, or RADS. Occupational asthma has also been reported to occur in few instances from repeated exposures to lower doses of ‘irritant’ or non-sensitizing agents.⁴ After exposure to an irritant, the asthmatic reaction occurs within a short time.⁵ Since there is no sensitization, the battery of diagnostic tests for RADS does not include specific immunological tests, and specific inhalation challenge (SIC) is not used. OA without latency is less common; it is believed to represent between 5 and 15 percent of all OA cases.^{1, 6, 7}

Table 1

Definitions of work-related asthma

Table 1

Definitions of work-related asthma

Term	Defining Features
Occupational asthma with latency of allergic or presumed immunological mechanism	There is an immunologic/hypersensitivity component and the diagnostic tests include measures of specific sensitization (e.g.,SIC, skin prick test, serum specific IgE).
Occupational asthma without latency	There is no allergic component and the worker is not “sensitized” to an agent, but rather, the agent causes an inflammatory response through an irritant mechanism.
Work-aggravated asthma (no latency period)	The worker has a previous or concurrent history of asthma that was not induced by an exposure found in the workplace. The worker is not sensitized to an agent at work, but is irritated by a “non-massive” exposure (e.g., cold, exercise, non-sensitising dust, fumes, or sprays) that provokes an asthmatic reaction.

Abbreviations: SIC = specific inhalation challenge

Through the remaining report, the term OA refers to OA with and without a latency period. This approach is similar to that of the American Thoracic Society guidelines.⁸ Where possible, we have noted where the results are derived from a RADS population. We recognize that some studies may also include individuals with work-aggravated asthma within their study population, although it is rarely possible to identify if this is so. When the individual studies provide adequate detail, we have included the proportion of workers with a previous history of asthma and this may indicate work-aggravated asthma rather than OA.

Epidemiology of Occupational/Work-Related Asthma

Amongst developed countries, OA has become one of the most prevalent occupational lung diseases.² The annual incidence of OA is estimated to be 50 per million United Kingdom (UK) workers and 140 per million Finnish workers.⁹ In France, the mean annual incidence rate was 24 per million workers.¹⁰ Among American workers belonging to a Health Maintenance Organization who were at risk for developing OA, the annual incidence of new-onset asthma was 1300 per million.¹¹ A review of 43 attributable risk estimates from 19 counties found the attributable risk of OA to be 9 percent (interquartile range [IQR] 5–19 percent).¹² Because of the high prevalence of OA, it has been recommended that OA be considered a diagnostic possibility in all adults undergoing initial asthma evaluation. Surveillance programs indicate that OA accounts for 26–52 percent of reported occupational lung conditions in the UK¹³ and British Columbia, Canada¹⁴. Perhaps of even greater concern is the suggestion that OA is potentially more common than studies have previously reported.¹⁵ This seems likely, as several factors make OA identification difficult: 1) most industries expose workers to a number of potential causative agents and exposure may vary widely within the same workplace setting, making exposure assessment complex; 2) OA symptoms are variable and non-specific, with late reactions often occurring after the working day has been completed; 3) specific diagnostic tests have limited availability making exact diagnosis difficult; and, 4) the symptom onset is unpredictable.¹⁵

Greater than 250 synthetic and naturally occurring agents in a variety of work settings have been identified as relevant workplace exposures¹⁶; one of the most common exposures being di-isocyanates used in the production of coatings, adhesives, and foams in a wide variety of settings. It is estimated that up to 11 percent of workers exposed to di-isocyanates will develop bronchial hyper-reactivity.¹⁷ Other commonly reported agents include flour, wood dusts, and latex.² While a large number of the causative agents have been identified, many more sensitizers and irritants are not well characterized and/or known, thus making the diagnosis of OA even more elusive.

The contribution of work-aggravated asthma cannot be ignored. Work-aggravated asthma occurs in workers with a previous or concurrent history of asthma and is characterized by worsening symptoms at work in response to chemicals or physical stimuli encountered at work, such as dust or cold air.² Among employed asthmatics in a Finnish city, Saarinen et al. found the prevalence of work-aggravated asthma to be approximately 30 percent.¹⁸ In a recent study, the annual incidence rate of work-aggravated asthma among employed workers with current asthma was 3.9 cases per 100,000 each year.¹⁹ The highest incidence rate occurred among workers employed in the public administration industry (14.2 cases per 100,000 each year). The most commonly responsible agents were mineral and inorganic dusts.

Because workers often need to be away from work due to their illness, and the cornerstone of treatment is removal from the workplace, OA creates a significant economic burden for individuals, industry, health care providers, and society. Using the human capital method, Leigh et al. calculated the cost of OA in the United States to be $1.6 billion annually.²⁰ Sometimes, impaired lung function does not improve or reverse, even after extended periods away from the causative agent²¹ and this can result in permanent unemployment. The United Kingdom's Surveillance of Work-related and Occupational Respiratory Disease program found that 30 percent of people reported to have OA between 1989 and 1992 were unemployed when contacted later.²²

Diagnosis of Occupational/Work-Related Asthma

An ideal diagnostic test has both a high sensitivity and specificity. Sensitivity refers to the test successfully identifying the proportion of the people who truly have the disease, i.e., the test detects true positives. A test has high sensitivity when it identifies nearly all of the people who truly have the disease. A test has high specificity when it successfully identifies most of the people who truly are disease-free as not having the disease, i.e., the test detects true negatives. In general, the sensitivities and specificities of diagnostic test comparisons are more robust if the test results are reproducible. OA is a difficult diagnosis to make for a variety of reasons. The first step is to determine that the patient has asthma and not a similar syndrome, such as upper respiratory tract irritation or vocal cord dysfunction. Having diagnosed asthma, the next step is to show a causative link with exposure to an asthmagen at work and this process often commences with obtaining a history or completing a questionnaire.

While SIC is often cited as a ‘gold standard’, as yet there is no definitive diagnostic test for OA. The applicability of SIC is currently limited by its availability.²³ In addition, the possibilities of false negative and positive results have been recognized.⁵ For example, a false negative can occur when a worker with OA is exposed to multiple asthmagens and is subsequently challenged with the non-offending asthmagen.² Further, SIC testing is usually not considered useful in determining a diagnosis of irritant induced OA. As a result, throughout this report we will refer to SIC as a reference standard.

There are several alternative techniques, used in isolation or in combination, which can be used to diagnose occupational and work-related asthma. We have summarized SIC and the alternative techniques in the categories below:

Specific inhalation challenge (SIC). There are several methodologies used to perform SIC; however, in general the approach is designed to expose the worker to a suspected asthmagen in a controlled, non-work, test environment. When a suspected asthmagen is a water-soluble compound, the test subject may inhale it in its nebulized form.²⁴ Initially, low concentrations are administered for safety reasons until a response is obtained or a maximum concentration has been tested. When the asthmagen cannot be nebulized because of low solubility, some have proposed tipping powders of the suspected asthmagen from one tray to another in an attempt to mimic the work environment and generate a dry aerosol.²⁵ More recently standardized techniques have been developed to deliver dry powder aerosols. Another method is to conduct a simulated occupational-type specific provocation test.²⁶ For example, when colophony (e.g., used as a solder flux by electronics workers) is suspected to cause OA, the worker is exposed to simulated work tasks in a controlled laboratory environment while lung function monitoring is conducted in an attempt to determine changes in flow volumes or hyper-responsiveness.

The first step of SIC is to perform a control challenge with a substance similar in physical characteristics to the suspected asthmagen, but not likely to cross-react immunologically. When the fluctuations in forced expiratory volume in one second (FEV₁) following the control challenge are less than 10 percent, it is generally considered safe for the worker to undergo formal SIC testing with the suspected agent.²⁷ The duration or intensity of exposure is increased progressively, usually commencing with a small dose. Lung function is measured for a period of time after each incremental challenge dose. When challenging with LMW agents, challenges of increasing time durations should occur on separate days, as LMW agents often produce isolated, late asthmatic reactions. When challenging with an HMW agent several doses may be given on the same day as they more typically produce an early asthmatic reaction.²⁷ Conducting SIC in a blinded fashion and ensuring that the results are reproducible can minimize false positive results. A typical definition of a positive response is a decrease from baseline FEV₁ (i.e., control exposure) of 15–20 percent.

The greatest advantage of SIC is that a positive test confirms a diagnosis of OA caused by the particular agent. When the appropriate agent is applied and the test is positive in workers with respiratory symptoms that are thought to be work-related asthma, SIC testing is considered confirmatory. However, there are several disadvantages to SIC. First, the suspected workplace agent may not have been identified, which precludes testing. Second, SIC can only be conducted in specialty facilities and these are rare. For example, a survey conducted in 2000 indicated that only 15 centers capable of performing SIC exist in the USA and Canada.²³ Moreover, there are many complexities associated with developing a SIC laboratory. These include the expense of purchasing and operating the laboratory and the ability to generate the appropriate concentrations of the putative asthmagens. Third, SIC can only be performed in relatively stable workers (FEV₁ >60–70 percent of predicted and/or >2L²⁸) because SIC may induce a very severe asthma attack.²⁹ A fourth disadvantage is the potential for false positives that can occur when SIC is performed in a worker with unstable asthma or when the worker suffers from marked non-specific bronchial hyper-reactiveness, as the non-specific irritant reaction can masquerade as an early reaction to SIC.²⁷ Finally, SIC testing can take several days. While SIC can be performed as an outpatient procedure, carefully monitoring is required for several hours after the challenge and oxygen, bronchodilators, steroids, and intubation equipment should be available.¹⁵

History and questionnaires. There are several questionnaires used to assess respiratory health.³⁰–³³ Questionnaires and histories may focus on specific job duties and work processes and inquire about improvement during weekends and/or holidays and worsening when returning to work.² The 1995 ACCP guidelines recommend three components of history taking: 1) history detailing onset of illness, temporal relationship between exposure and exacerbation, description of airway disease, and severity of asthma; 2) medical history describing pre- and co-morbidities and associated symptoms; and, 3) occupational and environmental history.¹⁵ In addition, material safety data sheets (MSDSs) can be collected for the chemicals used in the workplace. MSDSs describe the ingredients, concentration, and toxic properties of chemicals in the workplace.

Questionnaires and histories are easily administered. However, respiratory symptoms are a common complaint among all workers, thus questionnaires and histories tend to lack specificity. They may also lack sensitivity. The following circumstances can produce falsely negative results: exposure to the agent is indirect and/or sporadic; workers may be reluctant to disclose symptoms for fear of losing employment; or, the worker does not relate progression of OA to a workplace exposure.⁵ While the utility of history and questionnaires has been established for surveillance purposes and case finding in the event of symptoms of rhinitis and/or asthma, little research exists that examines their usefulness as a diagnostic test.

History and questionnaires are used to identify patients with relevant symptoms and workplace exposures known to cause OA. These patients will have a higher pre-test probability of testing positive to other tests of OA, as the subsequent test is being conducted in a pre-screened population. This is of importance in the later considerations in this review.

