The Western Australian Mesothelioma Registry: Analysis of 60 years of cases.

BACKGROUND AND OBJECTIVE: Australia introduced a partial ban on asbestos consumption in 1984. There is continuing concern about exposure to asbestos in the built environment and non-occupational exposures. The aim of this study was to describe epidemiological trends of mesothelioma in Western Australia (WA) over the 60 years since the first case was recorded. METHODS: Every case of mesothelioma notified to the WA Cancer Registry is reviewed by an expert panel. Data include demographic and clinical variables including principal mode of asbestos exposure and age at first exposure. Trends over time for survival, latency and pathological subtype of mesothelioma where analysed. Incidence rates for cases exposed during home renovation where calculated. RESULTS: Two thousand seven hundred ninety-six cases of mesothelioma were identified with males comprising the majority (n = 2368, 84.7%). The median (IQR) age at diagnosis was 70 (62-78) years, and median latency of 47 (38-55) years. Pleural mesothelioma was recorded in 2620 (93.7%) cases with the epithelioid subtype most prevalent (n = 1730, 61.9%). Overall, median survival was 298 (128-585) days and latency 46 (37-54) years, both effectively doubling over the study period. Non-occupational exposures were proportionally higher in females (52.6%), compared with males (9.5%). Home renovation was the primary exposure in 227 (8.1%) cases, with number of cases and incidence rate ratio peaking in 2005/09 but subsequently decreasing. CONCLUSION: The annual number of cases of mesothelioma in WA may have hit a plateau. The majority of females have non-occupational exposures and incidence rates from home renovation exposure may have peaked, suggesting the ban on asbestos has been effective.

Lung cancer screening an asbestos exposed population: Existing lung cancer risk criteria are not sufficient.

BACKGROUND AND OBJECTIVE: Asbestos is a major risk factor for lung cancer, with or without tobacco smoke exposure. Low dose computed tomography (LDCT) screening for early lung cancer is effective but only when targeting high risk populations. This study aimed to analyse the effectiveness of LDCT screening in an asbestos exposed population and to compare lung cancer screening program (LCSP) eligibility criteria. METHODS: Participants in an asbestos health surveillance program, the Western Australia Asbestos Review Program, underwent at least one LDCT scan and lung function assessment as part of annual review between 2012 and 2017. Lung cancer cases were confirmed through linkage to the WA cancer registry. Theoretical eligibility for different screening programs was calculated. RESULTS: Five thousand seven hundred and two LDCT scans were performed on 1743 individuals. The median age was 69.8 years, 1481 (85.0%) were male and 1147 (65.8%) were ever-smokers (median pack-year exposure of 20.0). Overall, 26 lung cancers were detected (1.5% of the population; 3.5 cases per 1000 person-years of observation). Lung cancer was early stage in 86.4% and four (15.4%) cases were never smokers. Based on current lung screening program criteria, 1299 (74.5%) of this population, including the majority (17, 65.4%) of lung cancer cases, would not have been eligible for any LCSP. CONCLUSION: This population is at raised risk despite modest tobacco exposure. LDCT screening is effective at identifying early-stage lung cancer in this population and existing lung cancer risk criteria do not capture this population adequately.

Incidence of NUT carcinoma in Western Australia from 1989 to 2014: a review of pediatric and adolescent cases from Perth Children's Hospital.

BACKGROUND: NUT carcinoma (NC), previously known as NUT midline carcinoma, is a rare and very aggressive cancer that occurs in both children and adults. NC is largely chemoresistant, with an overall survival of less than 7 months. Because the carcinoma is not restricted to a particular organ, diagnosis is often a challenge. In the absence of a clearly determined incidence for NC, we sought to study the diagnosis of patients in a well-defined population. METHODS: We systematically reviewed records of all patients that presented to the Oncology Department of the Princess Margaret Hospital for Children from 1989 to 2014. This institution in the geographically isolated state of Western Australia has a catchment population of around 2 million. We then identified all high grade undifferentiated sarcomas or carcinomas in the 0-16 year age group. RESULTS: Over 26 years, we found 14 patients of 16 years or younger with undifferentiated malignant tumors. Of these, five tumors were positive by immunohistochemistry for the NUT/NUTM1 (Nuclear Protein in Testis) protein and/or the translocation t(15;19). Three patients presented with thoracic tumors, one with a para-spinal tumor, and one had an upper airway nasopharyngeal carcinoma. In all five cases, there was an initial response to therapy and then progression. This 26-year survey was conducted in a geographically isolated state with a well-defined population, and we determined an estimated incidence of NC of around 0.41 per million child years (0-16 yrs. of age) at risk. From three patients it was feasible to derive cell lines for further genetic analyses and drug screening. CONCLUSIONS: For the first time, the incidence of NC could be determined in a well-defined geographic area. The calculated rate of NC incidence is consistent with a history of under-recognition for this malignancy. These findings indicate that improved diagnostic detection of NC would enable better management and counselling of patients. Our findings emphasize the heterogeneity of NC, and they highlight the need to develop personalised therapy options, and to consider a diagnosis of NC in undifferentiated malignant tumors.

