Table 1
Variables used in this study.
| VARIABLE ON DATABASE | COMMENT OR FURTHER DESCRIPTION | DERIVED VARIABLES (FOR EACH MINER) |
|---|---|---|
| Unique key | TEBA (PTY) Ltd randomly assigned industry number (not personal ID) | – |
| Country of origin | Country or office at which recruited or registered for a given contract | Country of origin* |
| Contract start date | Number of contractsStart date of first contractStart date of last contract | |
| Contract termination date | End date of first contractEnd date of last contract | |
| Contract duration (days) | Difference in days between contract termination date and contract start date plus 1 | Cumulative employment across all contracts in days or years Contract duration years where 1 year = 365.2425 days |
| Date of birth | Age at start of first contractAge at end of last contract |
[i] * Miners recruited from nominally independent “homelands” in the apartheid era are counted as South African. A small number of miners from Lesotho may have changed country of origin during this period.
Table 2
Contract characteristics of employees on the database of a large gold mining company, 1973–2018 (n = 300,774).
| VARIABLE | MEAN | SD | MEDIAN | IQR |
|---|---|---|---|---|
| Age at first contract (years) | 23.3 | 6.2 | 21.6 | 19.7, 25.0 |
| Age at the start of the last contract (years) | 40.3 | 11.7 | 40.4 | 30.4, 49.9 |
| Age at the end of the last contract (years) | 42.5 | 11.5 | 42.4 | 33.0, 51.8 |
| Calendar year of first contract | 1980 | 10.7 | 1981 | 1974, 1986 |
| Calendar year of last contract | 1997 | (9.4) | 1996 | 1989, 2003 |
| Cumulative employment at end of final contract (years) | 13.8 | (8.9) | 12.7 | 6.9, 19.4 |
[i] SD: standard deviation; IQR: interquartile range.
Figure 1
Years of cumulative employment* of employees of a major gold mining company, by final contract end date, 1973–2018**.
* Annual mean and median values calculated by calendar year of last observation.
** Includes service at other gold mining companies.
Table 3
Piecewise linear regression slope coefficients with four breakpoints*.
| DATE RANGE | NUMBER OF WORKERS | BREAKPOINT (95% CI) | SLOPE (YEARS/CALENDAR YEAR) (95% CI) |
|---|---|---|---|
| 1973-01-01 to 1985-09-16 | 15,673 | 1985-09-16 ± 84 days | 0.02 (–0.04, 0.07) |
| 1985-09-17 to 1997-05-11 | 130,522 | 1997-05-11 ± 75 days | 0.83 (0.77, 0.89) |
| 1997-05-12 to 2013-11-30 | 126,496 | 2013-11-30 ± 49 days | 0.34 (0.33, 0.36) |
| 2013-12-01 to 2014-11-06 | 8,208 | 2014-11-06 ± 21 days | 3.93 (3.26, 4.60) |
| 2014-11-07 to 2018-12-31 | 19,875 | – | –1.35 (–2.02, –0.67) |
[i] * Adjusted R2 of model = 0.283; CI: Confidence interval.
Figure 2
Proportion of annual exit cohort with ≥10, 15, or 20 years cumulative employment, by final contract end date, 1973–2018.
Figure 3
Years of cumulative employment by final contract end date: miners of South African origin versus those from neighbouring countries,* 1973–2018**.
* Botswana, Eswatini, Lesotho, Mozambique.
** Piecewise linear regression (thick lines) and annualised mean (thin lines), using the same breakpoints as in Figure 1.