Serial lung function testing. Comparing serial lung function at work and away from work has previously been used to diagnose OA, especially when SIC test is unavailable. Until the advent of modern compact flow based spirometers, there were practical problems with the portability of the required equipment. Modern compact spirometers will allow for the measurement of lung function parameters including FEV₁ and forced vital capacity (FVC). One solution to this was to measure serial peak expiratory flow rate (PEFR) as portable and inexpensive equipment to measure this has been available for many years. The current ACCP guidelines recommend that PEFR be measured at least four times each day: upon waking, at noon, at the end of the working day, and at bedtime.¹⁵ During each measurement, three PEFR measures are performed and recorded once all three readings are within 20 L/min.¹⁵ The optimal duration of PEFR recording appears to be at least 4 weeks with a minimum of eight readings each day and at least 4 consecutive days in each work period.³⁴ However, obtaining records of that quality for that period of time is cumbersome. It is noteworthy that the sensitivity and specificity of this procedure was similar to records that were collected for at least four readings per day and at least 3 consecutive days at work for approximately 18 days.³⁴

Serial lung function testing is an inexpensive diagnostic test that may show a temporal association between lung function and occupational exposure.³⁵ Despite these advantages, serial lung function testing is still associated with a number of difficulties. Unfortunately, this test is dependent upon the worker's effort, requires them to reliably perform a forced expiratory maneuver and record the results accurately, and assumes worker honesty in performing and recording the results of the test.² Spirometers and peak flow meters incorporating a data recording device have helped with this to some extent. Another disadvantage of serial lung function testing is that it may be difficult to measure lung function for long enough periods while the worker is not at work to exclude OA. It is important the worker understands that late asthmatic reactions may occur several hours after exposure ceases, and they may not be detected during the monitoring period if only cross-shift measurements are made. In addition, serial lung function testing may be of no value among patients who are no longer exposed to the causative agent, as an asthmatic response should not occur, and when a patient has advanced OA, their lung function may become relatively fixed and not substantially improve during fairly short periods away from work. Finally, there are no universally accepted, standardized methods, for interpreting the results.¹⁵

Non-specific bronchial provocation (NSBP) testing. There are several different protocols used to measure bronchial hyper-responsiveness (BHR).¹⁵ In general, the first step of an NSBP test is to measure baseline lung function. Then, a worker inhales increasing doses of a bronchoconstrictor agent; the most commonly used agent in North America is methacholine.¹⁵ Numerous other agents have been used for NSBP testing, including histamine, acetylcholine, hypertonic saline, bethanechol, and distilled water. In the most frequently described test method (used in North America) the inhalation of methacholine or histamine begins at a concentration of 0.03 mg/mL for 2 minutes and is typically increased in a doubling fashion up to 32 mg/mL.¹⁵ Following each inhalation, the worker's FEV₁ is measured. The predicted concentration causing a 20 percent decrement (PC₂₀) in FEV₁ can then be estimated by linear interpolation.³⁶ There is no standardized definition of a positive response; however, a commonly used definition is a decrease from baseline FEV₁ of 20 percent or more at a methacholine concentration of 8 mg/mL or less.³⁷ For other techniques, a predicted dose causing a 20 percent decrement (PD₂₀) in FEV₁ can be similarly estimated.

When compared to SIC, the main advantages of NSBP testing approaches are that they are reasonably inexpensive, easier to perform, available in a larger number of facilities, and have proven safety records. In addition, the test can usually be completed in an hour, or considerably less if using an abbreviated protocol. However, a positive NSBP test only proves that the worker has hyper-reactive airways, which is typical of asthma due to any cause and it is not a definitive test for OA.²⁷ Serial NSBP testing can be performed at work and away from work. If the test demonstrates airways hyper-reactivity while an individual is at work and a significant reduction after the worker has been away from work for several weeks, then it is more likely the worker suffers from OA. From consensus recommendations, a threefold or greater increase in PC₂₀ or PD₂₀ when away from work as compared to at work is often considered a positive test result for OA³⁸, while a twofold change is suspicious, but not universally accepted as definitive of OA.⁵

Immunological testing. Classically, asthma is often thought of as a specific IgE mediated disease, most prevalent in atopic individuals. Common immunological tests include SPT and estimation of specific IgE/IgG, which is often measured by using the enzyme-linked immuno sorbent assay (ELISA) or radio allergo sorbent test (RAST) techniques. For SPT, a positive response is often defined as a 3mm or greater wheal reaction to a skin prick.³⁹ SPT can be used to identify responses to common allergens, which would signify atopy, or responses to specific antigens. Similarly, a high total IgE would be used to signify atopy, while increased specific IgE or IgG can also be measured. OA caused by the majority of HMW agents is specific IgE mediated and a positive SPT or raised specific IgE will be present. In contrast, LMW-induced OA is generally not specific IgE mediated, and consequently SPT or a measurement of specific IgE in this situation is not usually helpful⁴⁰, nor is it useful for irritant induced asthma or work-aggravated asthma.

Further, while specific immunological testing confirms that a worker is sensitized to a particular workplace allergen, sensitization is not synonymous with asthma, since sensitization to a workplace exposure can exist in the absence of OA. A further disadvantage of immunological testing is that few allergens known to cause OA are available in a standardized commercial form to be used for allergy testing. Importantly, there is no immunological test available for the majority of LMW compounds.

Total IgE and atopic status are used as screening tools but are not specific for OA or a particular workplace exposure, but rather identify atopy. Atopy is a risk factor for developing some types of sensitizer-induced OA.⁴¹

Measures of airway inflammation. Nitrous oxide is produced by a number of cells located in the respiratory tract, such as inflammatory and epithelial cells, and is detectable in exhaled air.⁴² It is hypothesized that asthmatic workers have higher levels of exhaled nitrous oxide (eNO) caused by airway inflammation than the normal population. Measuring eNO is a non-invasive procedure; however, it has yet to be fully validated as an effective diagnostic test for OA, as eNO seems to be increased in a number of inflammatory lung disorders.⁴³ Also, eNO testing is confounded by inhaled corticosteroid usage, as workers receiving inhaled steroid treatment tend not to have increased eNO.^{42, 44} Finally, it is a difficult test to perform for workers, is costly, and its availability in North American is limited. Perhaps, once techniques to measure and record eNO are widely available and fully validated, such a method may be more appealing.

A second measure of airway inflammation is sputum induction with identification of eosinophils and other cells or inflammatory markers in the expectorated material. The direct cellular and chemical evaluation of airway inflammation has traditionally been undertaken using bronchial biopsies and bronchial alveolar lavage. In both these latter techniques, individuals are usually sedated and bronchoscopes are used to sample the airways either through biopsy or saline lavages, respectively. Given the cost, inconvenience, and invasive nature of these tests, other methods of identifying and quantifying airway inflammation have been developed. One alternative is the use of sputum to examine the cellular and chemical agents responsible for inflammation. In many cases, asthmatic workers cannot produce sufficient sputum for examination, and so induction is performed using nebulized saline. Once a sample is expectorated, it is prepared and examined microscopically for cellular composition and chemically analyzed using a variety of techniques. The most important cells in the sputum include eosinophils and neutrophils. Standard normal concentrations of both have been reported⁴⁵ and these can be used to evaluate the samples.

This technique is safe and the repeatability of the technique has now been shown.⁴⁶ In the correct setting (qualified technologist, laboratory, and interpretation) this technique has been shown to provide useful information about the underlying inflammatory activities in the airway and may be incorporated into the diagnostic testing options available for OA.^{47, 48}

In OA, induced sputum markers have been used to determine changes in cellular make-up between pre- and post-exposure samples.⁴⁹ This technique shows some promise for agents or circumstances in which SIC cannot be performed. A significant change in eosinophil and/or neutrophil concentration post-exposure signifies an airway response related to the asthmagen.

Management of Occupational/Work-Related Asthma

Once OA is identified, the general recommendation has been to remove workers from the workplace rather than introduce medications to control their symptoms.¹⁵ There are, however, several different approaches used to manage OA. Firstly, workers can be treated pharmacologically in a manner that is similar to those with non-occupationally induced asthma. However, additional lung function deterioration may not be prevented in workers who receive pharmacological treatment and yet remain exposed to the causative agent.⁵⁰ Secondly, various mechanisms can alter the worker's environment to reduce exposures to an “acceptable” level by using personal protective equipment (PPE), making engineering changes to the workplace (e.g., improved ventilation, changing production materials or processes, etc.), or administrative changes to work tasks. Additionally, the worker can be relocated to a different job, or to a different area of the workplace, or it may be possible to substitute a non-hazardous agent for the causative agent in use. The final, and most drastic, management option is to remove the worker entirely from workplace. Removal from the workplace should ensure that exposure to the causative agent is ended completely which is considered by many to be the cornerstone of therapy; however, workers may not wish to be removed for financial and social reasons. In practice, it seems that many workers need to be removed from the workplace. Within 6 years, approximately one-third of workers are unemployed after their initial confirmed diagnosis of OA.^{22, 41} For those who are able to return to work, close medical follow-up is required to ensure that lung function does not continue to deteriorate at a rate quicker than would be anticipated as due to age alone.²

Reducing exposure. Reducing workplace exposures may benefit not only workers with OA, but also those that may go on to develop OA in the future⁴⁰; however, not all workers with OA benefit similarly from this approach. Reducing exposure levels has been hypothesized to be of particular use in irritant-based OA where there is no allergic component and hence a more linear and predictable dose response relationship.⁴⁰ Workers with OA with latency (i.e., immunologically mediated) may often react to very low concentrations of the aetiological agent, and therefore effective management by reducing exposure levels is questionable.² The American Thoracic Society analyzed five studies that examined the effects of reducing workplace exposure on occupational and work-related asthma.⁵ They found that OA improved or remained unchanged in approximately two-thirds of workers when exposure was reduced, while the other workers' asthma worsened.

One method to decrease exposure in the workplace is to reduce or eliminate the use of potential asthmagens from the workplace.⁸ Compared to engineering changes, this method is advantageous because it does not require mechanical maintenance.⁸ Examples of this method include using latex-free rubber gloves, latex gloves with lower protein content, or switching to a spray paint that is di-isocyanate-free. However, it is not frequently applicable in practice as the substances used in a workplace rarely have an alternative that can easily be substituted.

Engineering or structural changes can decrease workplace exposure to an asthmagen. Such changes include building an enclosure for the process where the asthmagen is used, or the establishment of local ventilation that clears the asthmagen away from the worker's breathing space.⁸ Administrative changes, such as job rotation, may also be useful in reducing exposure.

