Methods old and new for analyzing occupational cohort data

The person-years approach to analyzing mortality data from occupational cohorts was introduced in the midtwentieth century. It cross-classifies all observed deaths and observation times into cells, computes the number of expected deaths for each cell based on referenced mortality rates, and then examines the ratio of total number of observed deaths to total number of expected deaths (the standardized mortality ratio). The maximum likelihood method of statistical inference was developed in the early twentieth century. However, only recently has it been applied to the analysis of occupational cohort data. When so applied, it provides estimates of measures of association between exposures and disease by maximizing the probability of the observed data. This paper shows how recent developments in the use of this tool justify and extend the person-years approach. In particular, problems with the standardized mortality ratio cited in the literature are shown to result from reliance on assumptions that are inappropriate for the data at hand. Methods for testing these assumptions are described. The discussion is illustrated with examples from occupational cohort studies of lung cancer.     

Fifteen years after the Gozan-Dong glass fiber outbreak, Incheon in 1995

Objectives

In 1995, an outbreak survey in Gozan-dong concluded that an association between fiberglass exposure in drinking water and cancer outbreak cannot be established. This study follows the subjects from a study in 1995 using a data linkage method to examine whether an association existed. The authors will address the potential benefits and methodological issues following outbreak surveys using data linkage, particularly when informed consent is absent.

Methods

This is a follow-up study of 697 (30 exposed) individuals out of the original 888 (31 exposed) participants (78.5%) from 1995 to 2007 assessing the cancer outcomes and deaths of these individuals. The National Cancer Registry (KNCR) and death certificate data were linked using the ID numbers of the participants. The standardized incidence ratio (SIR) and standardized mortality ratio (SMR) from cancers were calculated by the KNCR.

Results

The SIR values for all cancer or gastrointestinal cancer (GI) occurrences were the lowest in the exposed group (SIR, 0.73; 95% CI, 0.10 to 5.21; 0.00 for GI), while the two control groups (control 1: external, control 2: internal) showed slight increases in their SIR values (SIR, 1.18 and 1.27 for all cancers; 1.62 and 1.46 for GI). All lacked statistical significance. All-cause mortality levels for the three groups showed the same pattern (SMR 0.37, 1.29, and 1.11).

Conclusions

This study did not refute a finding of non-association with a 13-year follow-up. Considering that many outbreak surveys are associated with a small sample size and a cross-sectional design, follow-up studies that utilize data linkage should become standard procedure.     

Mortality among professional drivers

The mortality of truck drivers and taxi drivers was studied in Reykjavík. The national mortality rate was used for comparison, and the follow-up lasted until 1 December 1988. The 868 truck drivers (28,788.0 person-years) had an excess of lung cancer deaths [24 observed, 11.2 expected, standardized mortality ratio (SMR) 2.14], but fewer deaths than expected from respiratory diseases (15 observed versus 30.1 expected). The SMR from lung cancer did not steadily increase as the duration of employment increased, nor did it change with the length of follow-up. The SMR values did not deviate substantially from unity for the taxi drivers. Since the high mortality from lung cancer among the truck drivers did not seem to be due to their smoking habits, it might have been caused by one or more occupational factors, especially in light of this group's exposure to engine exhaust gases.     

Mortality and causes of death in eating disorders in Denmark 1970-1993: a case register study

OBJECTIVE: The aim of this study is to describe mortality rates and causes of death for patients with eating disorders. METHOD: By means of record-linkage, the study includes all patients admitted and diagnosed as suffering from an eating disorder according to the ICD-8 classification system during the period 1970-1993 at any Danish psychiatric (since 1970) or somatic department (since 1977). The study includes 2,763 cases, of which 237 are males. Maximum follow-up time is 23 years and mean follow-up time is 10.3 years. RESULTS: Crude mortality at follow-up is 8.4%. A significant excess mortality is demonstrated since the standardized mortality ratio (SMR) of the total patient population is 6.69 (CI 5.68-7.83) and the highest rate ratio (RR) of 14.92 (CI 9.66-22.03) relates to women aged 25-29. DISCUSSION: The study documents a significant excess mortality among eating-disordered patients.     