Autoimmune antibodies and asbestos exposure: Evidence from Wittenoom, Western Australia.

BACKGROUND: Studies comparing different forms of asbestos are rare, and limited by the failure to compare results with unexposed populations. We compare autoimmune responses among former workers and residents of the crocidolite mining and milling town of Wittenoom, Western Australia, with an unexposed population. METHODS: ANA testing using indirect immunofluorescence was performed on randomly selected serum samples from Wittenoom workers or residents and compared with those from participants of another unexposed cohort study. RESULTS: ANA scores were higher in the Wittenoom participants compared with Busselton and the odds of being ANA positive was fivefold greater among Wittenoom participants than Busselton (OR 5.5, 95%CI 2.3-13.0). CONCLUSIONS: This study is the first to report increased ANA positivity among persons exposed exclusively to crocidolite. This finding of a high frequency of positive ANA tests among crocidolite-exposed subjects may be an indicator for an increased risk of systemic autoimmune diseases and needs further scrutiny.

Comparison of outcomes following a cytological or histological diagnosis of malignant mesothelioma.

BACKGROUND: Survival with the epithelioid subtype of malignant mesothelioma (MM) is longer than the biphasic or sarcomatoid subtypes. There is concern that cytology-diagnosed epithelioid MM may underdiagnose the biphasic subtype. This study examines survival differences between patients with epithelioid MM diagnosed by cytology only and other subtypes diagnosed by histology. METHODS: Demographics, diagnosis method, MM subtype and survival were extracted from the Western Australia (WA) Mesothelioma Registry, which records details of all MM cases occurring in WA. RESULTS: A total of 2024 MM cases were identified over 42 years. One thousand seven hundred forty-four (86.2%) were male, median (IQR) age was 68.6 (60.4-77.0) years. A total of 1212 (59.9%) cases were identified as epithelioid subtype of which 499 (41.2%) were diagnosed using fluid cytology only. Those with a cytology-only diagnosis were older than the histology group (median 70.2 vs 67.6 years, P<0.001), but median survival was similar (cytology 10.6 (5.5-19.2) vs histology 11.1 (4.8-19.8) months, P=0.727) and Cox regression modelling adjusting for age, sex, site and time since first exposure showed no difference in survival between the different diagnostic approaches. CONCLUSIONS: Survival of cytologically and histologically diagnosed epithelioid MM cases does not differ. A diagnostic tap should be considered adequate to diagnose epithelioid MM without need for further invasive testing.

Correlation of ultra-low dose chest CT findings with physiologic measures of asbestosis.

OBJECTIVES: The correlation between ultra low dose computed tomography (ULDCT)-detected parenchymal lung changes and pulmonary function abnormalities is not well described. This study aimed to determine the relationship between ULDCT-detected interstitial lung disease (ILD) and measures of pulmonary function in an asbestos-exposed population. METHODS: Two thoracic radiologists independently categorised prone ULDCT scans from 143 participants for ILD appearances as absent (score 0), probable (1) or definite (2) without knowledge of asbestos exposure or lung function. Pulmonary function measures included spirometry and diffusing capacity to carbon monoxide (DLCO). RESULTS: Participants were 92% male with a median age of 73.0 years. CT dose index volume was between 0.6 and 1.8 mGy. Probable or definite ILD was reported in 63 (44.1%) participants. Inter-observer agreement was good (k = 0.613, p < 0.001). There was a statistically significant correlation between the ILD score and both forced expiratory volume in 1 second (FEV1) and forced vital capacity (FVC) (r = -0.17, p = 0.04 and r = -0.20, p = 0.02). There was a strong correlation between ILD score and DLCO (r = -0.34, p < 0.0001). CONCLUSION: Changes consistent with ILD on ULDCT correlate well with corresponding reductions in gas transfer, similar to standard CT. In asbestos-exposed populations, ULDCT may be adequate to detect radiological changes consistent with asbestosis. KEY POINTS: • Interobserver agreement for the ILD score using prone ULDCT is good. • Prone ULDCT appearances of ILD correlate with changes in spirometric observations. • Prone ULDCT appearances of ILD correlate strongly with changes in gas transfer. • Prone ULDCT may provide sufficient radiological evidence to inform the diagnosis of asbestosis.

