Chest physician-reported, work-related, long-latency respiratory disease in Great Britain

Discussion

This study provides an overview of chest physician-diagnosed LLRDs in GB over a 19-year period. NMPD and mesothelioma predominated (82% of cases), 95% of cases were in men, and asbestos was specified for 92% of the cases. Incidence typically increased until approximately 70–85 years and declined thereafter. The highest IRRs were observed for miners and quarrymen (COPD), plumbers and gas fitters (asbestosis), and shipyard and dock workers (all other categories). With a few notable exceptions, there was a clear correlation between the proportion of SWORD mesothelioma cases and PMRs by occupation.

LLRD incidence rates were typically higher than in previous SWORD reports [17, 18]. For example, mesothelioma rates in men were approximately five per 100 000 per year in the present study, compared to three (SWORD data 1996–1999) and two (1989–1991). In the present (but not earlier) study, the numerator was adjusted for cases not captured due to non-participation and non-response [12]. This approach might overestimate rates as it assumes that physicians not participating/responding would report, on average, at the same frequency as those that are. However, even with these adjustments, the true LLRD incidence in GB is probably underestimated. The SWORD “case capture” rate has likely declined since the estimate of 60% (2005–2007), because the number of GB chest physicians has increased, and SWORD reporter numbers have fallen slightly [12]. However, since much of this increase in physician numbers has been in the sub-speciality of respiratory cancer, the SWORD “case-capture” rate is likely higher for diagnoses such as asthma compared to diagnoses such as lung cancer, mesothelioma and pleural plaques. Difficulties in attributing a case to work, particularly for diagnoses such as COPD and lung cancer, may also result in under-reporting to SWORD. Estimates based on the attributable fraction suggest a much higher GB incidence of occupational lung cancer (5442 cases in 2004 compared to an estimated 131 reported to SWORD) [19].

Unsurprisingly, given the differences in occupational exposure, a higher incidence of LLRD was observed for men than for women. Incidence rates also generally increased with age, peaking at around 75 years and declining thereafter. A similar pattern is reflected in other data sources [20, 21]. The decrease in incidence in extreme old age is likely explained by the “healthy survivor effect”, where patients with LLRD die before reaching this point.

Temporal trends in LLRD incidence are not included here but are investigated annually and have been published previously [22, 23]. Contrary to other data sources [24, 25], SWORD data generally suggest LLRD incidence to be stable or declining, the exception being pneumoconiosis, for which an increase in incidence was observed. However, changes in clinical practice over time are likely to have substantially influenced the observed trends; therefore, the findings need to be interpreted carefully.

The high LLRD incidence for shipyard and dock workers (compared to all other occupations) was expected because, historically, this occupational group was amongst the most frequently exposed to asbestos [26]. The comparatively low PMR (based on mesothelioma deaths) for this group may be underestimated. The mortality register records the last known occupation, and because (GB) shipbuilding had largely ceased by the 1980s, many exposed in this industry may have taken up other work after dockyard closure (e.g. in construction). However, another occupational group closely associated with shipbuilding – metal plate workers, shipwrights and riveters – had one of the highest PMRs. In general, movement from the occupation in which asbestos exposure took place (typically earlier in working life) into other jobs prior to retirement will tend to dilute the true differences between mesothelioma PMRs, but this will particularly affect those groups associated with industries that have declined or disappeared altogether. Another example of this effect may be textile process operatives. Conversely, SWORD – where reporters typically record the main occupation thought to be the source of exposure – could potentially overstate the true risk in certain cases, as exemplified by the relatively high proportion of SWORD-reported cases among motor mechanics and auto engineers in contrast with the particularly low PMR for this group.

