METHODS

Case identification and follow up in the SENSOR asthma programme

Details on how WAA and NOA cases are identified in the SENSOR asthma programme have been published elsewhere.^5,8,20,21 Cases are identified from several different sources, including physician reports and hospital records. Programme staff administered questionnaires to identified cases. Core data elements about the case were collected by all states, including personal characteristics, pattern of asthma symptoms in relation to work, reported exposures associated with asthma symptoms, industry and occupation, history of allergies, and adverse impacts of having WRA.⁵

Case definition and classification

According to the standard SENSOR classification scheme, a person was defined as having WRA if he or she was diagnosed with asthma, and there was a temporal association between asthmatic symptoms and work. Cases were then further classified as either WAA or NOA. Individuals with WRA were classified as having WAA if the following conditions were met: (1) the affected individual experienced asthma symptoms or had treatment for asthma in the two years prior to entering a new work setting; and (2) they experienced an increase in asthma symptoms or increased use of their asthma medications after entering that new exposure setting.⁵ The new exposure setting could include starting a new job, a change of work processes, and/or the introduction of new agents in the workplace. WRA cases with no asthmatic symptoms or treatment for asthma in the two years before entering a new work setting were classified as having NOA, which included reactive airways dysfunction syndrome (RADS).

Average annual incidence rates

We calculated average annual incidence rates for major industrial and occupational categories for both WAA and NOA, as follows:

For each major industrial or occupational category:

We arrived at the number of employed persons with and without current asthma through a series of calculations. First, we obtained the current percentages of males and females in each major industrial or occupational category in the United States, which were available from the Bureau of Labor Statistics.²² We multiplied these percentages by the total number of persons working in each major industrial or occupational category for each of the four states.²³ The resulting estimated numbers of male and female workers in each industry or occupation for each state were then multiplied by the respective state specific percentages of males and females who had reported ever being told by a doctor that they had asthma, as indicated by the 2000 Behavioral Risk Factor Surveillance System (BRFSS) survey.²⁴ This yielded the estimated expected numbers of males and females who had ever had asthma in each major industrial or occupational category. Finally, we multiplied these numbers by the state specific percentages of males and females with current asthma who reported ever having asthma in the 2000 BRFSS survey. The resulting figures were the total estimated numbers of males and females in that state with current asthma for each major industrial and occupational category. These values were summed across the four states to obtain denominators for the overall WAA rates. We divided the number of WAA cases for all four states by this denominator to arrive at an estimated rate over the three year period. We then divided this rate by 3 to arrive at an estimated annual incidence rate. We arrived at the denominator for the NOA average annual incidence rate through the same series of calculations, except that the numbers of male and female workers in each major industrial and occupational category for each state were multiplied by the respective percentages of males and females who did not have current asthma, as indicated by the 2000 BRFSS survey.

Data management and analysis

Primary industry for each case had been coded by the states using the four digit 1987 Standard Industrial Classification (SIC) system developed by the US Department of Labor.²⁵ In addition, primary occupation had been coded following the 1990 Occupational Classification System from the US Bureau of Census.²⁶ Industrial and occupational categories had also been created by using major groupings previously developed by the US Department of Labor and the US Bureau of Census.

Agents had been coded by the states using the Association of Occupational and Environmental Clinics (AOEC) classification system.²⁷ Table 2 in Jajosky et al presented agents for both WAA and NOA.⁵ For WAA, the five most frequently reported agents in descending order were mineral and inorganic dusts, not otherwise specified (nos); indoor air pollutants; chemicals, nos; paint; and smoke, nos.⁵ The top five agents in descending order for NOA were indoor air pollutants; chemicals, nos; lubricants, which included metal working fluids; mineral and inorganic dusts, nos; and cleaning materials, nos.⁵ In the Jajosky et al article, agents were listed singly according to AOEC codes.⁵ For our analysis, we combined cleaning agents, nos (AOEC code 322.00), bleach (050.02), and household cleaners (322.04) into a single cleaning agents category. Similarly, we arrived at an inclusive diisocyanates category by combining diisocyanates, nos (221.00), toluene diisocyanate (221.01), and methylene diisocyanate (221.02). Cleaning agents and diisocyanates are two categories of exposure that have commonly been associated with work related asthma.