A third way to reduce exposure is to relocate the workers to a new job in an area of the building where the exposure is not present (or present in very low concentrations). The Americans with Disabilities Act recommends that large companies with adequate resources go to considerable effort to relocate workers with OA into an area or job with decreased exposure.⁵¹

Finally, exposure concentration can be reduced by the proper use of PPE. In order to be of value, PPE must be correctly worn, safely removed, properly maintained, and replaced as needed.⁴¹ By creating a better atmospheric breathing environment for the worker, air-supplying respirators provide the highest level of protection.⁵² The self-contained breathing respirator can protect against most exposures; however, it is cumbersome, expensive, and can not be comfortably worn for an entire work shift.⁵² There is general consensus that PPE is more appropriate for managing irritant-induced OA with brief and infrequent exposure than sensitizer-induced OA for the reasons mentioned above.⁸ PPE will reduce but not eliminate exposure and consequently is only recommended for short-term use.⁸

Removal from exposure. There have been numerous deaths reported among workers with OA who were not removed from exposure suggesting that complete asthmagen avoidance is important.⁵³ Previous authors have also suggested that for workers suffering from OA with latency (immunologically mediated OA), there is consistent evidence concluding that removal from the exposure results in an improved health outcome.^{2, 54} In addition, it appears that compared to late removal, earlier removal is associated with greater improvement in lung function and symptoms.^{55, 56} In contrast, workers with irritant-induced OA are more likely to be able to return to work and manage asthma pharmacologically.⁵

However, removal from the workplace has significant deleterious effects on the income and financial stability of the individual affected, as well imposing costs on the employer.^{57, 58} OA claims are also a financial burden to the government; the average accepted claim in Quebec, Canada in the early 1990's was approximately $50,000.⁵⁹

Pharmacological treatments. Pharmacological treatment of OA does not differ from chronic non-occupational asthma treatment. Asthma is managed through a variety of methods including trigger avoidance (reduction of trigger or complete avoidance), education, and pharmacological measures. Asthma is managed pharmacologically using two large groups of agents: relievers (bronchodilators) and preventers (anti-inflammatory agents).

Relievers. Relievers generally act on the beta-receptors in the airway to dilate the airways and relieve symptoms. Examples of these agents include short-acting beta-agonists (SABA) such as salbutamol (Ventolin®) or terbutaline, long acting beta-agonists (LABA) such as salmeterol (Serevent®) and formoterol (Oxese®, Foradil®), and anticholinergic agents such as ipratropium bromide (Atrovent®) and tiotropium (Spiriva®). These agents may be taken for relief of symptoms or prior to work to avoid the drop in FEV₁ that normally occurs in workers suffering from OA. There is some evidence that regular use of LABA agents, when administered with inhaled corticosteroids, can improve chronic asthma control.⁶⁰ Mono-treatment with LABA is not recommended. There is limited evidence that regular use of anticholinergic agents improve chronic asthma control.⁶¹

Preventers. Broadly, there are a number of preventers from which physicians and workers may choose to decide upon for therapy (corticosteroids, combination agents, leukotriene receptor antagonists or modifiers, and mast cell stabilizers):

• Corticosteroids (CS). Corticosteroids work by reducing inflammation, up-regulating beta-receptors, and decreasing airway edema by reducing capillary permeability. There are a variety of agents and delivery methods and doses of corticosteroids; however, the two main methods of delivery are systemic and inhaled. Systemic corticosteroids are effective in intravenous, intramuscular, and oral forms; oral is clearly the preferred method of delivery due to convenience, cost, and worker's adherence. However, long term systemic CS are associated with severe adverse effects, such as osteoporosis, skin changes, cataracts, impaired glucose regulation, and fluid retention (so called: moon-face and buffalo hump appearance). Prior to the introduction of inhaled corticosteroids (ICS), systemic CS agents were the mainstay for control of moderate-severe asthma. Since the introduction of ICS, these agents have largely replaced systemic CS as a treatment of choice.

• Combination agents. ICS and LABA agents may be taken separately as individual inhalational agents, or as new combination inhalers. The current recommendation states that LABAs should not be used without an ICS. Fluticasone combined with salmeterol (Serevent®) has been marketed as Advair® (Seritide® in Europe); budesonide in combination with formoterol (Oxese®) has been marketed as Symbicort®. These agents deliver both ICS + LABA in each inhalation and have the benefit of increased compliance and ease of use. In general, these agents are reserved for the treatment of moderate-severe work-related asthma unresponsive to increasing doses of ICS.

• Leukotriene receptor antagonists (LKTRA) or modifiers. Discovery of the cystenyl leukotriene pathway has been an important advancement in asthma care over the past decade. The pathway is particularly important in the inflammatory cascade involved in asthma, especially for children and in aspirin-sensitive workers. Agents that inhibit or block cystenyl leukotriene pathway, called LKTRA, have been marketed and are available for mild-moderate asthma in adults and children. In general, these agents are restricted to add-on treatment of moderate-severe work-related asthma unresponsive to increasing doses of ICS.

• Mast cell stabilizers. Infrequently, mast cell stabilizers are used to control asthma. These agents are often used in exercise-induced bronchoconstriction (EIB) and in children, due to their non-steroidal nature. Given the availability of more effective agents, these agents are infrequently used now and generally restricted to EIB and mild asthma in children.

• Other agents. A variety of other agents are available to treat asthma including methyl-xanthines, antibiotics (especially newer macrolides) and a non-traditional agents. Due to their general lack of effectiveness, these agents will not be described in detail in this report.

Literature Search

Two medical librarians identified appropriate databases to search and developed search strategies based on the following terms: asthma, lung disease, respiratory disease, occupational disease, occupational exposure, worker, work-related, leave work, reduce exposure, personal protective equipment, pharmacological treatment, inhalation challenge, peak flow, forced expiratory flow rates, bronchial provocation test, medical history taking, diagnostic techniques, sensitivity, specificity, predictive value, and, likelihood function. A filter was applied to the search output to remove studies pertaining to children (defined as those under the age of 18 years).

These search terms were adapted appropriately to search the following electronic resources: MEDLINE^®, EMBASE^®, Dissertation Abstracts^®, Expanded Academic^®, National Agricultural and Safety Database^®, CINHAL^®, Biological Abstracts^®, Agricola^®, and trials registries (http://clinicaltrials.gov/; http://www.centerwatch.com/; http://www.cochrane.org/index0.htm; http://www.controlled-trials.com/; http://www.update-software.com/National/; http://www.trialscentral.org/). Web of Science^® was searched by tracking the most sentinel articles forward. The Cochrane Airways Review Group has developed an “Asthma and Wheez* RCT” register through a comprehensive search of EMBASE^®, MEDLINE^®, and CINAHL^®. In addition, hand searching of 20 common respiratory care journals has been completed and relevant randomized controlled trials (RCT) have been added to the register. This database was examined for articles pertaining to therapy for OA. The detailed search strategies appear in Appendix A.^♦

Authors of included studies that had published at least two papers were contacted regarding any missing studies. The reference lists of all included articles and an internal report that was recently prepared for the Workplace Safety & Insurance Board of Ontario, Canada entitled “The Diagnosis of Occupational Asthma” were searched for any new studies we might have missed. Conference proceedings from ACCP, American Thoracic Society, and European Respiratory Society scientific meetings were hand searched for the years 2001-2003. In addition, clinical practice guidelines were examined for the following organizations: ACCP, American Thoracic Society, Canadian Asthma Consensus Guidelines, European Respiratory Society, Thoracic Society of Australia and New Zealand, British Thoracic Society, World Health Organization, International Labor Organization, International Commission on Occupational Health, and the Canadian Thoracic Society Guidelines on Occupational Asthma. The search was not limited by language or publication status and is considered current until February 2004. If potentially relevant studies were identified as part of the review process and met the inclusion criteria, they were included.

Selection and Inclusion

Table 3

Inclusion/exclusion criteria for review on diagnosing (more...)

Table 3

Inclusion/exclusion criteria for review on diagnosing and managing occupational asthma

Criterion	Diagnosis Review	Management Review
Study Design	Include: RCT, CCT, prospective or retrospective cohort, cross-sectional, case-series (>2 subjects)	Include: RCT, CCT, prospective or retrospective cohort, cross-sectional, case-series (>2 subjects)
	Exclude: case studies	Exclude: case studies
Participants	Include: de-novo asthma or a previous diagnosis of occupational asthma that is exacerbated at work	Include: de-novo asthma or a previous diagnosis of occupational asthma that is exacerbated at work
Intervention / Control	Reference Standard: SIC, supervised workplace challenge, serial lung function tests, serial measurement of non-specific airway reactivity, immunological testing, clinical expert diagnosis and exposure to an “asthmagen”.	Removal from the workplace, relocated to a position with decreased exposure to the “asthmagen” within the same workplace, PPE, engineering controls, or pharmacological treatment.
	Other Comparison: above tests and/or sputum, metabonomics, nitrous oxide.
Outcome Measures	Absolute numbers to construct a 2 × 2 (comparing two diagnostic techniques) or 2 × 1 (assessing one diagnostic technique in workers with a previous diagnoses of occupational asthma) table, sensitivity, specificity, or likelihood ratios, cost, time to complete diagnosis, adverse effects.	Pulmonary function, use of medication, healthcare utilization, frequency of exacerbations, QOL, symptoms, economic consequences, adverse events.

Abbreviation: RCT = randomized controlled trial

CCT = clinical controlled trial

SIC = specific inhalation challenge

PPE = personal protective equipment

QOL = quality of life

Using standard data forms, two reviewers independently applied the inclusion criteria. Disagreements were resolved through discussion and consultation with an occupational and asthma researcher. In situations where the studies were attempting to diagnose or manage workers with OA, respiratory symptoms, and/or rhinitis, studies were included if the results were stratified by specific disease and data for those with OA extractable, or if >90 percent of the included workers suffered from suspected OA, and data for just those subjects with OA could not be identified, all data were included. Case-reports and series involving less than two workers were excluded, as were studies that exclusively examine two different ways of measuring the same diagnostic test (e.g., Occupational Asthma System computer system versus visual assessment of PEFR records).

If needed, the investigators contacted the authors to clarify that individual publications reported on discrete workers. In cases where multiple publications involving the same or a portion of the same workers were identified, the most recent publication of the largest cohort was selected and any additional, unique information from previous publications was incorporated. If a study was published within the last 10 years and relevant health outcomes were not presented by ongoing exposure status, but the authors presented exposure status as an outcome (e.g., number remaining at same workplace), the authors were asked to provide outcome data by exposure status.

Quality Assessment

Assessment of the methodological quality of included studies was completed independently by two reviewers, using a variety of methods of assessment based on the review topic as follows:

Diagnosis review. The methodological quality of each included diagnostic study was completed using a quality tool developed from Lijmer's empirical research examining biases in diagnostic studies (Appendix B). The quality tool was comprised of 12 questions and pilot tested by the research team prior to employment. Empirically validated questions included study design (case control versus cohort), levels of blinding of measurements, use of appropriate reference or gold standard, an adequate description of the reference standard and test(s), thorough description of the study population, and the occurrence of differential reference bias.⁷⁰ In addition, information regarding the occurrence of partial verification bias, timing of data collection, method of worker selection, reporting of inter-rater reliability, and the method of reporting results was also captured. Discrepancies were resolved through discussion. Whether or not the authors mentioned that medication was terminated (or attempted) before testing was considered an additional marker of methodological quality.