Discrepancies between PMRs and proportions by occupations in SWORD could also arise from unclear/limited information provided to either data source. Within SWORD, such ambiguities tend to result in the oversubscribing of cases to more “general” occupational groups. Some of the information reported to THOR may also be erroneous. For example, 241 mesothelioma/NMPD cases in SWORD were not attributed to asbestos. This is likely misattribution – a limitation of establishing causation in LLRD, as, by definition, the exposure had occurred decades ago. Similarly, because the criteria for reporting to THOR are fairly inclusive, some of the reported associations with other agents have arisen from co-exposures and misclassification of the true cause.

Given the common exposure (asbestos) it is unsurprising that occupations with comparatively high IRRs (and their ranking) were similar for mesothelioma, NMPD, asbestosis and lung cancer. Of interest, plumbers and related occupations had the highest asbestosis IRR, with a further seven occupational groups also having comparatively high IRRs, suggesting high levels of exposure for these groups (the general consensus is that heavy asbestos exposures are required to produce clinically significant asbestosis within the lifetime of an individual) [27]. Lung cancer was the only LLRD for which a statistically significantly higher IRR for the group “other engineering and allied trades, foundry workers” was observed. However, information about other risk factors, most notably tobacco smoking, may potentially affect the attribution of individual lung cancer cases.

Pneumoconiosis attributed to coal and to silica was also investigated, with miners and quarrymen having a very high IRR for the former. This occupational group declined in number during the study period, so adjusting for this decline would likely increase the IRR further. Miners and quarrymen also had a relatively high IRR for silicosis, in addition to glass and ceramics makers and construction workers. An estimated 3.2 million EU (500 000 UK) workers are still routinely exposed to silica [28]. This estimate is likely to increase with the potential for workers to be exposed to silica in new industries such as stone or artificial stone composite countertop fabrication and hydraulic fracturing [29, 30].

COPD rates were significantly higher for miners and quarrymen than those in other occupations. Approximately 94% of cases were attributed to coal dust, an established cause of COPD that is compensable under IIDB [31]. This study, and others, also revealed an increased risk of COPD for welders [32]. Because welding trades account for approximately 79 000 workers in the UK (according to the 2014 LFS), this could imply that a significant proportion of the population are at risk.

Of the SWORD LLRD cases, 5% were in women. Occupations with relatively high IRRs tended to be in manufacturing, most notably textile workers (mainly asbestos manufacture). Interestingly, the suggested increased risk of LLRD for teachers [33] was not borne out by SWORD data, with only four cases in women reported (and two in men).

The denominator for the rates in table 1 and figure 1 was the total (i.e. working + non-working) adult British population (1996–2014). This approach may underestimate incidence among those originally exposed. However, attempting to calculate incidence among the smaller, originally exposed, historic populations (using lagged working population as the denominator) risks overestimating the originally exposed subgroup of the working population by an unknown amount. It also introduces other uncertainties by ignoring survival and requiring an assumption about an appropriate lag period. When comparing relative incidence by occupation, however, lagged denominators were deemed necessary to allow for the possibility that the sizes of some occupational groups had changed substantially since the exposure period. Hence, the denominator was the number employed (by occupation) according to the 1971 census. However, the use of data from a single year (historical workforce data is limited and not necessarily comparable over time) may still have led to under- or overestimation of IRRs for occupations for which the exposed population changed significantly during the estimated exposure period. This approach also makes assumptions about the lag period between exposure and disease onset, i.e. restricting it to 26–44 years (depending on the year the case was reported) and, for any 1 year, the same for all diagnoses. This is a relatively crude approach, and a limitation of this study. There may be considerable variation in this lag assumption, both within and between diseases [34].

In conclusion, this study provides a useful overview of the status of LLRD in GB over the past two decades and shows the continuing incidence of LLRD arising from historical exposure to agents such as asbestos, silica and coal. These findings are informative for other countries that have had similar exposure patterns. For example, the Netherlands had a historically early and high average asbestos use per capita, with shipbuilding a significant source of exposure, followed by the early introduction of control measures [35]. Furthermore, the identification of occupations at increased risk from agents still encountered or in active use, such as welders exposed to welding fumes or construction workers exposed to silica, provides useful information for the targeting of preventative activities.