The mortality of lead smelter workers: an update.

OBJECTIVES. Mortality studies of lead workers have shown excesses of nonmalignant renal disease and cerebrovascular disease. Animal studies and one human study have shown excess kidney cancer. We have updated a mortality study of male lead smelter workers (n = 1990). METHODS. An analysis was conducted using standard life table techniques. The updated analysis added 11 years of follow-up and 363 new deaths. RESULTS. The original study had found elevated but nonsignificant risks for kidney cancer, stroke, and nonmalignant renal disease, probably attributable to lead exposure. Deaths from accidents and nonmalignant respiratory disease were significantly elevated, but probably not as a result of lead exposure. In the updated study, no new deaths from nonmalignant renal disease occurred (9 observed, standardized mortality ratio = 1.21). Three more deaths from kidney cancer were observed, yielding a standardized mortality ratio of 1.93 (9 observed, 95% CI = 0.88, 3.67), which increased for those who had worked in areas with the highest lead exposure (8 observed, standardized mortality ratio = 2.39, 95% CI = 1.03, 4.71). Cerebrovascular disease remained elevated for those with more than 20 years of exposure (26 observed, standardized mortality ratio = 1.41, 95% CI = 0.92, 2.07). CONCLUSIONS. This cohort with high lead exposure showed a diminishing excess of death from nonmalignant renal disease, a continued excess from kidney cancer, and an excess of cerebrovascular disease only in those with longest exposure to lead.     

Mortality and cause of death among first-generation Japanese in S?o Paulo, Brazil

The death certificates of first-generation Japanese in the city of S?o Paulo, Brazil, were analysed for underlying causes of death, and standardized mortality ratio (SMR) values were calculated using the age-specific mortality rates for Japan and S?o Paulo. Mortality rates for all causes were lower than those of the general population in S?o Paulo for all age groups and both sexes, and were almost equal to those in Japan. Compared with the mortality in Japan, significantly higher SMR values were obtained for diabetes mellitus, ischaemic heart disease, respiratory diseases (females), motor vehicle accidents, and homicide (males), whereas SMR values were low for all causes (males), other heart diseases, cerebrovascular disease, chronic liver diseases (males), suicide, and malignant neoplasms (males). On the other hand, compared with the mortality for S?o Paulo, SMR values were significantly low for most causes of death. The potential cultural and social factors contributing to these changes in mortality pattern are discussed.     

Association of Body Weight Variability with Adverse Cardiovascular Outcomes in Patients with Pre-Dialysis Chronic Kidney Disease

To investigate the association of body weight variability (BWV) with adverse cardiovascular (CV) outcomes in patient with pre-dialysis chronic kidney disease (CKD), a total of 1867 participants with pre-dialysis CKD from Korean Cohort Study for Outcomes in Patients With Chronic Kidney Disease (KNOW-CKD) were analyzed. BWV was defined as the average absolute difference between successive values. The primary outcome was a composite of non-fatal CV events and all-cause mortality. Secondary outcomes were fatal and non-fatal CV events and all-cause mortality. High BWV was associated with increased risk of the composite outcome (adjusted hazard ratio (HR) 1.745, 95% confidence interval (CI) 1.065 to 2.847) as well as fatal and non-fatal CV events (adjusted HR 1.845, 95% CI 1.136 to 2.996) and all-cause mortality (adjusted HR 1.861, 95% CI 1.101 to 3.145). High BWV was associated with increased risk of fatal and non-fatal CV events, even in subjects without significant body weight gain or loss during follow-up periods (adjusted HR 2.755, 95% CI 1.114 to 6.813). In conclusion, high BWV is associated with adverse CV outcomes in patients with pre-dialysis CKD.     

Interhospital variations in admission severity-adjusted hospital mortality and morbidity.

