Of 3,680 subjects identified initially from the data bases, subjects were excluded for the following reasons (in this order): missing year of birth (6), born after 1977 (13), missing year of accident (1), year of accident before 1980 (1), age at time of “accident” greater than 65 yr (32, most of whom were AP), or age less than 17 yr at time of accident (21). The remaining 3,605 subjects included 1,556 injury claimants, 402 adult asthma patients, and 1,647 work-related asthma claimants. Of this latter group, 107 WCB files could not be located for abstraction and, on abstraction 428 files were not found to have sufficient information for the research assistant to adequately complete the standardized format, leaving 1,112 work-related asthma claimants. Of these, 844 with OA were included in this morbidity study; the remainder had AA and were not included.

The characteristics of the subjects, by subgroup, are shown in Table 1. Smoking data were available for 810 (96%) of OA and 392 (98%) of AP but not for INJ; there were few smokers in the AP group. The commonest causative agents among the OA claimants (known for 810 [96%]) were isocyanates (in 52%), flour (6%), metals (4%), cedar (3%), latex (3%), grains, plastics, noncedar wood dusts, and welding fumes (2% each), paint, chlorine, and solvents (1% each), unspecified dusts (2%), unspecified fumes (4%), and other agents (15%).

The number of subjects who were ever hospitalized over the study period, according to subgroup and broad diagnostic categories is shown in Table 2; overall approximately one-third of the subjects had been hospitalized, most frequently among AP, intermediate among OA, and least frequently among the injury claimants.

The injury claimants were hospitalized significantly less frequently than the general population for all causes combined (SMR 0.73), malignant neoplasms, and circulatory diseases (Table 3). On the other hand, the number of hospitalizations for all causes combined among both the OA (SMR 1.45) and AP (SMR 1.66) groups significantly exceeded that expected in the general population (Table 3).

Among INJ, hospitalizations for respiratory diseases were close to expectation. The increase in hospitalizations among OA and AP groups was due largely to admissions for respiratory diseases that were increased many times more than expected (SMRs 9.2 and 17.7, respectively), due in turn to even greater excess admissions for all COPD and allied conditions, and particularly for asthma (45-fold and 81-fold greater, respectively; Table 3).

OA compared with injury claimants. The results from both the stratified analyses and proportional hazards regression models are presented in Table 4. The results for the two types of analyses are similar, but the magnitude of the risk estimates was lower with the latter approach. As compared with the injury claimants, after adjustment for period of birth, sex, and time of accident, the OA claimants were significantly more likely to be hospitalized for all causes combined, due largely to an increased RR of hospitalization for respiratory diseases (5-fold higher) and for asthma (28-fold higher), both highly statistically significant. OA claimants were also 40% and 20% more likely to be admitted for circulatory disease and IHD, respectively, in the adjusted proportional hazards models, although this was not statistically significant. There was no difference between the groups in the risk of hospitalization for malignant neoplasms or external causes.

OA compared with asthma patients. The results for the stratified and proportional hazards analyses were very similar (Table 5). As compared with the tertiary referral AP, the OA claimants were 30% less likely to be hospitalized overall, owing to significantly lower risks for hospitalization for all respiratory diseases and asthma. The stratified analyses and unadjusted RRs from the proportional hazards analyses provided estimated RRs for these outcomes of approximately 0.4 to 0.5. However, both sets of analyses indicated significant interaction of “exposure group” with smoking (for respiratory disease), and with smoking and sex (for asthma). The stratum-specific RRs indicated that among nonsmokers or ex-smokers, after adjustment for period of birth, period of accident, and sex, OA claimants were approximately 50% as likely as AP to be admitted for respiratory diseases, whereas current smokers were only 10% as likely. For asthma, after adjustment, among males, OA claimants were similarly approximately 40% to 50% as likely as AP to be admitted across all smoking groups, although the 95% confidence interval (CI) included unity among current smokers. Among females, OA claimants were 20% to 30% as likely to be admitted as compared with AP if nonsmokers and ex-smokers, and very unlikely if current smokers (RR 0.04). The CIs for these comparisons excluded unity.

There was no difference between these groups in the risk for hospitalization for external causes or circulatory diseases (Table 5). In contrast to the observations for all causes combined and for respiratory diseases, the OA claimants were 30% more likely than the asthma patients to be hospitalized for IHD, but the CI included unity.

Factors associated with asthma hospitalization among the OA group. Compared with those OA claimants who were not hospitalized, those who were hospitalized for asthma were significantly older, less likely to be current smokers, slightly more likely to be male, but had been followed for a similar length of time since their claims (Table 6). Moreover, those who were hospitalized for asthma were only about half as likely to have been exposed to isocyanates as the cause of OA compared with other agents (p = 0.003); this association was little changed when adjusted for age, sex, calendar period of claim, or smoking status (Table 6).

OA versus injury claimants. The OA claimants remained at greater risk of hospitalization for asthma than the injury claimants throughout the follow-up, but the magnitude of the OR declined sharply at 5 or more years after the accident date, to one-fifth of that during the first 2 yr and to one-third of that during the middle time period (Table 7). These were based on very few admissions among the injury claimants.

