Respiratory health of brickworkers in Cape Town, South Africa. Radiographic abnormalities.

A cross-sectional study utilizing internal controls based on dust exposure determinations was performed on 268 brickworkers. Smoking, age, and other information from a detailed respiratory questionnaire and results from a physical examination and pulmonary function tests were investigated in relation to radiographic abnormality. The prevalence of pneumoconiosis was near 4%. The roles of smoking, workplace dust exposure, and age as factors predicting radiographic abnormality are discussed.

The re is a noticeable absence in the liter ature of investigatio ns into the prevalence of respiratory abnormalit ies other than pneumoconiosis among brickworkers; this lack is sur prising given the high levels of du st expo sure recorded. Two studies ha ve attempted to investigate symptoms (3,9), while four have included lung function parameters (3,17 ,19) for brickworkers. None have investigated the interrelationships between radiographic abnormality, lung funct ions , symptoms , and du st expo sur e. The purpose of thi s st udy was to apply standardized techniques for determining du st levels and respiratory abnormality pre valences.
Th is study of five bri ckworks in the Cape Province of South Africa was accordingly designed to (i) ascertain the levels and types of dust exposures in different wor kplaces, (ii) determine the prevalence of respiratory abnormalities among brickworkers (abnormalities of interest comprised symptoms, signs, reductions in lung fun ction values, pneumoconiosis, and other radiographic changes), (iii) ascertain the interrelationships between the various respiratory abnor malities detected in the sur vey, (iv) investigate the relationships between respiratory abnormalities, dust exposure, smoking history, occupational history , and past medical histo ry with a view to establishing their relati ve contributions to the pre valen ce of respiratory abnormalities, and (v) investigate the adequacy of the Amer ican Conference of Go vernmental Indust rial Hygienists' threshold limit values (20) for different component s of brickworks dust for the prevention of respiratory ill effects. Workers and trade union and man agement personnel participated in the study .
Th is paper present s results from an ind ustrial hygiene surv ey in which the du st levels to which wo rkers were exposed in different parts of the brickworks were measured, including total dust (all airborne particles irrespective of size), respirable dust (particle diameter less than 5 to 7 urn), and the free-silica percentage composition of the respirable fraction. The measurements were then used to construct objective dust indicators which could be used to divide the study population into different exposure groups and assign exposures to individuals within the groups. The relationships between the subjective and objective indicators of dustiness, and the interrelationships between the different objective measures of dustiness, were examined with a view to assessing the validity of possible screening measures for detecting a silica exposure hazard in this industry.

Production process
General production characteristics for bricks and tiles are provided in the encyclopedia of the International Labour Office (9). Briefly described, the raw materials comprise different sorts of clay, shale, fly ash, and ground coal, and they are transported by a loader from the quarry or raw materials storage to the preparation department. They travel by conveyer belt via a box feeder, where checks are conducted for large metallic or other foreign bodies that might damage machinery in the preparation plant. Several rolling and grinding machines reduce the size of the material, and some water is added to dampen the mix, which passes in plastic form to extruding machines from which it emerges as a continuous brick shape and is cut by wire into individual bricks. Offsetters then sort the "green" bricks and place them onto trolleys, which are wheeled into drying tunnels and exposed to air heated by coal fires. Dryer operators load the wet bricks into the dryers, remove them when dry, and are responsible for cleaning the tunnels of dust and soot. In summer, bricks may be stacked outside by packers, where they are dried by wind and sun. Trammers transport the bricks in trolleys or barrows from the making section to the dryers and from the dryers to one of three types of kilns for firing. The old-fashioned Hoffmann kilns and the Downdraught kilns use finely ground coal for fuel. Bricks may be loaded either manually or by a hyster into these kilns. Setters stack the kilns manually with "green" bricks, while drawers manually unpack the baked bricks. The automatic oil-fired kiln constitutes a continuous process system with all the bricks packed on moving trolleys which proceed at a predetermined rate through a kiln tunnel in which they are first dried and then baked. No workers are involved in this process. Bricks are also layered with sand and coal into pyramids called clamps by clamp packers in the yards, where they bum for several weeks. Some baked and "green" bricks travel by conveyer to the rockface department, where machines render their surfaces uneven or remove their comers. Finished bricks are sorted and packed in the yard. There are various types of cleaning work done around the factory. The work process is mainly labor-intensive and has not changed 2 over the years in the various plants. Even the most modem plant with the automatic kiln has been in operation for 35 years.