Management review. The quality of each included cohort therapy study was independently assessed using Downs and Black's partially validated “Checklist of the assessment of methodological quality of both randomized and non-randomized studies of health care interventions” (Appendix B).⁷¹ This tool is composed of 28 questions that evaluate reporting (10 questions, total score 11), external validity (three questions, total score three), internal validity-bias (seven questions, total score seven), internal validity-confounding (six questions, total score six), and power (two questions, total score two). The maximum total score was 29 indicating high quality and the lowest possible score was 0 indicating low quality. The reviewers, asthma researchers, and occupational health specialist developed a priori guidelines regarding the application and implementation of the quality tool. In the event that a question was not applicable to the study design, the question was answered “no”. Also, the funding source was recorded for each study. Controlled clinical trials (CCT) were assessed with a second tool, the Jadad scale.⁷² This validated five-point scale assesses randomization, double-blinding, and the reporting of withdrawals and dropouts. In addition, concealment of allocation was evaluated to be “adequate”, “inadequate”, or “unclear”.⁷³ When there were multiple publications of the same workers, methodological quality was assessed on the most recent or largest study. Discrepancies were resolved through discussion.

Data Extraction

Two data extraction forms were developed, piloted, and used to extract data from the group of diagnostic or therapy studies (Appendix B). Data were extracted by one reviewer and checked for completeness and accuracy by a second reviewer. When data were presented graphically, the graphs were scanned into CorelDraw® (Version 9; Ottawa, Canada) to facilitate extracting the data points with greater accuracy. Discrepancies were resolved through discussion and input from the task leaders. Whenever possible, a task leader classified the exposures as high and low molecular weight based on published information.⁷⁴

Diagnosis review. Data were extracted regarding the study population characteristics, diagnostic tests, and results (e.g., sensitivity, specificity, likelihood ratios [LR], etc.). The following patient characteristics were recorded: probable cause of OA, patient source, sex, race, duration of exposure and symptoms, history of atopy/allergy diagnosis, smoking status, history of asthma, medication usage, and baseline pulmonary function. For each diagnostic test, the methodology, timing, inclusion/exclusion criteria, medication use, and definition of a positive test result were extracted.

The reference standard was assigned an ‘evidence grade’ as per our Levels of Evidence (Appendix C) designed in consultation with the technical expert panel (TEP). Additional outcomes included cost of diagnosis, time to complete diagnosis, and presence of adverse events. Where possible, a 2 × 2 diagnostic table was constructed for the reference standard versus the comparison test(s). Where a comparison involved two ‘reference’ tests the highest ranked test for evidence grade (see Appendix C) was used as the reference test. The sensitivity and specificity were extracted or calculated from the 2 × 2 table.

Management review. Details concerning worker characteristics, interventions, tests used to measure outcomes, and outcomes were recorded. Workers were described by their sex, age, race, smoking status, history of asthma, history of atopy, probable cause of OA, diagnostic tests used to determine OA, duration or exposure and symptoms, current medication, and severity of asthma. Where applicable, appropriate workers characteristics, such as smoking status, at follow-up were documented. Interventions were described by change in exposure (decreased or removal from exposure), type of protective equipment employed, or the type, dosage, route, and timing of pharmaceutical treatment. The type of test, method followed, and when appropriate, definition of a positive test (e.g., follow-up NSBP test) was recorded for each of the tests used to measure outcomes. Outcomes extracted included lung function, symptoms and medication scores, and economic status. The length of follow-up was also recorded.

Data Analysis

All data analyses were performed using SAS® (Version 8.2, Cary, USA). Graphics were produced using S-Plus® (Version 6.0, Seattle, USA).

Diagnosis review. Whenever possible we re-created a standard 2 × 2 table (or 2 × 1 if only reference standard positive or reference standard negative subjects were included) for each comparison test or combination of tests. This was not possible when individual patient data (IPD) was presented without a documented cut-off value indicating the presence (i.e., a positive result) or absence of OA. When sensitivity and specificity could be calculated for more than one cut-off value or definition for the same diagnostic test, we included the table that produced the highest test efficiency defined as the proportion of correctly identified patients (reference standard positive and reference standard negative) by the comparison test.⁷⁵

Sensitivity and specificity were calculated for each study using standard formulas. Results were pooled using the inverse variance method for random effects to calculate an overall estimate of sensitivity and specificity.¹ We pooled data from 2 × 1 tables separately (reference standard positive or reference standard negative only) from studies that presented data from which a 2 × 2 table could be generated (both reference standard positive and negative). Furthermore, results were only pooled for studies of similar molecular weight (HMW or LMW) asthmagens. When sensitivity and specificity could both be calculated, these values were plotted in receiver operator curve (ROC) space. When there were a sufficient number of studies (n>10), a summary ROC (SROC) curve was derived and added to the plot.

One drawback of pooling sensitivity and specificity is that it does not take between study heterogeneity into account. There are many sources of heterogeneity in any review, and a specific source for diagnostic reviews is that variation in sensitivity and specificity that results from heterogeneous definitions of a positive test result.^{76, 77} When the information was reported, we recorded the exact definition used to define a positive test and this information is presented in an evidence table (Appendix E). The inclusion of different asthmagens within each molecular weight category was an additional source of heterogeneity in this review. The study specific asthmagen is listed with the results on the test specific plots described later in this report.

Management review. Due to the extreme heterogeneity of the reporting of outcomes in the treatment articles, no meta-analytic techniques were employed and the summaries are descriptive in nature. Nonetheless, some re-organization of the data as reported was required to homogenize the information provided in the descriptive summary.

For continuous outcomes such as FEV₁ and some measures of BHR, a mean was used as the measure of central tendency when available. In cases where the mean was not reported or could not be obtained from an accompanying graph, the following substitutions (in order of preference) were used: median and midpoint of the range or IQR. When possible, a mean was calculated from IPD when it was available in a tabular format or plotted on a graph.

Similarly, standard deviation (SD) was used as the measure of variation. In cases where the SD was not reported or could not be determined from a graph, the following substitutions were used: (75^th percentile - 25^th percentile)/1.35 when IQR was reported; (maximum - minimum)/4 when range was reported and sqrt(n)*standard error of the mean (SEM) when SEM was reported. If individual data were available in tabular or graphic form, the SD was directly calculated.

Baseline characteristics and outcomes were grouped in various ways for the studies. When possible the data were grouped according to exposure status in follow-up as follows: continued exposure, reduced exposure, or ceased exposure. We used the following formulas to combine data when necessary:

Exposure Group mean=(n₁*mean₁)+(n₂*mean₂ )+...(n_k*mean_k )/(n₁+n₂...n_k)

Exposure Group SD=sqrt(((n₁-1)*(SD₁)²+(n₂-1)*(SD₂)²+...(n_k-1)*(SD_k)²)/(n₁+n₂...+n_k-k))

Because it was usually not reported, the mean difference between baseline (diagnosis) and follow-up was calculated as the difference between the mean at follow-up and the mean at baseline. Standard deviation of the difference was estimated assuming a between patient correlation of rho=0.5 between the baseline and follow-up values using the formula:

Var(A-B)=sqrt(VarA+VarB-2*rho*sqrt(VarA*VarB)) where Var(X)=SD(X) ²

For economic outcomes, currency measures were converted into US dollar (USD) equivalents based on exchange rates for the reported currency in the year of the studies' publication (http://www.oanda.com/convert/classic).

Description of Included Studies

Table 6

Description of studies for the diagnosis of OA

Table 6

Description of studies for the diagnosis of OA

	References
Published in language other than English
Italian	180,181,182,107,183,184,185,186
German	103,85,187,188,189
French	190,191,192,193
Chinese	194
Finnish	104
Country of study origin
United Kingdom	34,195,101,99,196,197,198,102,199,91,200,201,202,203,98,204,205,206,207,208,209
Canada	210,211,212,213,214,96,215,94,216,111,35,110,146,217,93,43,218,95,219
Italy	180,181,182,220,221,222,113,107,183,223,184,224,225,185,186,226,227,228
Germany	106,103,85,229,230,187,231,188,232
Korea	233,234,235,236,151,86,237,236,238
Spain	88,87,239,97,240,241,90,242
United States	243,244,245,115,246,247,114,248
Finland	249,250,251,104,252,160,253
France	190,191,254,192,193,255
Reference standard other than SIC
Clinical diagnosis	34,243,182,91,200,98,233,184,256,257,241,209,160,248,102,115,246,207
Serial PEFR	258
Suspected asthmagen
Low molecular weight
Di-isocyanates	243,103,210,99,196,244,214,181,222,194,215,245,249,233,255,183,223,247,224,205,206,186,226,237,227,114,90,209,232
Chemical	198,102,91,203,115,204,230,246,97,256,151,259,207,248
Wood dust	192,96,260,239,216,43,225,218,241,228,219
High molecular weight
Flour	87,211,180,254,229,257,187,188,189
Animal/Bird	261,262,104,160,253
Latex	85,263,264,105
Mixed	34,190,106,191,265,220,221,113,200,94,201,193,202,107,266,111,35,146,217,93,184,112,95,251
Comparison tests
Single NSBPT	88,87,190,106,103,210,191,211,265,196,197,180,213,214,254,181,192,198,220,260,221,222,113,91,215,201,193,85,245,115,234,258,255,261,262,216,266,111,35,204,110,146,217,183,223,267,229,230,184,247,224,97,256,250,205,206,43,225,185,186,226,218,268,235,236,151,86,237,238,227,257,240,114,228,90,263,231,89,259,264,105,252,232,189,253
Specific skin prick test	88,87,195,211,26,180,212,213,214,181,192,182,198,199,220,260,91,194,94,193,202,239,203,85,249,234,261,262,107,216,111,35,204,110,146,217,183,223,267,229,230,224,97,256,206,43,112,225,185,226,218,268,235,236,151,86,237,236,238,227,257,240,241,251,228,187,231,188,89,259,207,208,242,104,209,264,105,252,232,189,253,248
Serum specific IgE	88,87,190,195,210,211,244,212,213,214,254,102,199,220,194,215,239,203,85,245,249,233,234,261,262,267,229,246,256,225,185,218,268,235,236,86,237,236,238,227,240,241,251,228,90,188,259,242,104,209,219,252,160,232,189,253,248
Eosinophil counts	87,26,192,220,260,113,94,202,258,262,111,110,246,250,43,112,268,86,114,228,259
Serial PFT	34,101,99,96,200,98,35,110,217,93,97,250,151,95,228,207,255,246

Methodological Quality of Included Studies

Table 7

Methodological quality of studies for the diagnosis (more...)