OA claimants versus asthma patients. We confined these analyses to male nonsmokers and ex-smokers because there were too few female smokers among AP for analysis. The OR adjusted for year of birth, period of accident, and smoking status was lower 5 or more years after the accident (0.29; 95% CI 0.10 to 0.85) than during the first 2 yr (0.59; 95% CI 0.25 to 1.41) or during 2 to 5 yr (0.48; 95% CI 0.17 to 1.35), although the CIs overlapped. This suggested that OA claimants were relatively more likely to be admitted during the earlier period after the accident date. The CI excluded unity only in the last time period. The MH-ORs and logic estimates were almost identical (data not shown). Further adjustment of these models for admission during the first 2-yr interval changed the ORs only slightly (data not shown).

These analyses indicated that the comparison group of injury claimants had a pattern of hospitalizations similar to what might be expected in a working group, with fewer than expected admissions for all causes, circulatory and respiratory diseases, consistent with the healthy worker effect (18-20). On the other hand, the OA claimants, although drawn from a similar working population, did not display such a pattern, but instead mirrored closely that of the asthma patients, with more admissions than expected in the general population. Most striking were the markedly increased SMRs in these two asthma groups for respiratory diseases (9- and 17-fold, respectively), and for asthma (45- and 81-fold, respectively). These excesses are comparable in magnitude to the increased mortality from asthma (observed/expected of 66 in males, 50 in females) among patients with asthma compared with that expected in the population of England and Wales reported by Alderson and Loy (21), and RR among asthmatics compared with nonasthmatics for all causes of 1.6, respiratory diseases of 11.1, COPD of 14.8, and asthma of infinity, reported by Markowe and colleagues (22). More recently, Lange and coworkers (23) similarly found higher risks of death for all causes (1.7 in women, 1.5 in men), respiratory diseases (7.5-fold), and asthma (52-fold) among adults with self-reported asthma compared with nonasthmatics in Denmark, and Huovinen and colleagues (24) reported increased mortality among adult asthmatics in Finland compared with nonasthmatic persons, with adjusted hazard ratios for all causes of approximately 1.5 and for COPD of 9 in men and 20 in women.

As compared with the general population, we observed more admissions for asthma than expected among the Injury claimants (SMR 1.87), despite admissions for all respiratory diseases being close to expectation (SMR 0.96; Table 3). This increase was limited to males (observed/expected, 18/4.4). As explained in the Methods section, the SMRs are calculated using all observed admissions (i.e., including readmissions). In the direct comparisons, there were only 4 individuals (3 males, 1 female) in the Injury group with asthma admissions (see Table 7), suggesting multiple readmissions for asthma among a few males in the Injury claimants group as an explanation for this finding.

OA versus injury claimants. These findings paralleled the comparisons to the general population. The OA group was significantly more likely to be hospitalized, due largely to very much greater risks for admission for all respiratory diseases and asthma (approximately 5- and 28-fold greater, respectively in the adjusted proportional hazards analyses; Table 4). The small nonsignificant increase in risk for admissions from circulatory disease of 40% and from IHD of 20% is consistent with the approximately doubling of deaths from IHD observed in our mortality study (16). This observation makes biologic sense, given that uncontrolled asthma may lead to episodic hypoxia and pulmonary hypertension or arrhythmias, which may compromise myocardial function in those with underlying IHD. We are not aware that increased IHD morbidity or mortality associated with OA has been reported previously, and this needs to be confirmed.

The subjects with OA were no more likely than those with injuries to be admitted as a result of malignancies or external causes, outcomes that are not related to asthma, which strengthens the internal validity of our findings. Thus, we have documented that subjects with OA are hospitalized at rates far greater than expected in the general population and many times greater in adjusted direct comparison to a working group with INJ who were likely similar in socioeconomic status to the OA claimants, and who passed through the same filter of the compensation system. The magnitude of these risks is both statistically and clinically significant. The injury claimants represent the more appropriate comparison group in our study.

OA claimants versus nonwork-related asthmatic patients. The OA claimants were less likely than the AP seen in a tertiary hospital outpatient setting to be admitted during follow-up, displaying approximately half the risk of hospitalization for respiratory diseases and asthma. The RR estimates were modified by smoking and sex, with OA claimants who were female current smokers being much less likely to be admitted. We do not have an explanation for this, but these analyses were limited by the very small number of female smokers in the AP group. Given that the asthma patients were seen at a tertiary referral center, they may not have been the most appropriate comparison group with respect to asthma severity. Patients from community asthma practices may have been more suitable, although these patients, including very mild asthmatics, may be less likely to have had routine objective confirmation of the diagnosis of asthma by demonstration of a bronchodilator pulmonary function response or methacholine/ histamine challenge response. It should also be emphasized that the inclusion of the asthma patients as a referent group may control for changes over time with respect to policies and practices for admissions for asthma, given hospital restructuring and reduction in inpatients beds that were underway in the province during this time period.