Functional abilities in children and adults with the CDKL5 disorder.

Functional abilities in the CDKL5 disorder have been described as severely impaired, yet some individuals are able to run and use phrases for speech. Our study investigated gross motor, hand function, and expressive communication abilities in individuals with the CDKL5 disorder. Data for 108 females and 16 males registered with the International CDKL5 disorder database and with a pathogenic CDKL5 mutation were analyzed. Relationships between functional abilities, age, genotype, and gender were analyzed using regression models. Over half of the females could sit on the floor and nearly a quarter could walk 10 steps. Fewer males could complete these tasks although one boy was able to sit, walk, and run. Most females and few males were able to pick up a large object. Females mostly used gestures to communicate while males mostly used other forms of non-verbal communication. Compared to those with no functional CDKL5 protein, individuals with truncating variants after aa 781 were more likely to be able to stand (OR 5.7, 95%CI 1.2, 26.6) or walk independently (4.3, 95%CI 0.9, 20.5), and use more advanced communication methods such as words (OR 6.1, 95%CI 1.5-24.2). Although abilities were markedly impaired for the majority with the CDKL5 disorder, some females and a few males had better functional abilities. This variability may be related to underlying gene variants, with females with a late truncating variant having better levels of ability than those with no functional protein. © 2016 Wiley Periodicals, Inc.

Ultra low dose CT screen-detected non-malignant incidental findings in the Western Australian Asbestos Review Programme.

BACKGROUND AND OBJECTIVE: Computed tomography (CT)-based studies of asbestos-exposed individuals report a high prevalence of lung cancer, but the utility of low dose CT (LDCT) to screen asbestos-exposed populations is not established. We aimed to describe the prevalence of indeterminate pulmonary nodules and incidental findings on chest LDCT of asbestos-exposed subjects in Western Australia. METHODS: A total of 906 subjects from the Western Australian Asbestos Review Programme underwent LDCT of the chest as part of regular annual review. An indeterminate (solid) nodule was defined as >50 mm3 and part-solid/non-solid nodules >5 mm. The presence of asbestos-related diseases was recorded with a standardized report. RESULTS: Subjects were mostly (81%) men with a median age of 70 years. Fifty-eight (6.5%) participants were current smokers, 511 (56.4%) ex-smokers and 325 (36.4%) never-smokers. One hundred and four indeterminate nodules were detected in 77 subjects (8.5%); of these, eight cases had confirmed lung cancer (0.88%). Eighty-seven subjects (9.6%) had incidental findings that required further investigation, 42 (4.6%) from lower airways inflammation. The majority of nodules were solid, 4-6 mm and more common with age. Five hundred and eighty (64%) subjects had pleural plaques, and 364 (40.2%) had evidence of interstitial lung disease. CONCLUSION: The prevalence of LDCT-detected indeterminate lung nodules in 906 individuals with significant asbestos exposure was 8.5%, lower than many other CT studies. Clinically important incidental findings were found in 9.4%, predominantly related to lower respiratory tract inflammation. LDCT appears to effectively describe asbestos-related diseases and is likely to be an acceptable modality to monitor asbestos-exposed individuals.

Early mortality from external causes in Aboriginal mothers: a retrospective cohort study.

BACKGROUND: Maternal loss can have a deep-rooted impact on families. Whilst a disproportionate number of Aboriginal women die from potentially preventable causes, no research has investigated mortality in Aboriginal mothers. We aimed to examine the elevated mortality risk in Aboriginal mothers with a focus on external causes. METHODS: We linked data from four state administrative datasets to identify all women who had a child from 1983 to 2010 in Western Australia and ascertained their Aboriginality, socio-demographic details, and their dates and causes of death prior to 2011. Comparing Aboriginal mothers with other mothers, we estimated the hazard ratios (HRs) for death by any external cause and each of the sub-categories of accident, suicide, and homicide, and the corresponding age of their youngest child. RESULTS: Compared to non-Aboriginal mothers and after adjustment for parity, socio-economic status and remoteness, Aboriginal mothers were more likely to die from accidents [HR = 6.43 (95 % CI: 4.9, 8.4)], suicide [HR = 3.46 (95 % CI: 2.2, 5.4)], homicide [HR = 17.46 (95 % CI: 10.4, 29.2)] or any external cause [HR = 6.61 (95 % CI: 5.4, 8.1)]. For mothers experiencing death, the median age of their youngest child was 4.8 years. CONCLUSION: During the study period, Aboriginal mothers were much more likely to die than other mothers and they usually left more and younger children. These increased rates were only partly explained by socio-demographic circumstances. Further research is required to examine the risk factors associated with these potentially preventable deaths and to enable the development of informed health promotion to increase the life chances of Aboriginal mothers and their children.