Job ranking for dustiness
The survey team performed detailed inspections of all the workplaces in the five factories. Before the industrial hygiene survey, questionnaires had been administered to the workers to obtain their subjective estimates of dustiness for different workstations and information relating to their current and previous employment histories. On the basis of this information jobs were ranked by dustiness into three groups. As the factories differed slightly in the combination of tasks undertaken to produce bricks, but the jobs themselves were generally similar, it was decided to adopt a job-based approach in characterizing dust exposure rather than a factory-based approach. For instance, a worker in a preparation plant in one factory works under conditions similar to those of someone in a similar department in another factory. The same applies to most jobs except those for which the production technology is unique, as with the automatic kiln in one of the factories.

Sampling
The team planned the sampling strategy, and the job categories that were by consensus subjectively characterized as involving high, medium, and low dust exposure were selected for monitoring. The numbers of people in these job categories in each factory were supplied by the management. The selection of the sample subjects was made according to procedures described by the National Institute for Occupational Safety and Health (NIOSH) (21) and resulted in a sample size large enough to be able to detect with a confidence level of 95 070 at least one worker from the highest exposure subgroup (the top 20 % of the range) for the job category concerned. The numbers of measurements are presented in table I. For all the factories the numbers of workers to be monitored were chosen to cover all workers in a typical day shift. The differences in the sample proportions across factories relate to differing shift-work schedules. The resulting total for the respirable dust samples was 135. In addition, it was decided to determine simultaneously the free-silica content of at least two of the aforementioned samples from each worker category in table I, a total of 35 samples being obtained for free-silica content.

Measurements
The measurement procedures were based on methods published by NIOSH (21). The monitoring was jointly supervised by the team throughout the survey. Worker participation in the team meant that the likelihood was reduced that the conditions monitored would deviate too greatly from usual standards of dustiness, housekeeping, shift lengths, and job content. Measurements were felt to be maximally representative of the average work conditions. At each factory preliminary meetings of the survey team were held to select randomly the requisite number of workers for personal monitoring, to determine how long each measurement should take (up to 8 h), and whether repeat samples were required in the case of noncontinuous, irregular, batch, or task work in order to obtain average exposures for discontinuous work. Every effort was made to obtain full-shift samples for as close to 8-h periods as possible. The overwhelming majority of the measurements are representative of an 8-h workshift. In some cases work stopped early for the day, and the sampling time was less. In cases in which, after consultation with the worker monitored, we were unsatisfied that the period of measurement was in every way typical and representative of the work, the sample was discarded. As such, these shorter measurements were felt to be reasonably accurate time-weighted averages. Twenty-three constant flow Casella and du Pont (P 2000 and P 4000) pumps were used for the measurements, all of which were personal samples. The pumps were calibrated before and after the measurements, and periodic checks were made to see that they were working efficiently. Where technical difficulties (eg, substantially decreased flow rate) were encountered, measurements were discontinued. Abnormal work routines by workers or absences from production were also noted, and all these readings were discarded. Pumps were connected to SKC dust cyclones which deliver dust particles in the respirable range (less than 5 to 7 urn in diameter) to the filter. Glass fiber filters were used for gravimetric measurements unless free-silica determinations were also made, in which case cellulose nitrate filters were used. The latter were desiccated under standard conditions before and after the dust measurement. Filter weights were in milligrams and accurate to four decimal places. The scale was calibrated before each daily weighing session and then zeroed after every three readings. A mean value of five determinations was recorded as the weight of respirable dust. 182 Nonrespirable dust in the collecting chamber of the cyclone holders was removed and weighed. This weight was added to the weight of respirable dust trapped on the filter, and the sum was recorded as the value for total dust in milligrams. Free silica was determined as alpha-quartz by means of X-ray diffraction (8). Five replicate counts were done on each of the 35 samples, and the mean value was taken as the weight of free silica. Two samples were excluded from the analysis because of indeterminate job category, and four for unavailability of a respirable dust measurement for that sample.