In this study hospital admissions are categorized into admission severity groups based on key clinical findings. Severity of illness is determined again later in the hospital stay after treatment has been initiated. High severity on this second review is labeled major morbidity or morbidity, depending on the severity level, and these rates serve as a health outcome indicator along with in-hospital mortality. This study's findings show, for ten hospitals randomly selected from MedisGroups users, considerable interhospital variation in standardized mortality and morbidity ratios for ten frequently occurring DRGs on the adult medical service. After adjusting for admission severity and case mix, three of the ten study hospitals have a statistically significant (p less than .01) difference between the hospital's standardized mortality ratio and 1.0. Such a significant difference exists for the standardized major morbidity ratio of four hospitals and for the standardized morbidity ratio of three hospitals. At the DRG-specific level, our results show that 8.9 percent, 4.4 percent, and 15.0 percent of the hospital-specific mortality, major morbidity, and morbidity ratios, respectively, are statistically significant. Most hospital outliers have fewer deaths or morbid cases than expected. We caution that the study hospitals may not be representative of a larger group of U.S. hospitals.     

A retrospective cohort mortality study of a phosphate fertilizer production facility

A retrospective cohort mortality study of phosphate fertilizer production workers was undertaken to determine whether this group is at increased risk of dying from any cause, particularly from lung cancer. A total of 3,199 workers who had ever been employed at one facility were included in this investigation. These workers were followed for vital status ascertainment from their first date of employment up to December 1, 1977, or the date of death, whichever occurred first. Overall, no statistically significant elevations in cause-specific mortality were observed for the entire study population. However, when the analysis was stratified by duration of employment, and length of follow-up, a statistically significant (P less than .05) excess in lung cancer mortality was observed among workers with more than 10 yr of employment and follow-up (standardized mortality ratio = 411). Because of the small number of deaths involved, and because we had prior knowledge of a lung cancer cluster at this plant, we believe that these findings should be viewed as suggestive, and that other investigations in plants with similar exposures are needed to clarify whether an occupationally related lung cancer excess truly exists.     

Correlation of cerebrovascular disease standardized mortality ratios with dietary sodium and the sodium/potassium ratio among the Japanese population

One thousand seventy-two 24-hr diet duplicate samples were collected from inhabitants of 49 regions in various parts of Japan during the winters of 1977-1981. An additional 238 samples were collected in an adjacent summer. The samples were analyzed for sodium (Na) and potassium (K) by flame atomic absorption spectrometry and for chloride (Cl) with a chloride counter. The winter-summer differences in Na, Cl, and Na/K were essentially negligible. When the regional means of Na, K, Cl, and Na/K were compared with the 1969-1978 standardized mortality ratios of each region, positive and significant correlations were observed between winter Na and the standardized mortality ratios for cerebrovascular disease (P less than 0.01), cerebral infarction (P less than 0.01), and subarachnoid hemorrhage (P less than 0.05) in both males and females. The correlation (P less than 0.01) with the cerebrovascular disease standardized mortality ratio was further confirmed by the values for 1978-1982. In the case of the Na/K ratio, the correlation with the standardized mortality ratio for each of the three diseases was significant for men (P less than 0.01 or 0.05, depending on the disease) but not for women (P greater than 0.05). Both Na and Na/K showed significant associations with the ischemic heart disease standardized mortality ratio in men (P less than 0.05) but not in women (P greater than 0.05). In contrast, no positive association was found between Na, K, Cl, or Na/K and standardized mortality ratios for diabetes mellitus, liver cirrhosis, tuberculosis, or liver cancer (P greater than 0.05). Current blood pressure did not appear to correlate with any of the Na, K, Cl, or Na/K measurements. The validity of the present observation is discussed.     

秦山核电厂周围居民1988～2000年呼吸系统病死因分析

目的:建立秦山核电厂周围20公里范围内居民健康状况数据库。方法:“居民死亡医学证明书”,结合回顾性调查,获取相关数据。结果:1988～2000年居民因呼吸系统疾病死亡居全死因首位,年均死亡率为218.49/10万(标化死亡率为154.72/10万);死因依次为支气管炎、肺气肿和哮喘,标化死亡率为124.34/10万;肺恶性肿瘤标化死亡率为21.33/10万;肺炎标化死亡率为8.98/10万;尘肺标化死亡率为0.07/10万。肺恶性肿瘤占全恶性肿瘤死亡的20.99%。肺炎、肺恶性肿瘤、支气管炎等呼吸系统疾病死亡随着年龄的增加而上升,在40岁以后出现明显的峰值。支气管炎、肺气肿和哮喘和肺炎的死亡总体呈下降趋势,肺恶性肿瘤则呈逐年上升趋势。结论:呼吸系统疾病已是居民的主要死因,秦山核电站在正常运行过程中排放的少量放射性物质没有对周围环境造成显著的影响。     