Changes in Caesarean Delivery Rates in Western Australia from 1995 to 2010 by Gestational Age at Birth.

BACKGROUND: The caesarean delivery rate in the developed world has been increasing. It is not well understood how caesarean delivery rates have changed by gestational age at birth in Western Australia, particularly in relation to the introduction of the early-term delivery guidelines in Australia in 2006. METHODS: Data from the Western Australian Midwives Notification System were used to identify 193,136 singletons born to primiparous women at 34-42 weeks' gestation during 1995-2010. Caesarean delivery rates were calculated by gestational age group (34-36 weeks, 37-38 weeks, and 39-42 weeks) and stratified into pre-labour and in-labour caesarean delivery. The average annual percent change (AAPC) for the caesarean delivery rates was calculated using joinpoint regression. Log-binomial regression was used to estimate the risk of having a caesarean delivery while adjusting for maternal and antenatal factors. RESULTS: Caesarean delivery rates rose steadily from 1995 to 2005 (AAPC = 5.9%, [95% confidence interval (CI) 4.9, 6.9]), but stabilised since then (AAPC = 0.9%, [95% CI -1.9, 3.8]). The rate of in-labour caesarean deliveries rose consistently from 1995 to 2010 across all gestational age groups. The pre-labour caesarean delivery rate rise was most dominant at 37-38 weeks' gestation from 1995 to 2005 (AAPC = 6.8%, [95% CI 5.4, 8.2]), but declined during 2006-10 (AAPC = -4.5, [95% CI -6.7, -2.3]), while at the same time the rate at 39-42 weeks rose slightly. CONCLUSIONS: The rise in pre-labour caesarean deliveries during 1995-2005 occurred predominantly at 37-38 weeks' gestation, but declined again from 2006 to 2010. This suggests that the recently developed Australian early-term delivery guidelines may have had some success in reducing early-term deliveries in Western Australia.

Effect of N-acetylcysteine supplementation on oxidative stress status and alveolar inflammation in people exposed to asbestos: a double-blind, randomized clinical trial.

BACKGROUND AND OBJECTIVE: Many of the pathological consequences in the lung following inhalation of asbestos fibres arise as a consequence of persistent oxidative stress and inflammation. Inflammatory responses can be observed in asymptomatic asbestos-exposed individuals. There are currently no interventions to reduce inflammatory or oxidative responses to asbestos before disease develops. We investigated the effects of oral N-acetylcysteine (NAC) on indicators of inflammation or oxidative stress in asymptomatic people previously exposed to asbestos. METHODS: A double-blind, randomized, placebo-controlled study was conducted to assess the effectiveness and safety of 1800 mg of NAC given orally over a period of 4 months. This was a proof of principle study. Effectiveness was assessed using indicators of inflammation or oxidation as primary end-points. Serum levels of total combined thiols (cysteine, cysteinylglycine, glutathione and homocysteine) were used to monitor the NAC supplementation. RESULTS: Thirty-four subjects were randomly allocated to NAC and 32 to placebo. Serum levels of total combined thiols were similar between the groups after intervention. There were no differences in levels of inflammatory or oxidative stress end-points between the groups. No adverse effects were identified. CONCLUSIONS: No evidence was found that NAC supplementation replenishes total combined thiols in the blood of healthy subjects with a history of asbestos exposure. There was also no evidence of reduced indicators of inflammation or oxidative stress. Further studies should determine the conditions required to increase levels of total anti-oxidant capacity in the blood and in the lungs of subjects with either asbestos-related diseases or subclinical lung inflammation.

Early respiratory infection is associated with reduced spirometry in children with cystic fibrosis.