Derivation of the exposure indicators
On the basis of the percentage concentration of free silica in the different workstations, the respirable dust threshold limit value-time-weighted average (TL V-TWA) (20) was calculated. The means and standard deviations for the total and respirable dust measurements in different job categories were then calculated, and the overall weighted means for the three levels of dustiness obtained. Jobs that were not measured were allocated on the basis of work process observations to one of the three dustiness groups. Job categories inserted in this way into the low-exposure group would be expected to be exposed to even lower levels of dust than those monitored. As far as the high-exposure group is concerned, maintenance workers are exposed to high levels of dust as they frequently but intermittently spend time mending machinery in the dusty sections, while cleaners are generally exposed to high dust levels partly generated by themselves when sweeping. The categories inserted into the mediumexposure group would be typically exposed at this level.
Where the average group means of the respirable dust levels exceeded the free-silica TL V, a silica hazard was presumed to exist, while dust levelsof 5 mg/rn' for respirable (22) and 10 mg/rn" for total (23) dust were taken to constitute general dust hazards. The whole number nearest the weighted group mean was used as the dust score for the subjects in this group.
These scores were then used to generate obj ective measures of individua l du st exposure.

Prior dustiness assessment
Generally speaking, the conditions were rather du sty with much pollution fro m coal fires in the kilns and from burning clamp s. Housekeeping was poor with dust accumulations on most surfaces. There was no evidence of any dust extraction mechanism or exhaust ventilation of either a local or general nature in any of the fact ories, and the use of personal protective equipment was not evident. There app ear to be seasonal fluctuations in general du stine ss. In the dry season the workers were generall y exposed to a continuou s cloud of du st generated by continuous tru ck and trolle y movement along roadways that are poorly maintained and not kept damp. Such exposure particularly affects outside wor kers at the clamp s and workers in the preparation department, where fro ntend loaders generate much du st by their movements. Inside the factory poor hou sekeeping led to the generation of dust even in the cleaner departm ents when settled du st was disturbed or swept. Dur ing the survey, which took place in the dry season, housek eeping, especially in the making section, was reportedly better than avera ge, while outside roadways were kept damper than usual.  2 shows the grouping system deri ved by the team for job categori es in the low, medium, and high dust exposure groups.

Raw materials
The raw material consisted of 59 to 79 % free or bound silica and of 12 to 25 % aluminum trioxide . Talc was onl y present in trace quantities.

Mean values and threshold limit values by job category
The results of the free-silica determinations and respir able dust levels for 29 samples are presented in table 3. The TL V-TWA values for respirable dust contain ing free silica are shown. Table 4 shows the mean dust values and the presence or absence of dust and silica hazards for different factories and workstations. Where the mean respirable dust level exceeded the TLV-TWA in table 3, a silica hazard (20) was deemed to exist. A general du st hazard was considered to exist in areas where the to tal dust concentration exceeded IO mg/rn" (23).  means to the nearest whole number. The weighted mean percentage of free silica for the medium (3.9070) and high (3.2 %) exposure groups together (where a silica hazard existed) was 3.6 %. In these two categories the average brickworker was exposed to a weighted mean respirable dust levelof 3.82 mg/rn", which is well above the TLV of 1.79 mg/m'. The weighted mean total dust for these workers was 25.5 mg/m'. The weighted means for the study group as a whole were 2.22 and 15.16 mg/m' for the respirable and total dust, respectively, with a 2.1 % free-silica content for the respirable fraction. The average brickworker was thus exposed to respirable dust at a level just less than the TLV of 2.44 mg/m', while the total dust exposure exceeded the "nuisance" value of 10 mg/m'.