We analysed all data using the SAS System for Windows.²⁸ Tests of statistical significance were conducted using the χ², continuity corrected χ², and Fisher’s exact test for categorical data, and Student’s t test for continuous data. We considered all p values that were less than or equal to 0.05 to be statistically significant.

RESULTS

Source of case reports, demographic characteristics

Between 1993 and 1995, 1101 cases of WRA were identified by the four participating state SENSOR programmes. Of these, 210 (19.1%) were classified as having WAA, with the remaining 891 cases classified as NOA. In comparison with NOA cases, WAA cases were significantly more likely to be younger (mean age 38 years v 42 years), female (69% v 53%), and non-white (31% v 21%) (table 1). In addition, WAA cases were significantly more likely to have ever been told they had allergies (75% v 42%) or had reported a family history of allergies or asthma (57% v 35%). However, WAA cases were significantly less likely to have ever smoked cigarettes (39% v 53%) than NOA cases. About 90% of both WAA and NOA cases were identified from physician reports, with the remaining cases coming from hospital discharge records and other sources.

Primary industry and primary occupation

WAA cases were most likely to be employed in the service industry (40%), followed by manufacturing (22%) and public administration (16%) (table 2). The three most common industries for NOA cases were the same as WAA cases; however, the first two were reversed in order relative to the WAA cases: manufacturing (46%), services (29%), and public administration (10%). By primary occupation, WAA cases were most commonly employed in technical, sales, and administrative support professions (34%), followed by managerial and professional specialities (22%), operators, fabricators, and labourers (18%), and service (17%) professions (table 3). In contrast, NOA cases differed significantly (p < 0.0001) from WAA cases, with over a third employed as operators, fabricators, and labourers (36%), followed by technical, sales, and administrative support (18%), managerial and professional specialities (17%), and precision production, craft, and repair occupations (16%).

Average annual incidence rates for WAA and NOA

We calculated average annual incidence rates of WAA and NOA among employed persons with and without current asthma, respectively. The rate of WAA among employed persons with current asthma was 3.9 cases/10⁵/y, and the rate of NOA among employed persons without current asthma was 1.3 cases/10⁵/y. We also calculated average annual incidence rates for WAA and NOA for each major industrial and occupational category. For WAA cases, the highest rate was found among persons working within the public administration industry (14.2 cases/10⁵/y) (table 2). Manufacturing and public administration industries had the highest rates among NOA cases (3.2 cases/10⁵/y and 2.9 cases/10⁵/y, respectively). Incidence rates of WAA were similar among all major occupational categories, although operators, fabricators, and labourers, as well as those in service occupations, had slightly higher incidence rates (table 3). Operators, fabricators, and labourers had the highest average annual incidence rate for NOA cases by occupation (3.4 cases/10⁵/y) (table 3).

Agents

In California, Massachusetts, and New Jersey, up to three agents were recorded for each WRA case. In Michigan, up to two agents were recorded for each case. Overall, more than one agent was recorded for 64 (30%) WAA cases and 257 (29%) NOA cases. The most common agents for WAA cases included, in descending order: mineral and inorganic dusts, nos; indoor air pollutants; chemicals, nos; cleaning agents; paint; smoke, nos; and glues, nos (table 4). The most common agents for NOA were: diisocyanates; indoor air pollutants; cleaning agents; chemicals, nos; lubricants, nos; mineral and inorganic dusts, nos; and smoke, nos.

Consequences of work related asthma

A majority of both WAA (75%) and NOA (79%) cases reported still experiencing breathing problems at the time of the interview. Over half of all WAA and NOA cases (59% and 52%, respectively) had visited an emergency department, and about a fourth (23% of WAA and 25% of NOA) had been hospitalised for work related breathing problems.

A significantly higher proportion of NOA cases had applied for workers’ compensation at the time of the interview—with 40% of NOA cases and 21% of WAA cases, respectively (p < 0.0001). However, among the individuals whose applications had been decided, there was little difference between the two groups in the percentage of claims awarded, with 72% for WAA and 77% for NOA. WAA (73%) and NOA (71%) cases were equally likely to no longer be exposed to suspected agents in the workplace. Of those no longer exposed, NOA cases were twice as likely as WAA cases to have left the company (47% v 23%). Among both types of cases, medical reasons, such as workers’ compensation, disability, sick leave, and physician’s advice, were the most commonly reported motivations for leaving the company.