Table 7

Methodological quality of studies for the diagnosis of OA

	References
Subject selection
Random or consecutive	34,96,221,98,249,261,266,110,146,230,250,43,104,105,160,253
Not reported	190,195,103,191,197,244,180,213,192,182,220,222,194,193,239,107,183,223,184,256,206,185,186,226,235,151,95,227,251,114,187,90,263,219,252,232,189
Data collection
Retrospective	34,99,244,213,249,35,217,93,224,250,225,240,242,209
Could not be determined	201,85,107,266,188,189
Blinded assessment of outcome
Reference standard and comparison test	111,35,110
Reference standard or comparison test	87,210,99,265,212,182,96,113,94,201,98,258,262,204,217,256,43,268,236,209,105
Inadequate	194,226,252,248
Adequately described reference standard	88,87,34,106,103,210,211,26,101,99,265,180,212,213,214,254,181,192,96,199,220,260,221,91,194,215,94,201,193,202,239,203,85,245,115,249,234,255,261,262,216,266,111,35,110,146,93,183,223,267,229,230,184,246,224,97,256,250,205,206,43,225,185,186,226,218,235,236,151,86,237,236,238,227,240,251,228,90,231,89,242,104,209,264,105,252,160,232,189,253,248
Differential bias	88,195,211,212,113,200,201,203,234,261,107,35,146,93,246,205,236,86,236,228,90,208,242,219,264,160
Could not be determined	190,103,26,180,214,181,199,91,194,98,216,250,186,218,227,241,187,263
Partial verification bias	88,190,195,211,214,182,199,113,91,200,98,85,233,234,204,110,146,183,184,225,185,236,86,251,228,231,188,104,219,264,189
Could not be determined	243,103,26,180,181,198,220,194,258,107,216,250,206,186,227,241,187,263
Asthma medication stopped prior to testing	88,87,265,254,96,220,221,113,91,215,94,202,245,115,255,261,216,35,146,223,267,229,230,247,224,97,250,43,226,218,268,237,227,240,251,90,231,89,242,104,253
Source of funding
Government	243,211,96,113,94,115,234,266,223,268,227,219
Foundation	195,216,111,256,43,240,160
Internal	261,235,86,236,114
Private	205,206,238,105
Other (including multiple sources)	34,210,244,212,214,200,245,247,253,248,88,254,203,229,237,257,264,232

Differential bias is likely to have occurred in 26 of the studies and could not be assessed in 18 studies; the other studies did not have this bias. Partial verification bias was present in 31 studies and it was unclear if it had occurred in a further 18 studies. Partial verification bias did not occur in the remaining 75 studies. Forty-one of the studies reported that workers stopped, or attempted to stop, asthma medication prior to diagnostic testing. Funding was not reported in 78 of the 124 studies. Among studies with one source of funding, funding was most commonly provided by a government agency; 10 studies had multiple sources of funding.

Quantitative Results

One hundred thirty-one publications reporting data on 124 cohorts met the inclusion criteria for the diagnostic component of this review. After selecting the most efficient comparison within a study for each comparison test, and ensuring that a comparison test was only reported once for a cohort, the most frequently reported comparison tests included single NSBP test (n=61), specific SPT (n=47), serum specific IgE antibodies (n=41), clinical diagnoses (n=9), serial pulmonary function tests (generally PEFR) (n=9), eosinophil count (n=6), and serial NSBP test (n=6).

In addition, some studies reported sufficient information to investigate the utility of combining comparison tests. These combinations primarily consisted of single NSBP test with SPT and/or serum specific IgE. For the most frequently reported comparisons, results based on the molecular weight of the suspected agent are summarized below and individual study results have been plotted in figures. In addition, pooled results from less frequently reported comparisons and combinations of comparison tests are reported. However, due to the between-study heterogeneity, which can be seen in the figures, pooled results should be interpreted with caution. Finally, because the majority of the workers included in the individual studies were screened-in, the results presented below are probably best interpreted as the patients also being “positive” for history, questionnaire, and/or referral to a specialist. For example, the comparison should be interpreted as screening and single NSBP test versus screening and SIC.

A summary of sensitivity and specificity of comparison tests that used SIC as the reference test is provided in Appendix H.

Figure 1. Sensitivity and specificity extracted from (more...)

   Figure 1. Sensitivity and specificity extracted from studies comparing single NSBP test to SIC among LMW asthmagens

Figure 2. Sensitivity and specificity extracted from (more...)

   Figure 2. Sensitivity and specificity extracted from studies comparing single NSBP test to SIC among HMW asthmagens

Figure 3. Sensitivity and specificity extracted from (more...)

   Figure 3. Sensitivity and specificity extracted from studies comparing single NSBP test to SIC among mixed asthmagens

Figure 4. Sensitivity and specificity pairs from studies comparing (more...)

   Figure 4. Sensitivity and specificity pairs from studies comparing NSBP test with SIC

The size of the plotting character is proportional to the number of patients in the group and the color indicates the molecular weight of the suspected asthmagen (white=HMW, black=LMW, grey=mixed).

Figures 1

2

3

Figure 4

Among the 37 studies that investigated patients exposed to LMW agents, 24 reported both sensitivity and specificity. The pooled estimate of sensitivity was 66.7 percent (95% confidence interval [CI]: 58.4 to 74.0 percent) and of specificity was 63.9 percent (95% CI: 56.1 to 71.0 percent). Pooled estimates for studies that reported only sensitivity were higher (n=13; 76.6 percent; 95% CI: 59.0 to 88.2 percent).

Of the 13 studies that reported results from investigations of HMW agents, 10 reported sensitivity and specificity. The pooled estimate for sensitivity was 79.3 percent (95% CI: 67.7 to 87.6 percent) and for specificity was 51.3 percent (95% CI: 35.2 to 67.2 percent). The estimated sensitivity in the five studies reporting only this data was similar (75.5 percent; 95% CI: 56.4 to 88.1 percent).

Nine studies reported results for various suspected agents of differing molecular weights and five studies reporting both sensitivity and specificity. For these mixed populations, the pooled estimate of sensitivity was 83.7 percent (95% CI: 66.8 to 92.9 percent) and specificity was 48.4 percent (95% CI: 25.9 to 71.6 percent). Sensitivity was lower in the three studies reporting only this value (43.7 percent; 95% CI: 10.9 to 83.0 percent).

Specific skin prick test versus SIC. Forty-seven studies reported comparisons of specific SPT to SIC.

Figure 5. Sensitivity and specificity extracted from (more...)

   Figure 5. Sensitivity and specificity extracted from studies comparing specific SPT test to SIC among LMW asthmagens

Figure 5

Figure 6. Sensitivity and specificity extracted from (more...)

   Figure 6. Sensitivity and specificity extracted from studies comparing specific SPT test to SIC among HMW asthmagens

Figure 6

Figure 7. Sensitivity and specificity extracted from (more...)

   Figure 7. Sensitivity and specificity extracted from studies comparing specific SPT test to SIC among mixed asthmagens

Figure 7

Figure 8. Sensitivity and specificity pairs from studies comparing (more...)

   Figure 8. Sensitivity and specificity pairs from studies comparing SPT to SIC

The size of the plotting character is proportional to the number of patients in the group and the color indicates the molecular weight of the suspected asthmagen (white=HMW, black=LMW, grey=mixed).

Figure 8

Figure 9. Sensitivity and specificity extracted from (more...)

   Figure 9. Sensitivity and specificity extracted from studies comparing serum specific IgE test to SIC among LMW asthmagens

Figure 10. Sensitivity and specificity extracted from (more...)

   Figure 10. Sensitivity and specificity extracted from studies comparing serum specific IgE test to SIC among HMW asthmagens

Figure 11. Sensitivity and specificity extracted from (more...)

   Figure 11. Sensitivity and specificity extracted from studies comparing serum specific IgE test to SIC among mixed asthmagens

Figure 12. Sensitivity and specificity pairs from studies comparing (more...)

   Figure 12. Sensitivity and specificity pairs from studies comparing specific IgE with SIC

The size of the plotting character is proportional to the number of patients in the group and the color indicates the molecular weight of the suspected asthmagen (white=HMW, black=LMW, grey=mixed).

Figure 9

Figure 10

Figure 11

Figure 12

Eleven out of the 21 studies considering LMW agents reported both sensitivity and specificity; the pooled estimate of sensitivity was 31.2 percent (95% CI: 22.9 to 40.8 percent) and of specificity was 88.9 percent (95% CI: 84.7 to 92.1 percent). Of the 10 studies that only reported sensitivity, the pooled estimate was 35.9 percent (95% CI: 23.2 to 50.9 percent).

Sensitivity was higher in the studies where HMW agents were examined; the pooled estimate of sensitivity was 73.7 percent (95% CI: 63.9 to 81.0 percent) for the nine studies reporting sensitivity and specificity and 81.7 percent (95% CI: 57.8 to 93.5 percent) for the nine studies reporting sensitivity alone. The pooled estimate of specificity was 79.0 percent (95% CI: 50.5 to 93.3 percent).

The two studies using a variety of molecular weight agents reported both sensitivity and specificity. The pooled estimate for sensitivity was 85.1 percent (95% CI: 40.3to 98.0 percent) and of specificity was 61.2 percent (95% CI: 7.0 to 97.1 percent).

Combined results with single NSBP test, serum specific IgE, and specific SPT compared to SIC. When possible, results were combined for the most frequently reported comparison tests. In the first assessment, all tests in combination had to be positive for the combined result to be considered positive. If any result was negative, the combination testing was considered negative. We report results from studies reporting sensitivity and specificity for LMW and HMW agents.

When a single NSBP test and specific SPT were considered in combination, four studies investigated patients exposed to HMW agents.⁸⁵–⁸⁸ The pooled estimate of sensitivity was 60.6 percent (95% CI: 21.0 to 89.9 percent) and of specificity was 82.5 percent (95% CI: 54.0 to 95.0 percent) respectively. Sensitivity was 100 percent (95% CI: 74.1 to 100 percent) and specificity was 80 percent (95% CI: 49.0 to 94.3 percent) in the one study that investigated green tea (a LMW agent).⁸⁹

Single NSBP test and serum specific IgE results could be combined for only three studies reporting both sensitivity and specificity. For two studies of HMW agents, the pooled sensitivity was 35.6 percent (95% CI: 1.2 to 96.1 percent) and 84.6 percent specificity (95% CI: 48.2 to 97.0 percent). ^{86, 87} The third study examined OA caused by isocyanates and resulted in 0 percent sensitivity (95% CI: 0 to 49.0 percent) and 100 percent specificity (95% CI: 61.0 to 100 percent).⁹⁰

In a second analysis, the combination of a positive single NSBP test and either a positive specific SPT or positive serum specific IgE was considered positive. Three studies of HMW agents yielded results which could be combined in this manner ⁸⁶–⁸⁷ The pooled estimate of sensitivity was 60.4 percent (95% CI: 11.8 to 94.5 percent) and specificity 81.5 percent (95% CI: 47.8 to 95.5 percent).