RATIONALE: Pulmonary inflammation, infection, and structural lung disease occur early in life in children with cystic fibrosis. OBJECTIVES: We hypothesized that the presence of these markers of cystic fibrosis lung disease in the first 2 years of life would be associated with reduced lung function in childhood. METHODS: Lung function (forced expiratory volume in the first three-quarters of a second [FEV0.75], FVC) was assessed in individuals with cystic fibrosis diagnosed after newborn screening and healthy subjects during infancy (0-2 yr) and again at early school age (4-8 yr). Individuals with cystic fibrosis underwent annual bronchoalveolar lavage fluid examination, and chest computed tomography. We examined which clinical outcomes (pulmonary inflammation, infection, structural lung disease, respiratory hospitalizations, antibiotic prophylaxis) measured in the first 2 years of life were associated with reduced lung function in infants and young children with cystic fibrosis, using a mixed effects model. MEASUREMENTS AND MAIN RESULTS: Children with cystic fibrosis (n = 56) had 8.3% (95% confidence interval [CI], -15.9 to -6.6; P = 0.04) lower FEV0.75 compared with healthy subjects (n = 18). Detection of proinflammatory bacterial pathogens (Pseudomonas aeruginosa, Staphylococcus aureus, Haemophilus influenzae, Aspergillus species, Streptococcus pneumoniae) in bronchoalveolar lavage fluid was associated with clinically significant reductions in FEV0.75 (ranging between 11.3 and 15.6%). CONCLUSIONS: The onset of lung disease in infancy, specifically the occurrence of lower respiratory tract infection, is associated with low lung function in young children with cystic fibrosis. Deficits in lung function measured in infancy persist into childhood, emphasizing the need for targeted therapeutic interventions in infancy to maximize functional outcomes later in life.

Correlation of lung function with ultra-low-dose CT-detected lung parenchymal abnormalities: a cohort study of 1344 asbestos exposed individuals.

INTRODUCTION: Deliberate exposure to medical ionising radiation should be as low as reasonably practicable but the reduction of radiation from CT should be balanced against diagnostic image quality. The ability of ultra-low-dose CT (uLDCT: similar radiation to chest X-ray) to demonstrate low contrast abnormalities (emphysema and interstitial lung abnormality (ILA)) is unclear.The aim of this cross-sectional study was to analyse the lung parenchymal findings from uLDCT scans against physiological measures of respiratory function. METHODS: WA Asbestos Review Programme participants were eligible if they had an uLDCT scan and lung function assessment between Janary and December 2018. All scans were performed using a single CT machine and reported using a standardised, semiquantitative synoptic report which includes emphysema and linear fibrosis (ILA) scores. RESULTS: Of 1344 participants, median (IQR) age was 72.0 (65.0-78.0) years, the majority were males (84.9%) with mixed occupational asbestos exposure (68.1%). There were 721 (53.6%) with no abnormality, 158 (11.8%) with emphysema, 465 (34.6%) with ILA. Mean radiation dose was 0.12 mSv. There was statistically significant between group differences for all physiological parameters of lung function compared with controls. For instance, the emphysema score significantly correlated with obstructive forced expiratory volume in 1 s (FEV1)/forced vital capacity ratio (r=0.512), per cent predicted FEV1 (r=0.24) and lower diffusion of carbon monoxide (DLCO) (r=0.337). Multivariate modelling demonstrated that increasing age, emphysema and fibrosis scores predicted reduced DLCO (adjusted R2=0.30). DISCUSSION: uLDCT-detected parenchymal lung abnormalities correlate strongly with significant changes on lung function testing suggesting the observed CT abnormalities are of physiological and clinical significance.

Interpregnancy intervals and adverse birth outcomes in high-income countries: An international cohort study.

BACKGROUND: Most evidence for interpregnancy interval (IPI) and adverse birth outcomes come from studies that are prone to incomplete control for confounders that vary between women. Comparing pregnancies to the same women can address this issue. METHODS: We conducted an international longitudinal cohort study of 5,521,211 births to 3,849,193 women from Australia (1980-2016), Finland (1987-2017), Norway (1980-2016) and the United States (California) (1991-2012). IPI was calculated based on the time difference between two dates-the date of birth of the first pregnancy and the date of conception of the next (index) pregnancy. We estimated associations between IPI and preterm birth (PTB), spontaneous PTB, and small-for-gestational age births (SGA) using logistic regression (between-women analyses). We also used conditional logistic regression comparing IPIs and birth outcomes in the same women (within-women analyses). Random effects meta-analysis was used to calculate pooled adjusted odds ratios (aOR). RESULTS: Compared to an IPI of 18-23 months, there was insufficient evidence for an association between IPI <6 months and overall PTB (aOR 1.08, 95% CI 0.99-1.18) and SGA (aOR 0.99, 95% CI 0.81-1.19), but increased odds of spontaneous PTB (aOR 1.38, 95% CI 1.21-1.57) in the within-women analysis. We observed elevated odds of all birth outcomes associated with IPI ≥60 months. In comparison, between-women analyses showed elevated odds of adverse birth outcomes for <12 month and >24 month IPIs. CONCLUSIONS: We found consistently elevated odds of adverse birth outcomes following long IPIs. IPI shorter than 6 months were associated with elevated risk of spontaneous PTB, but there was insufficient evidence for increased risk of other adverse birth outcomes. Current recommendations of waiting at least 24 months to conceive after a previous pregnancy, may be unnecessarily long in high-income countries.