DISCUSSION

Strengths and limitations of study

The strengths and limitations of using the SENSOR database for WRA have been discussed elsewhere.⁵ Although the number of states participating in the programme is limited, SENSOR is currently the only available surveillance programme in the United States for detecting and monitoring trends in WRA. SENSOR has encouraged case identification and reporting by providing educational materials to physicians for detecting individuals with WRA, as well as keeping them up to date on the latest SENSOR surveillance and intervention activities.⁵ However, SENSOR was never intended to be a comprehensive or representative sample of all WRA cases in the United States; many cases are not reported due to lack of recognition by a healthcare professional or incomplete reporting.⁵ Accordingly, the average annual incidence rates reported in tables 2 and 3 are assumed to be underestimates of the “true” rates.

Differences between NOA and WAA cases in industry and occupation

It should first be noted that comparisons between WAA and NOA were made only across major industrial and occupational categories, which may fail to identify similarities and differences in more specific industries and occupations. However, there are several possible explanations as to why a major industrial or occupational category could account for a disproportionate number or rate of WAA or NOA cases. One reason might be the varying types and levels of exposures across workplaces. For example, industries with very potent agents and/or high level exposures would be expected to have high incidence rates for both WAA and NOA, which might explain the rates observed for manufacturing. However, those with asthma may self-select out of particular industries or occupations where they feel harmful exposures are likely (for example, the manufacturing industry) and self-select into industries or occupations where they feel they will not be exposed (for example, the service industry). Therefore, people with asthma who comprise the at-risk pool for WAA may not be equally distributed by industry and occupation, and consequently, exposure levels alone may not predict the frequency of WAA. In calculating incidence rates for WAA, we assumed that people with asthma were equally distributed by industry and occupation. If this assumption is incorrect, perhaps due to the healthy worker selection effect, some rates we presented would be underestimates and other rates would be overestimates.

Another question is why the average annual incidence rates of WAA and NOA were so high for the industrial category public administration. One potential explanation is the fact that public administration includes those working in fire and police protection (SIC 9224 and 9221, respectively). Henneberger and colleagues found that public administration had the highest rate of non-fatal work related inhalation injuries (16.4 inhalations per 10 000 workers) among all major industries.²⁸ When only firefighters and police were considered, this rate almost doubled to 31.5 inhalations per 10 000 workers.²⁹ Of the 1101 cases reported to SENSOR between 1993 and 1995, 122 cases were employed in public administration. Of those, 24% (n = 29) were in police or fire protection. According to private market research statistics, only 11.6% of those in public administration are employed in police or fire protection.³⁰ The apparent over-representation of police and firefighters in our data could have contributed to the increased rates for public administration.

Race could have also contributed to the high rate of WAA in the public administration industry. According to Mannino and colleagues, asthma proportionately affects more non-whites than whites.³¹ Recent labour statistics indicate that the public administration industrial category, on the whole, has proportionately more non-whites than the overall workforce (16.7% v 11.3%, respectively).²² Additionally, SENSOR data indicated that WAA cases in public administration were about 50% more likely to be non-white than WAA cases in all other industries (42% v 29%, respectively).

Prevention

At least two different approaches could be taken to target primary prevention efforts. One approach would be to focus on those industries that have the greatest number of persons with WAA or NOA. The industry groups with the most cases were services (40%) for WAA and manufacturing (46%) for NOA. Services and manufacturing combined comprised 62% of WAA cases and 75% of NOA cases. A second approach would be to focus on those industries which had the highest risk as indicated by average annual incidence rates. Public administration and manufacturing had high average annual incidence rates for both WAA and NOA (table 2). By focusing efforts on specific industries within services, manufacturing, and public administration with a high risk of work related asthma, a large proportion of both WAA and NOA could potentially be averted. Secondary prevention efforts could be directed towards increasing healthcare provider awareness of the potential for WAA, including educating healthcare providers about which worker populations are at greater risk.