Other comparisons. Nine studies reported serial pulmonary function tests versus SIC; all but one study⁹¹ recorded PEFR. Of these, five studies investigated mixed agents and reported both sensitivity and specificity.^{35, 92}–⁹⁵ The pooled estimate of sensitivity was 63.6 percent (95% CI: 43.4 to 79.9 percent) and of specificity was 77.2 percent (95% CI: 66.5 to 85.2 percent). One study of a LMW agent reported 86.7 percent (95% CI: 59.5 to 96.6 percent) sensitivity and 90 percent (95% CI: 53.3 to 98.6 percent) specificity.⁹⁶ Two other studies only reported sensitivity resulting in a pooled estimate of 56.2 percent (95% CI: 17.2 to 88.8 percent).^{91, 97} Finally, one study of a HMW agent reported 100 percent (95% CI: 56.6 to 100 percent) sensitivity and no results for specificity.⁹⁸

All nine studies that compared clinical diagnosis to SIC reported both sensitivity and specificity. Clinical diagnosis ranged from physician assessment to a combination of tests, which may have included pulmonary function tests, NSBP test, symptom questionnaires, etc. Five studies investigated LMW agents and the pooled estimate of sensitivity was 93.6 percent (95% CI: 85.0 to 97.5 percent) and of specificity was 68.9 percent (95% CI: 54.7 to 80.3 percent).⁹⁹–¹⁰³ The pooled estimates of sensitivity and specificity were 93.7 percent (95% CI: 69.3 to 99.0 percent) and 32.3 percent (95% CI: 7.5 to 73.8 percent) respectively in the two studies reporting results for HMW agents.^{104, 105} Combined results of the two studies considering agents of various molecular weight yielded a sensitivity of 95.1 percent (95% CI: 86.8 to 98.3 percent) and specificity of 47.7 percent (95% CI: 26.7 to 69.7 percent).^{106, 107}

Six studies reported serial NSBP tests compared to SIC and all reported sensitivity and specificity. Pooled results from the three studies investigating mixed agents yielded 50 percent sensitivity (95% CI: 35.5 to 64.5 percent) and 66.8 percent specificity (95% CI: 53.3 to 78.0 percent).^{92, 95, 108} Two studies involved only patients exposed to LMW agents; pooled sensitivity was 67.5 percent (95% CI: 42.6 to 85.3 percent) and specificity was 65.6 percent (95% CI: 41.1 to 84.0 percent).^{109, 110} A study of OA caused by oil seed rape flour (a HMW agent) yielded 100 percent sensitivity and specificity.⁸⁷

Six studies reported eosinophil counts from sputum, blood, or broncho-alveolar lavage versus SIC of which four considered various agents causing OA. Of these, three^{94, 111, 112} reported sensitivity and specificity. The pooled estimate of sensitivity was 54.9 percent (95% CI: 23.7 to 82.7 percent) and of specificity was 72.3 percent (95% CI: 54.1 to 85.3 percent); one study reported 100 percent sensitivity only.¹¹³ Two studies of LMW agents reported sensitivity only.^{26, 114} The pooled estimate of sensitivity was 53.1 percent (95% CI: 10.3 to 91.8 percent).

RADS. Only one study examined workers with RADS.¹¹⁵ Fifty-six hospital workers were exposed to 100 percent acetic acid; within 24 hours eight workers met the RADS criteria. Eight months later, a questionnaire was administered to determine the degree to which workers experienced respiratory symptoms after the chemical spill. Fifty-one workers completed the survey. Approximately 9 months after the spill, a NSBP test was performed among 24 of the exposed workers, including seven of the eight workers who met the criteria for RADS (persistent respiratory symptoms). Four of the seven workers had a positive methacholine challenge, defined as provocative dose causing a 20 percent drop in FEV₁ [(PD₂₀) FEV₁] >20 percent (no dose cut-off was reported).

None of the included studies reported cost of diagnosis, time to complete diagnosis, or the presence of adverse events.

Description of Included Studies

Table 8

Description of studies for the management of OA

Table 8

Description of studies for the management of OA

	References
Published in language other than English
German	172
Italian	140
Spanish	169
Country of study origin
United Kingdom	158,131,58,163,141,22,159,177,176,269,173
Canada	134,136,143,55,147,132,179,4,270,135
Italy	133,145,166,144,130,153,137,56,148,140
United States	142,167,164,174,175,178
Finland	116,152
France	57,139
Germany	170,172
Korea	151,171
Spain	165,169
Other	271,149,154,150
Suspected asthmagen
Low molecular weight
Di-isocyanates	142,163,143,141,133,169,144,130,153,171,137,139,148,269
Chemical	158,174,175,4,170,165,151,116,56,173
Other	131,150,172,149,55
High molecular weight
Latex	167,178,138
Animal/Bird	177,152
Flour	154,176
Other	271,132,159
Mixed	134,57,58,164,136,147,179,270,145,166,135,22,140
Subject source
Clinic	134,57,158,131,58,164,154,178,136,143,55,141,147,179,4,145,166,165,169,151,171,135,137,56,139,150,140,152,138
Workplace	142,175,163,170,172,149,153,116,159,177,269,173
Not reported	167,174,271,132,270,133,144,130,22,148,176
Exposure Status
Removed	134,57,158,167,131,58,164,154,174,175,163,136,143,55,141,147,132,179,4,270,133,170,172,166,165,169,144,130,153,151,171,135,116,137,56,139,22,148,159,150,138,173
Reduced	57,142,167,131,218,175,163,141,170,166,153,272,116,137,139,176,138,269
Continued exposure	57,167,58,55,133,145,170,166,149,144,56,139,150,176
Used PPEs	57,167,178,165,169,139,177,152
Medications (+/- removal)	271,140
Outcomes
NSBP tests	134,142,158,131,43,59,154,141,147,132,4,270,133,145,170,172,166,165,149,144,130,153,151,155,135,116,137,56,139,148,150,140,138,173
Pulmonary function tests	134,142,158,131,59,154,174,175,163,55,141,147,132,4,133,145,170,166,165,144,135,116,137,56,139,150,177,140,152,138,269,173
Questionnaires	134,57,142,158,131,43,59,58,164,154,174,175,136,143,141,147,132,170,166,135,56,139,22,138,269,173
Specific skin prick test	158,167,168,154,136,143,141,170,165,144,135,137,56,177,173
SIC	178,136,143,141
	133,156,165,144,153,56,148,150,138
Serum specific IgE	158,154,174,175,271,136,143,132,171,272,159,173

The most commonly identified asthmagens in these studies were chemicals, of which 14 of the studies examined di-isocyanates. Thirteen studies included workers with OA caused by various agents. Workers were most often recruited from an OA clinic or the workplace. The median sample size of the included studies was 26 workers (range: 3–1011). Length of follow-up was variable within and between studies.

Methodological Quality of Included Studies

Table 9

Methodological quality of studies for the management (more...)

Table 9

Methodological quality of studies for the management of OA

	References
Report IPD	142,158,174,163,271,178,136,143,132,4,133,165,144,130,153,151,171,135,116,56,139,148,159,150,177,152,173
Data collection
Prospective	134,57,142,158,167,131,55,58,164,154,174,175,163,271,178,136,143,141,147,132,179,4,270,133,145,170,172,166,165,149,169,144,130,153,151,171,135,116,137,56,139,148,159,150,177,176,140,152,138,269,173
Retrospective	22
Source of funding
Government	142,158,167,164,271,166,130,153,22,148
Other	174,175,178,136,143,4,133,145,171,135,138
Not reported	134,57,131,55,58,154,163,141,147,132,179,270,170,172,165,149,169,144,151,116,137,56,139,159,150,177,176,140,152,269,173

Table 4

Summary of components of Downs and Black score

Table 4

Summary of components of Downs and Black score

	Reporting	External Validity	Bias	Confounding	Power	Overall
Maximum Score	11	3	7	6	2	29
Actual Score Mean (SD)	8.2 (1.8)	1.6 (1.0)	3.9 (1.1)	2.3 (1.1)	0.5 (0.5)	16.4 (4.0)

Abbreviations: SD = standard deviation

Apart from one study, data were collected prospectively. Approximately half (27/52) of the studies provided some IPD. When reported, the most common funding source was provided by a government agency; 31 of the studies did not disclose their funding source and few were industry sponsored.

Qualitative Results

Cohort Studies

Figure 13. Average percent predicted FEV₁ at baseline (more...)

   Figure 13. Average percent predicted FEV₁ at baseline for groups of patients who remained exposed, were removed from contact, or reduced exposure to the suspected asthmagen

The size of the plotting character is proportional to the number of patients in the group and the color indicates the molecular weight of the suspected asthmagen (white=HMW, black=LMW, grey=mixed).

Lung function. Twenty-seven studies reported FEV₁ at diagnosis (see Appendix E: Evidence Table E-7) and we attempted to synthesize the change in lung function over time. In order to assess this outcome, we examined the difference between the mean percent predicted FEV₁ at follow-up and baseline in relation to average length of follow-up. Baseline mean percent predicted FEV₁ of the included studies is shown in Figure 13.^{55, 58, 130}–¹⁴⁷

Severity at diagnosis. Of the 17 studies (n=666 patients) where patients were completely removed from exposure to the offending asthmagen, 16 reported a mean baseline percent predicted FEV₁ >80 percent, indicating primarily mild impairment to normal pulmonary function. Among the studies examining patients who remained fully exposed, six of seven studies reported mean baseline percent predicted FEV₁ >80 percent. Four of the five studies describing subjects with reduced exposure reported mean baseline percent predicted FEV₁ >80 percent. Based on these findings, it does not appear that lung function at diagnosis was associated with exposure status during follow-up. No specific pattern emerged when we considered the molecular weight of the asthmagen studied.

Figure 14. Difference between average percent predicted FEV (more...)

   Figure 14. Difference between average percent predicted FEV₁ in follow-up and average percent predicted FEV₁ at baseline plotted against average length of follow-up by exposure status in follow-up

Improved percent predicted FEV₁ over time is indicated by values greater than zero (0). No change in percent predicted FEV₁ in indicated by a dashed line at percent predicted FEV₁=0. Observations from cohorts with multiple follow-up visits are joined by a dotted line. The size of the plotting character is proportional to the number of patients in the group and the color indicates the molecular weight of the suspected asthmagen (white=HMW, black=LMW, grey=mixed).

Follow-up. Less than half of the studies where patients were removed from exposure^{58, 132, 138}–^{140, 144, 146} or had reduced their exposure^{131, 138, 139}, had improved FEV₁ over time (change >0) (Figure 14). Only one study where patients remained exposed to the work environment had a positive change in mean percent predicted FEV₁.¹⁴⁴ No specific pattern emerged when we considered the molecular weight of the asthmagen studied.

NSBP test. A variety of test characteristics were used to describe non-specific BHR. Generally, hyper-responsiveness was defined as the provocative concentration or dose of histamine or methacholine required to elicit a pre-determined change in FEV₁, usually a 15 or 20 percent decline (PC_15/20 and PD_15/20, respectively). In most cases, PC₂₀ values <16 mg/mL were considered to reflect significant BHR; however, cutoff levels of <8 mg/mL and <32 mg/mL were also reported. These results were eventually reported as either the number of patients who were hyper-responsive at a specified concentration cut-off value (e.g., <8 mg/mL, <16 mg/mL, <32 mg/mL) or as the mean or geometric mean of the concentrations that produced the required FEV₁ drop. Of the 13 studies that reported the presence of hyper-responsiveness at diagnosis, between 36 percent and 100 percent of patients with OA were classified as hyper-responsive.^{56, 130, 132, 134, 135, 141, 144, 148}–¹⁵²

Follow-up. The change in hyper-responsiveness over time (i.e., from baseline) was investigated by calculating the ratio of average hyper-responsive concentration at follow-up to the average baseline concentration. This measure was chosen because it is independent of the measurement unit and because the baseline measures of NSBP test varied. Therefore, all studies that reported or provided sufficient information to calculate a mean or geometric mean of a hyper-responsiveness measure at baseline and at follow-up were included. A ratio greater than 1.0 indicates improved hyper-responsiveness (follow-up/baseline) because the average concentration to achieve the specified bronchial response was greater at follow-up than at diagnosis. Conversely, a ratio less than 1.0 indicates worsening hyper-responsiveness because the average concentration to elicit a bronchial response was lower at the follow-up visit, signifying that the airways of patients were more hyper-responsive.