Survival with Rett syndrome: comparing Rett's original sample with data from the Australian Rett Syndrome Database.

AIM: rett syndrome is a severe neurodevelopmental disorder that typically affects females. Little is known about the natural history and survival time of these females. METHOD: we compared the survival of all Austrian female participants from Rett's historical cohort (1966) with that of affected females registered in the Australian Rett Syndrome Database. The analysis included both Kaplan-Meier analysis and a log-rank test for equality of survivor functions. RESULTS: of females in the original Austrian group, three are still alive. The median age at death was 13 years 4.8 months. The probability of survival up to the age of 25 years was 21%, compared with 71% in the Australian cohort (p<0.001). We found no practical or statistically significant differences in survival between the various birth year groups within the Australian cohort. INTERPRETATION: our data indicate that survival of females with Rett syndrome has improved since the late 1960s but that there has been shown no change in survival over the last 30 years, possibly because the follow-up time has been too short.

Pregnancy and birth outcomes of mothers with intellectual disability and their infants: Advocacy needed to improve well-being.

BACKGROUND: Pregnancy in women with intellectual disability (ID) is increasingly recognised, along with their increased likelihood of experiencing risk factors for adverse pregnancy and infant outcomes. OBJECTIVES: We aimed to compare risks of socio-demographic, pregnancy and infant outcomes of women with ID to other women. METHODS: All mothers with children born in Western Australia from 1983 to 2012 were linked to the population-based Intellectual Disability Exploring Answers database. Mothers with ID were matched by age and Aboriginality to a population sample of mothers without ID. Pregnancy and birth outcomes were compared for both groups and logistic regression was used to compare outcome risks. RESULTS: Compared to non-Aboriginal mothers without ID, non-Aboriginal mothers with ID were more likely to be of low socio-economic status, be without partner, smoke antenatally and have pre-existing asthma. They had a 1.5 times increased risk of pregnancy complications, specifically pre-eclampsia, urinary tract infection, threatened preterm labour and post-partum haemorrhage. After adjustment for maternal medical conditions and pregnancy complications, infants of Aboriginal mothers with ID had twice the risk of preterm birth and 1.6 times the risk of having percentage of optimal head circumference <95% compared to infants of Aboriginal mothers without ID. Infants of both Aboriginal and non-Aboriginal mothers with ID were more likely to have percentage of optimal birth weight <85% compared to those without ID. CONCLUSIONS: For mothers with ID, modifiable risk factors for adverse outcomes need addressing. They may require additional assistance during pregnancy, including more frequent consultations and support to assist with pregnancy management.

Fractional polynomials and model selection in generalized estimating equations analysis, with an application to a longitudinal epidemiologic study in Australia.

In epidemiologic studies, researchers often need to establish a nonlinear exposure-response relation between a continuous risk factor and a health outcome. Furthermore, periodic interviews are often conducted to take repeated measurements from an individual. The authors proposed to use fractional polynomial models to jointly analyze the effects of 2 continuous risk factors on a health outcome. This method was applied to an analysis of the effects of age and cumulative fluoride exposure on forced vital capacity in a longitudinal study of lung function carried out among aluminum workers in Australia (1995-2003). Generalized estimating equations and the quasi-likelihood under the independence model criterion were used. The authors found that the second-degree fractional polynomial models for age and fluoride fitted the data best. The best model for age was robust across different models for fluoride, and the best model for fluoride was also robust. No evidence was found to suggest that the effects of smoking and cumulative fluoride exposure on change in forced vital capacity over time were significant. The trend 1 model, which included the unexposed persons in the analysis of trend in forced vital capacity over tertiles of fluoride exposure, did not fit the data well, and caution should be exercised when this method is used.