Evaluation of autopsy data for occupational lung disease surveillance: this is the place where death delights to help the living
Background There is little systematic occupational health surveillance in South Africa, and past attempts have been unsuccessful. Routine administrative databases, such as compensation databases, are attractive for occupational disease surveillance. South African law allows for examination of the cardio-respiratory organs of current and former miners for occupational lung disease (OLD) compensation at the National Institute for Occupational Health and the findings are captured into the Pathology Automation (PATHAUT) database. The data are generated primarily for compensation but have also been used extensively for research such that the PATHAUT database is recognised as a unique resource for OLD research worldwide. Aim This thesis evaluated the attributes and usefulness of the administrative, PATHAUT autopsy database for its use as a tool for OLD surveillance in the South African mining industry. Methods Data were analysed by population group because of differences in employment and access to autopsy services. Paper 1 described demographic and occupational information, and OLDs in black and white miners from 1975 to 2013. Paper 2, used autopsy diagnoses of asbestos-related diseases (ARDs) as the gold standard to determine sensitivity, specificity and predictive values for clinical diagnoses of Asbestos Relief Trust claimants who had autopsies at the NIOH in 2010 and 2011. Factors associated with false-negative diagnoses of asbestosis were investigated using binary logistic regression modelling. Paper 3 investigated pulmonary tuberculosis (PTB) trends (1975-2014) in black and white miners. The analyses were segmented to represent the pre-HIV, rapid HIV spread and ART periods. Factors associated with PTB (year of autopsy, age, gold employment duration, silicosis and HIV) were determined using binary logistic regression modelling. The fourth paper (Paper 4) compared silicosis and PTB proportions female and male gold miners who had autopsies from 2005 (when sex was introduced in the PATHAUT database) to 2015 and who had started gold mining after the introduction of an intervention to increase the participation of women in underground mining in 2002. The evaluation used findings from the four epidemiological studies and was guided by the CDC Guidelines for Evaluating Public Health Surveillance Systems. Results Paper 1 analysed 109,101 records in the PATHAUT database (1975-2013): 72,348 (66.3%) of black and 34,794 (31.9%) of white miners. The annual numbers of autopsies declined from above 3,000 from 1975 to 1998, to 1,118 in 2013. The mean ages of black and white miners were 39.7±11.4 and 60.9±14.6 years, respectively. Fields for sex, employment status and place of origin of the organs were added to the database in 2005, 2007 and 2010, respectively. There were 320 (0.3%) women. At the time of death, 78.6% (n=5,077) black and 13.2% (n=428) white miners were in employment. Most organs were submitted from the North West (n=1,774; 34.8%), Free State (n=1,150; 22.6%) and Gauteng (n=1,139; 22.3%) provinces. Many miners had worked in more than one mining commodity (n=14,444; 13.2%) and, of the 34 commodities worked, most were gold (n=70,296; 72.7%), platinum (n=9,586; 8.8%), coal (n=7,741; 7.1%) or asbestos (n=2,560; 2.3%) miners. Over the study period, the major diseases diagnosed at autopsy were emphysema (15.5% in black and 49.5% in white miners), silicosis (12.0% and 20.5%) and PTB (13% and 2.4%). The proportions of silicosis and PTB increased over time. One hundred and forty nine ARD cases were identified (Paper 2). More ARDs were diagnosed at autopsy (77 asbestosis, 27 mesothelioma and 22 lung cancer) than clinically (52 asbestosis, 14 mesothelioma and 3 lung cancer). The sensitivity values for the clinical diagnoses for asbestosis, mesothelioma and lung cancer were 50.6%, 40.7% and 13.6%, and the specificity values were 81.9%, 97.5% and 100.0%, respectively. False-negative diagnoses of asbestosis were more likely if chest radiographs were of acceptable compared to good quality, and in cases with autopsy-diagnosed PTB. The proportions of male miners with PTB increased from 4.62% in 1975 to 43.1% in 2007 and decreased to 27.18% in 2014. For white miners, PTB increased from 2.07% in 1975 to 9.51% in 2007 and declined to 5.19% in 2014. The adjusted odds ratios (AORs) for PTB differed in magnitude and significance by population group and calendar period. PTB was positively associated with silicosis in both population groups and in all calendar periods. The AORs for year of autopsy (a surrogate for unmeasured confounding, e.g. unknown HIV status and tuberculosis transmission) increased over time. PTB was also associated with increasing gold employment duration (a proxy for cumulative dust exposure) and HIV. Paper 4 included 847 gold miners: 68 (8.0%) women and 779 (92.0%) men. There were no statistical sex differences in autopsy-diagnosed pulmonary silicosis [n=3 (4.4%) in women and n=54 (6.9%) in men; p=0.61], lymph gland silicosis [n=11 (16.2%) and n=171 (22.0%); p=0.27] or PTB [n=29 (42.6%) and n=254 (32.6%); p=0.09]. All three women with silicosis had slight disease. Two had worked for 3.3 and 3.8 years and were 26 and 35 years old, respectively. The third woman had 10.9 years of employment and was 47 years old. Information in the four papers was used to evaluate seven system attributes (simplicity, flexibility, data quality, representativeness, sensitivity, stability and timelines) of the autopsy process and data. Acceptability and some aspects of other attributes were not assessed because stakeholder engagement was not included in this thesis. Predictive values were not assessed due to the lack of suitable external data. The usefulness of the autopsy data for use in surveillance was assessed by its abilities to detect OLDs, provide disease estimates, detect trends, determine factors associated with disease occurrence, identify sentinel cases and assess the effectiveness of prevention and control interventions. Conclusion The high levels of OLDs in miners are indicative of ineffective silica dust control programmes. The autopsy data provide the only source of long-term information to monitor OLDs and the effectiveness of prevention and control interventions in the South African mining industry. The evaluation findings indicate that the autopsy processes and data cover the core elements of surveillance, that is, ongoing and systematic reporting, data collection, analysis, interpretation and dissemination of information. Hence, the use of the PATHAUT database could be extended to include a surveillance function in addition to its use for compensation and research. The development of a surveillance programme using these data is relevant because of the scarcity of successful OLD surveillance systems globally.
A thesis submitted to the Faculty of Health Sciences, University of the Witwatersrand, in fulfilment of the requirements for the degree of Doctor of Philosophy Johannesburg, 2019
Ndlovu, Ntombizodwa (2019) Evaluation of autopsy data for occupational lung disease surveillance, University of the Witwatersrand, Johannesburg, <http://hdl.handle.net/10539/29701>