Figure 15. Ratio of mean NSBP test at follow-up visit (more...)

   Figure 15. Ratio of mean NSBP test at follow-up visit to mean NSBP test at baseline plotted against average length of follow-up by exposure status in follow-up

Improved hyper-responsiveness is indicated by values greater than 1. No change in responsiveness is indicated by a dashed line where the ratio=1. Observations from cohorts with multiple follow-up visits are joined by a dotted line. The size of the plotting character is proportional to the number of patients in the group and the color indicates the molecular weight of the suspected asthmagen (white=HMW, black=LMW, grey=mixed).

Figure 15 demonstrates that 14 of 15 studies of individuals who were removed from exposure had decreased hyper-responsiveness at follow-up.¹³⁰–^{133, 135}–^{137, 143, 148, 153}–¹⁵⁷ Conversely, patients who remained exposed were overall more hyper-responsive at follow-up testing, although two of the five studies showed improved hyper-responsiveness at follow-up.^{133, 150} There were insufficient data (three studies) to draw conclusions about change in hyper-responsiveness among patients with reduced exposure. No specific pattern emerged when we considered the molecular weight of the asthmagen studied.

Medications. Medication needs were used as a proxy measure for disease severity and continued asthma symptoms. Twenty-two studies reported some medication data at baseline or diagnosis. The number of patients taking specific types of medications (bronchodilators, corticosteroids, etc.) was most frequently reported in insufficient detail to form an analytic approach.^{58, 134, 136, 146, 158}–¹⁶² Eight studies reported the percent of patients using asthma medications at baseline; however, did not specify type.^{49, 123, 131, 132, 141}–^{143, 163} Two studies categorized patients by frequency of medication use^{130, 138} and three studies reported a mean score based on the frequency of medication use.^{56, 100, 137} Follow-up data were reported in a similar manner.

There was no clinically meaningful way to combine the continuous measure outcomes with the frequency data. In addition, not all studies reported medication use at baseline and follow-up, thus we explored the percent of patients taking medications at follow-up. When the usage level was specified, only those that reported daily or frequent use of medications were counted.^{138, 141, 142, 147, 158, 164}

Figure 16. Percentage of patients taking asthma medications in (more...)

   Figure 16. Percentage of patients taking asthma medications in follow-up plotted against average length of follow-up by exposure status in follow-up

Using medication need as a surrogate for disease severity, decreasing percentages of subjects on medication over time may indicate decreasing disease severity. Observations from cohorts with multiple follow-up visits are joined by a dotted line. The size of the plotting character is proportional to the number of patients in the group and the color indicates the molecular weight of the suspected asthmagen (white=HMW, black=LMW, grey=mixed).

Follow-up. Of the patients removed from exposure, the percentage of patients using medication at follow-up ranged from 17 to 100 percent (Figure 16).^{56, 98, 116, 130}–^{132, 134}–^{136, 138, 141, 147, 154, 158, 162, 165}–¹⁶⁸ Within this group of 17 studies, there was no indication that fewer patients were using medication as time from removal increased. Only nine studies reported medication use among patients who remained exposed^{131, 139, 141, 157, 162} or had reduced exposure to the asthmagen^{131, 138, 139, 142, 165}; no clear pattern of changes in medication use emerged.

Symptoms/improvement. Thirty publications reported a variety of symptom outcomes by intervention status of continued exposure, reduced exposure, and cessation of exposure. Outcome measures included mean symptom scores^{137, 138, 145}, categorical symptom scores^{164, 166}, and the number of subjects who were asymptomatic or recovered, remained symptomatic, or had specific symptoms.^{50, 56, 58, 131}–^{133, 139, 142, 144, 148, 157}–^{159, 169}–¹⁷⁶ Four studies included statements regarding non-quantified “improvement”.^{136, 149, 154, 163} Due to the disparity of outcomes, only a qualitative assessment was possible.

Among the studies that examined workers removed from exposure the majority reported some improvement in workers' symptoms following removal.^{50, 56, 58, 131}–^{133, 136}–^{138, 144, 148, 154, 157}–^{159, 163, 164, 166, 169}–¹⁷⁴ This was measured by either a symptom score or reporting that the majority of patients had improved symptoms or were considered “asymptomatic”. The same pattern emerged in the nine studies describing symptoms for subjects whose exposure had been reduced by a workplace intervention.^{131, 138, 139, 142, 149, 163, 170, 175, 176} Very few studies reported a complete resolution of symptoms among the majority of workers.^{148, 159, 174}

Three groups examined the effectiveness of respiratory protective equipment in reducing symptoms caused by workplace exposures. In general, these papers reported that while respiratory protective equipment did reduce the severity of acute symptoms, they did not eliminate symptoms altogether.^{152, 169, 177, 178}

Finally, most of the studies of workers who remained exposed to the asthmagen at work showed that symptoms either remained stable or deteriorated with continued exposure.^{50, 58, 133, 137, 139, 144, 145, 157, 163} Only two studies reported improved symptoms among those who remained exposed to the asthmagen.^{166, 176}

Table 5

Economic consequences of occupational asthma

Table 5

Economic consequences of occupational asthma

Citation	Removed/Unemployed	Exposed or Reduced
Employment
Gassert et al. 164	38*/55	17+/55
Dewitte et al.59	33*/134
Income
Ameille et al.57	-50 percent (69/82^)	-19 percent (20/104^)
Moscato et al.166	-$4,203.72 USD/year	-$268.71 USD/year
Gannon et al.58	-$5,863.88 USD/year	-$3,820.27 USD/year
Gannon et al.58	-54 percent (56/78^)	-35 percent (14/34^)
Marabini et al.50	-$368.14 USD/month	-$256.38 USD/month
Marabini et al 50	-$609.96 USD/month*
	-20 percent (10/16^)	0 percent (6/20^+)
Insured
Gassert et al. 164	49/55
WCB Claim Acceptance
Vandenplas et al.138	3/7	1/5
Marabini et al.50	17/20 / 42/45	31/33
Pharmaceutical Costs/Month
Moscato et al.166	-$12.46 USD	+$13.17 USD

Abbreviations: USD = United States dollars

WCB = Workers' Compensation Board

Note: * = unemployed

+ = reduced exposure

^ = perception of reduced income

Socioeconomic consequences. Seven studies examined socioeconomic outcomes among workers with OA (Table 5). In all but three studies^{138, 164, 179}, the subjects were comprised of a group primarily removed from workplace exposure and a continued exposure group. Vandenplas et al.¹³⁸ compared two groups of workers: reduced or removed from exposure. In two instances, all of the workers were removed from the offending workplace.^{164, 179} Five of the studies included mixed causes of OA, while one study examined latex-induced OA¹³⁸, and the other included patients with red cedar asthma.⁵⁰ Time since OA diagnosis ranged from 1 year to 5.5 years.

Of the four studies that measured financial situation after OA diagnosis, all found that workers who were removed from the workplace causing their OA suffered a loss in income. In one study conducted, 85 workers who were removed from exposure were compared to 117 workers who continued to work for the same employer (46 used PPE, 20 had the same job, 38 workers were relocated, and 13 were on chronic sick leave).⁵⁷ The response rate to economic questions was high (89 percent). When compared to workers who were completely removed from the causative workplace, exposed workers were less likely to have suffered diminished earnings; the mean loss of annual income was significantly less among the exposed workers. Another study examined 25 workers with SIC-confirmed OA; 13 workers were removed from exposure and 12 workers continued their exposure.¹⁶⁶ After one year, a significantly greater number of removed workers reported deterioration in their economic situation. Compared to 1 year earlier, both monthly and annual income significantly decreased among the removed workers. Earnings also decreased in the workers with continued exposure; however, this difference was not significant. Interestingly, this study also reported that pharmaceutical expenses significantly decreased for those removed from exposure, while they increased among workers with continued exposure. A third study compared 78 workers with OA removed from the workplace to 34 workers with OA who continued to be exposed.⁵⁸ While 140 participants were identified, 112 completed the follow-up questionnaire. Fewer workers who were still exposed felt they had lost money than workers who were removed. The perceived median annual loss of income and the perceived percentage loss of annual income were higher among the workers who ceased workplace exposure. The authors did not perform any tests of statistical significance. The final study examined red cedar asthma workers who were exposed workers (48), non-exposed workers (27), and unemployed workers (53).⁵⁰ Monthly income among the unemployed cohort was significantly less than the working exposed and working non-exposed groups.

Vandenplas et al. compared 20 workers who had reduced levels of latex exposure to 16 workers who were removed from latex.¹³⁸ Among the reduced exposure group, 7/20 workers reported work disability, defined as changed or left work, while 11/16 of workers removed from latex exposure reported work disability and the remaining five workers were not working because of their latex-induced OA. More workers who were removed from latex reported a decrease in income compared with workers with reduced latex exposure. The actual reduction in earning was 20 percent among the non-exposed workers, while there was no reduction in earnings for the workers with reduced exposure.

Two studies assessed the rate of worker's compensation claims and associated acceptance.^{50, 138} In Vandenplas' et al. study, only 12 of the 36 workers stated they had attempted to seek compensation. The Belgian Workers' Compensation Board (WCB) more frequently approved claims among workers who were removed from the workplace than those who reduced their exposure. In contrast, the Canadian WCB approved more claims than its Belgian equivalent. Among those who filed for compensation, the acceptance rate was similar between workers with continued exposure, workers who ceased exposure, and the unemployed.⁵⁰

Two studies followed a cohort of workers who were removed from exposure. Gassert et al. conducted a study of 55 patients with “definite/probable” OA of mixed origins.¹⁶⁴ Seventy-two patients were identified and 55 were interviewed. At follow-up, approximately one-third of the subjects were employed and their exposure was reduced (17/55). The remaining subjects were unemployed. Prior to a diagnosis of OA, 54 workers had health insurance; at follow-up, 49/55 patients were not insured and had to pay for their asthma medication and care. The final study reported findings on 211 Quebec workers with OA caused by mixed exposures were awarded workers' compensation between 1986 and 1988. One hundred and thirty-four of the 211 eligible workers were followed up 2 years after removal from exposure. The average cost of disability/impairment for each worker was $35,529 USD. A significant portion of the workers were either unemployed (11/134) or took an early retirement (22/134).

Quality of life (QOL). Two studies measured quality of life. Vandenplas et al. examined 36 subjects with SIC confirmed latex induced OA.¹³⁸ At a median follow-up time of 56 months, 16 subjects were no longer exposed to latex and 20 subjects had reduced their exposure. Two methods to reduce latex exposure were employed: using less than 20 pairs of latex gloves in the room or department each week; and using low allergen sterile and latex-free examination gloves. QOL was measured using a French version of the Asthma Quality of Life Questionnaire (AQOL). AQOL did not significantly differ among those removed from exposure versus those who reduced their exposure.