Dust exposure indicators
The values in table 5 were used to construct objective dust indicators for investigating the dose-response relationships with biological variables reported elsewhere (24)(25)(26).
Continuous variables. The group mean values for the respirable and total dust concentrations in milligrams per cubic meter derived in table 5 for each job category constitute the values of the variables respirable dust, respirable dust 1, respirable dust 2, and total dust, total dust 1, total dust 2, respectively, and represent typical dust levels (both respirable dust and total dust) in the current job and in the two previous jobs, respectively, in brickworks. The product of the variables respirable dust or total dust and length of service (in years) in the current job and in the two previous jobs yield the.variables respirable dust-years or total dustyears, respirable dust I-years or total dust l-years, and respirable dust 2-years or total dust 2-years, and these variables represent respirable and total dust concentration-years in milligram-years per cubic meter for the three job levels. If these exposure times are summed over all the jobs in brickworks for respirable and total dust, two cumulative indicators of dust exposure time are obtained, ie, cumulative respirable dust exposure and cumulative total dust exposure.
Categorical variables. Respirable dust exposure was also used as a categorical variable. The total population was divided into tertiles on the basis of its cumulative frequency distribution. Objective dustiness, objective dustiness 1, and objective dustiness 2 correspond to the objective dustiness in the current and previous two jobs in brickworks, respectively, and have three levels corresponding to low, medium, and high exposure. They represent qualitative objective esti-mates of current and past job dustiness. The questionnaire responses yielded three subjective indicators of dust exposure, ie, subjective dustiness, subjective dustiness 1, and subjective dustiness 2, representing current subjective dust exposure, and subjective dust exposure in the previous two jobs in brickworks, respectively. These subjective dustiness indicators have three levels corresponding to low, high, and medium dustiness in the current and past two jobs held in brickworks v

Discussion
Problems with reliable workplace information due to misinformation, sudden changes in production tempo, breakdown of technical equipment, constraints on access to workers , and other problems that were difficult to anticipate and rule out in advance gave rise to some discrepancies in the number of samples for different dust types. Misclassification from the assigning of job categories to wrong dustiness groupings may have resulted in an unde restimation of the mean differences between the groups, and therefore in an underestimation of the dose-response relationships for respiratory health outcomes.
The shortage and high cost of monitoring equipment and laboratory services in South Africa limited the number and type of samples taken. It was not possible to perform triplicate measurements for respirable dust , total dust, and free silica for each subject. The free-silica and respirable-dust measurements were combined only at the cost of making determinations more laborious by introducing the need for elaborate filter preparation. A compromise was made when the total dust concentration was determined in that the nonrespirable fraction was collected from the cyclone holder. Although some dust may have adhered to the walls of the container after being emptied, there is no reason why this procedure should significantly underestimate the weight of dust any more than dust adhering to the inner walls of the filter holder underestimates the respirable dust concentration. Given the fact that a standard procedure was followed throughout, these determinat ions could, at worst, be slight systematic underestimates . The inability to test for trid ymite and cristobalite may have resulted in an underestimation of the silica hazard.
In cases in which the total dust level exceeded 10 mg/rn ", there was almost always a silica hazard as well (table 4). With regard to this particular industry, therefore, a strong case could be made for using total dust estimations as a screening procedure in future monitoring programs. These tests would be the cheapest option, as the use of cyclones would not be necessary and more time-consuming and multiple gravimetric measurements would be circumvented. The ability to dispense with X-ray diffraction analysis would result in substantial savings in time and cost. This and other simple screening devices for free silica based on the respirable dust level (27) may have particular preventive application in countries with less-developed resources, if an initial validation of the test has been performed for the situation to be screened on a routine basis.
The overall levelsof dustiness measured in this study were not exceptional. They were lower than previous levels recorded in similar brickworks in South Africa.' What is interesting however is that the ratio s (1:2:7) between the levels for the different dustiness groups in the factories were all similar. The mean respirable dust levels in brickworks in Great Britain (I) and Canada (2) were similar to those found in this study . The percentages of free silica were probably similar in Great Britain, but appeared to be higher in Canada and the United States (4, 7). They may however be similar in that problems with the analytical method resulted in an almost certain upward bias for both the Canadian and American results. They are very similar to percentages in a Polish red brick study (10). These results, taken together with the lack of preventive action at the workplace, indicate the existence of both general and silica dust hazards.
works' management and health personnel, and Dr D Christiani of the Occupational Health Program, Harvard School of Public Health , for their generous assistance.