The groups were equally likely to be hospitalized for external causes that are unrelated to asthma. Despite the lower risks of hospitalization for all causes combined and respiratory diseases (and the presence of the healthy worker effect), the OA claimants were as likely to be admitted for circulatory diseases and 30% more likely for IHD. This also mirrors the findings in our previous report on the mortality in these groups (16) in which there were nine deaths due to IHD among the work-related asthma claimants and none among the AP group. These increases in morbidity and mortality from IHD among occupational asthmatics versus nonoccupational asthmatics, although not statistically significant, would support the hypothesis that an industrial working population, which would be expected to have exposure to airborne fine and ultrafine respirable particles, may be at increased risk of IHD. It has been suggested that such an effect may relate to an inflammatory response in the lung (25), and might result from an increase in interleukin-6 (IL-6) and subsequently fibrinogen (26). Epidemiologic investigations have demonstrated an association between fine and ultrafine particles in outdoor air and cardiorespiratory morbidity and mortality, although the mechanism of this association is unclear, as recently reviewed by Brunekreef (27).

Factors associated with asthma hospitalization among the OA group. We found that among the OA claimants, those hospitalized for asthma were significantly less likely to have been exposed to isocyanates (versus other agents) than those not hospitalized. Adjustment for potentially confounding factors did not affect this association. This is consistent with our earlier findings that compared with those with OA due to other causes, those with isocyanate-induced OA had an earlier diagnosis (significantly shorter duration of symptoms before diagnosis) and a tendency to better outcomes at follow-up assessment (15). This may be attributable to the medical surveillance program for workers exposed to iscocyanates that has been in effect in Ontario since the 1980s.

We found some evidence that OA claimants became less likely to be admitted for asthma with increasing time after the accident, relative to both INJ and AP. This trend was consistent with out hypothesis that the reason underlying the hospitalizations among those with OA was related to the “accident,” that is, to the work-related induction of sensitization, and that the effect might decline after diagnosis and probable removal from exposure (although we do not have data on time of removal). The 5-yr period for this effect to appear is consistent with the findings at follow-up of subjects with OA reported by Perfetti and coworkers (28), who observed that the proportion of subjects having normal provocative concentration of methacholine causing a 20% fall in FEV₁ (PC₂₀) at follow-up was significantly higher among those removed from exposure for > 5 yr than among those removed for ⩽ 5 yr.

There are several possible explanations as to why the decline with time was not more dramatic. First, we started follow-up from the date of the accident, which generally preceded the documentation and allowance of the claims. Thus, in many cases workers continued to work and to be exposed after filing claims, and perhaps become clinically worse for some time after the “accident date,” which would dilute the trends for hospitalization. Second, it is well known that the symptoms and bronchial hyperresponsiveness may persist for long periods, even after removal from exposure (11). Third, there may have been misclassification of the coding of the cause of hospitalization, for example, between asthma versus other COPD, as was observed in our mortality study (16). Whereas this may apply to both OA claimants and AP, it is possible that for subjects known to have OA, subsequent hospitalizations (for respiratory disease in general) may be more likely to be coded by their physicians as asthma for compensation purposes. If so, this would be expected to diminish the evidence of a trend over time.

The limitations of our study include the use of tertiary care asthma patients, which may not represent the most appropriate comparison group, as previously discussed. The number of asthma admissions in the injury group was small and thus limited the power of the comparisons. Because we initially included all claims for OA in the province during the study period, selection bias is unlikely. By reviewing the files of the OA claimants, we attempted to include only those with somewhat better evidence of actually having had OA. The classification into the three “exposure” groups was performed independently of our study, before follow-up began, and was not influenced by the subsequent outcome of admission, or the cause of admission.

We adjusted in the analysis for important confounding factors including period of birth, sex, and period of accident. In addition, we obtained smoking data for the OA and AP but not the INJ groups. The data for the OA group indicated that the proportion of current smokers was 26%. In Canada in 1986, the midpoint of our cohort recruitment period, the proportion of current smokers was 39% and 36% among males age 25 to 44 and 45 to 64, respectively, and 36% and 29% among females (29). Furthermore, there was no difference in risk of hospitalization for all cancers (RR 1.0; Table 4); and for lung cancer in particular, the SMR among the OA group was 0.19 (observed 2; expected 10.3), whereas among the INJ group, the SMR was somewhat greater in magnitude, 0.45 (observed 5; expected 11.2). This suggests that smoking is very unlikely to be an important confounder in comparisons to the injury claimants.

The strengths of our investigation include the relatively large number of OA claimants, the large number of hospital admissions (in the two asthma groups), the adjustment for potential confounders including smoking where available, and the inclusion of comparisons both to the general population and directly between groups.

In summary, this cohort morbidity study provides, for the first time to our knowledge, evidence that workers with OA suffer increased hospitalization rates, mainly from respiratory diseases and asthma, as compared with other workers, at rates that approach those seen among asthmatic patients referred to a tertiary care center. Our findings also suggest that those with OA may have small increases in morbidity from IHD and circulatory disease as compared with workers without OA, although similar to that in the general population. Given that OA is preventable, and that serious sequelae may develop, our findings underscore the need for greater attention directed toward primary prevention of this condition.