In the second study, 211 workers with OA, 134 participated in a follow-up survey assessing QOL.⁹³ SIC or serial PEFR was used to diagnose Quebec workers with OA; all the workers were removed from the exposure and received compensation from the WCB. Ninety-one workers with OA were matched to 91 similar patients with non-occupational asthma. This study also utilized the AQOL that considers five components of asthma: self-determined activities that were limited by OA, other activities, symptoms, emotional function, and exposure to environmental stimuli. When compared to the control group, both the total and component scores of the QOL survey were significantly lower among workers with OA than those with non-occupational asthma.

RADS. Piirila et al. followed six men who were exposed to sulfur dioxide during an explosion at a pyrite mine.¹¹⁶ At follow-up, three patients continued to work in the same workplace, two had retrained and no longer worked underground, and one had retired because of respiratory disability sustained after the explosion; 13 years later, chest radiographs, spirometry, and NSBP testing were assessed. All of the workers, except one of the men who was retrained, continued to suffer from non-specific BHR. NSBP testing was not performed in one worker who continued in the same workplace because of dyspnea, wheezing, and moderate obstructive ventilatory impairment. There were no changes in chest radiographs.

A second study reported findings among a group of 29 men who were repeatedly exposed to chlorine over three months and subsequently diagnosed with RADS.⁴ The men were no longer exposed to chlorine and 20 men were assessed one-year post diagnosis. While percent predicted FEV₁ was not significantly different between diagnosis and follow-up, some of the workers showed improvement in non-specific BHR. Six workers had a significant improvement in PC₂₀ (3.2 fold difference from baseline to follow-up); non-specific BHR significantly deteriorated in one worker. Compared to diagnosis, significantly fewer workers required medication to control their OA symptoms.

Cohort

Figure 10

Figure 11

Third, similar results are demonstrated for the small group of studies where NSBP were recorded at baseline and at follow-up. Once again, those groups who remained exposed experienced continued deterioration in NSBP test results compared to baseline and over time. Almost all of those groups who report being removed from their workplace appeared to have improved compared to NSBP test results at the time of diagnosis. Improvement in non-specific BHR appears to be more impressive and progressive with time than FEV₁ results. Few groups appeared to deteriorate, suggesting that continued exposure may be required for non-specific BHR to continue to decline. Conclusions based on reduced exposure groups are complicated by the paucity of studies included in this comparison. No clear trend was identified based on the LMW versus HMW asthmagen division.

Fourth, while many studies reported symptoms and/or improvement, the definitions and measurements were too variable to analyze quantitatively. Symptoms appeared to persist and potentially worsen among workers who continued exposure. The picture is less certain among workers who were removed; there was evidence that while some workers' symptoms improve after removal from exposure, other workers failed to improve. Among workers who reduced their exposure, symptoms abated in some workers but the overall effect seemed to be persistence of symptoms. Because symptoms were often measured subjectively and using different methods, this outcome is problematic in determining the effect of removal, reduction, and continuation of exposure.

Medication use was a confounder and outcome that was incompletely reported in the majority of studies; however, using available data, we attempted to examine the three exposure status groups based on their medication use at follow-up. The conclusions from these studies are harder to draw, due to small study numbers. Overall, workers with OA, irrespective of subsequent exposure, often require medication treatment long after diagnosis. No clear trend was identified based on the LMW versus HMW asthmagen division.

Finally, the economic consequences of developing OA are impressive. From the published literature, those who leave the workplace clearly suffer economic repercussions of reduced income and/or unemployment. Workers who reduce their exposure or stay employed at the same workplace still appear to lose income over time, and their costs of medication increase. Furthermore, medical insurance coverage among workers, an important consideration in the United States, is reduced after the diagnosis of OA.

Trials

There were a limited number of clinical trials performed in the OA field. Overall, the 13 included clinical trials were of only moderate methodological quality. For example, while eight (62 percent) were randomized, none reported their method of randomization; seven (54 percent) were double-blinded, yet none described their methods of ensuring double-blinding; and only two (15 percent) demonstrated adequate concealment of allocation. All studies reported losses to follow-up and withdrawals.

From the 13 clinical trials examining OA interventions, the total number of study patients is 210, with the largest trial enrolling only 32 patients. One study involved immunotherapy, two involved reducing exposure via protective measures and the remaining were medication studies. Overall, the immunotherapy results, based on a small overall sample (n=30 patients) examining wheat flour antigen, suggest spirometric, immunological, and symptomatic improvement when workers experiencing OA are treated with immunotherapy. A comprehensive Cochrane Review supports benefit from this therapeutic approach. From 75 trials involving 3,188 patients with asthma, Abramson et al. demonstrated that immunotherapy reduced asthma symptoms and the use of asthma medications and improved bronchial hyper-reactivity.²⁷⁷ While this therapy may be as effective as ICS, it is complicated by its long duration (10–20 months), limited availability, and that many OA patients have a disease for which the relevant antigen for immunotherapy has yet to be identified. Finally, the possibility of adverse effects, such as allergic reactions and anaphylaxis, is a concern for some patients. It is likely most patients either would not qualify or elect to use other forms of therapy.

Trial evidence examining reduction in exposure is limited to only two clinical trials. Evidence does suggest that protective devices can reduce bronchial obstruction; however, they failed to provide complete protection and workers were required to be compliant with their use. In the situation of latex allergy-induced OA, the use of non-latex or low protein, powder-free latex gloves in the work environment appears to be a successful method of improving OA symptoms and outcome measures.¹²³

Finally, medication research in OA is limited and conclusions are difficult to draw. These medication trials suffer from small sample sizes, limited duration of treatment, and various dissimilar comparisons. Consequently, statistical pooling was not possible. In general, there is evidence that corticosteroid agents (both systemic and inhaled) are effective in the treatment of OA, although this was primarily evaluated in patients with di-isocyanate induced OA. Well recognized and parallel evidence is available regarding the effectiveness of corticosteroid treatments for chronic asthma from a variety of resources including guidelines and the Cochrane Library. Theophylline, a weak bronchodilator, reduced the severity of asthma exacerbations after SIC; however, airway responsiveness did not decrease. Other agents such as non-steroidal anti-inflammatories, calcium channel blockers, cromolyn, or placebo demonstrated limited or no benefit in the acute treatment of OA. Once again, these general effectiveness trends appear to be similar for OA and chronic asthma.

Diagnosis Review

The evidence concerning the harms (false positive and false negative) of diagnosis of OA is even less developed than the evidence for the benefits. Essentially, these patients have not been followed for the anxiety of an incorrect diagnosis, nor have the consequences been followed in any comprehensive manner. Costs of asthma care are largely borne by the patients, families, and the employers, and the costs of OA care are largely borne by the same groups. Falsely diagnosing respiratory symptoms as OA may force patients to inappropriately change work or become unemployed. Not diagnosing respiratory symptoms as OA when such is the case may lead patients to further asthmagen exposure, worsening health, impaired QOL, and later unemployment without compensation.

While SIC is considered the reference standard test in the diagnosis of OA, some studies did not have this result available for all patients, if at all. Some studies used a clinical “consensus” diagnosis to determine the presence of OA, which may or may not have included SIC in the diagnostic process, and it was usually not clear which patients had undergone SIC and why. In other studies, only data from patients who had a positive SIC result were reported. Two forms of bias may be present as a result of these characteristics: different reference standard bias and partial verification bias.⁷⁰ For HMW agents, we noted that the sensitivity generated from SIC positive subjects was similar to that generated from workers with suspected OA in two of the three comparison tests we considered in depth (single NSBP testing and specific SPT). Among LMW asthmagens, the sensitivity of SPT was substantially lower in studies when only SIC positive subjects were included in the study.

The studies included in the diagnosis review display considerable heterogeneity. This heterogeneity likely arises because many different asthmagens can cause OA and the diagnostic tests do not behave identically among the various asthmagens. Unfortunately, there were not enough studies to pool sensitivities and specificities by the specific asthmagen; however, in an effort to reduce heterogeneity, the results are presented by HMW and LMW and subgrouped by the specific asthmagen. Because of the heterogeneity between studies, pooled results for sensitivity and specificity are presented separately for each of the comparison tests. A drawback to this approach is that because the calculations are done independently, the pooled results of sensitivity and specificity are not explicitly paired as they are for each of the contributing studies. The plots of sensitivity/specificity pairs in ROC space demonstrate this artifact but do not allow for a link to the detail on specific asthmagen which we believe is valuable to the reader. Controversy exists with respect to pooling of heterogeneous data. Some would argue that the pooling in this setting is unhelpful and potentially misleading, while others believe this approach provides the best estimate of the test property. We have utilized random effects modeling, as appropriate for pooling heterogenous studies. However, we advise the reader to use caution in interpreting the results presented in this report, but believe that the utility of these values should be judged by the reader. We have provided sufficient data to recalculate the sensitivity and specificity for a number of specific asthmagens. This will allow clinicians to calculate the most appropriate pooled result for use in their practice.

Among the included studies, there were various definitions of a positive test result and different protocols were used to conduct the same test. For example, a positive test result for a single NSBP test was frequently reported as PC₂₀ or PD₂₀ FEV₁ ≤ 8 or 16 mg. Similarly for SPT, studies reported a positive SPT as ≥ 3 mm or compared the size of the reaction among workers with suspected OA to the size among a control population. For the purpose of this review, we treated one SIC methodology to be equivalent to another; this report does not attempt to evaluate the various methodologies used to conduct SIC.

A further limitation was that not all studies we identified were designed to be diagnostic studies and the data presented were not in a useful form to evaluate the diagnostic accuracy of a comparison test. That is, it had to be possible to generate a 2 × 2 or 2 × 1 table of the reference standard test result with a comparison test result based on one or more cut-offs of the comparison test. It is not possible to use results presented as a difference between mean values of the comparison test when grouped by the reference test result. In other cases, IPD data were available; however, the absence of an established cut-off value to define a “positive” test excluded these results. Resources and/or the long length of time between the publication of the results to the writing of this report precluded contact with most authors to obtain the necessary data in a usable form. This would have increased the number of studies that could be pooled in some of the comparisons.

Management Review

The main limitation of the management review is that the design of the included studies was weak. For example we found very few RCTs and the methodological quality of these was, at best, moderate. Most importantly, the interventions were generally divided into removal, reduced exposure, or continued exposure; however, the definition of these approaches differed and allocation was non-randomized.

Another limitation of the management studies is that the populations differed considerably. For example, while all groups attempted to confirm the diagnosis of OA, the methods used varied and the types of asthmagens that workers were exposed to differed. We attempted to retain the HMW versus LMW division employed in the diagnostic section of the review to investigate the heterogeneity of the results; however, this was not helpful. Treatment co-interventions were also incompletely reported in these studies.

Finally, the outcome assessments were often not comparable and tended to focus on short-term spirometric results (e.g., pulmonary function tests, NSBP test, SIC, etc.) rather than QOL. There was variability in length of follow-up within and between studies, making pooling difficult. Only four studies had repeated outcome measurements at different follow-up times for the same cohort.