Cancer mortality and incidence of mesothelioma in a cohort of wives of asbestos workers in Casale Monferrato, Italy

BACKGROUND: Family members of asbestos workers are at increased risk of malignant mesothelioma (MM). Although the hazard is established, the magnitude of the risk is uncertain, and it is unclear whether risk is also increased for other cancers. Few cohort studies have been reported. OBJECTIVE: The "Eternit" factory of Casale Monferrato (Italy), active from 1907 to 1986, was among the most important Italian plants producing asbestos-cement (AC) goods. In this article we present updated results on mortality and MM incidence in the wives of workers at the factory. METHODS: We studied a cohort of 1,780 women, each married to an AC worker during his employment at the factory but not personally occupationally exposed to asbestos. Cohort membership was defined starting from the marital status of each worker, which was ascertained in 1988 from the Registrar's Office in the town where workers lived. At the end of follow-up (April 2003), 67% of women were alive, 32.3% dead, and 0.7% lost to follow-up. Duration of exposure was computed from the husband's period of employment. Latency was the interval from first exposure to the end of follow-up. RESULTS: The standardized mortality ratio (SMR) for pleural cancer [21 observed vs. 1.2 expected; SMR = 18.00; 95% confidence interval (CI), 11.14-27.52] was significantly increased. Mortality for lung cancer was not increased (12 observed vs. 10.3 expected; SMR = 1.17; 95% CI, 0.60-2.04). Eleven incident cases of pleural MM were observed (standardized incidence ratio = 25.19; 95% CI, 12.57-45.07). CONCLUSIONS: Household exposure, as experienced by these AC workers' wives, increases risk for pleural MM but not for lung cancer.     

Properties of the geometrically averaged ratio, an alternative standardized measure.

In epidemiology, the comparative mortality figure and the standardized mortality ratio are standardized measures in common use. Both are weighted averages of rate ratios (or observed/expected death count ratios) on the arithmetic scale. I propose a new standardized measure, the geometrically averaged ratio (GAR), which is defined through simple averaging on the logarithmic scale. I show that, in addition to providing a valid comparison between populations, the geometrically averaged ratio possesses the following desirable properties: (1) invertibility and invariance of standardized sex ratios and (2) interpopulational comparability with different standards.     

The mortality of young offenders sentenced to prison and its association with psychiatric disorders: a register study

BACKGROUND: We studied the nationwide mortality in Finland of young offenders sentenced to prison, with the advantage of a long-term follow-up in an unselected population. In addition, we aimed to clarify the relationship between psychiatric disorders requiring hospital treatment and early death in young offenders sentenced to prison. METHODS: All offenders sentenced to prison between 1984 and 2000 in Finland and aged 15-21 years when the crime was committed were selected for this study. The mortality of the young offenders was compared with the age- and sex-matched mortality data of the general population, obtained from Statistics Finland. Information on hospital treatment periods for psychiatric diagnoses was collected from the Finnish Health Care Register and linked to the mortality data. RESULTS: The study population consisted of 3,743 young male and 89 young female offenders. Of these, 435 (11.4%) had died by the end of the follow-up period, including 3 girls. The standardized mortality ratio for young male prisoners was 7.4 (95% confidence interval 6.7-8.1). There was a higher mortality rate among young offenders convicted in the later years of the study period. The causes of death were mostly unnatural and often violent. Hospitalization for a psychiatric disorder or substance abuse was significantly associated with the risk of death. However, hospitalization for emotional disorders with an onset specific to childhood and adolescence were associated with a lower death risk. CONCLUSION: The mortality rate in the population of young offenders sentenced to prison is alarmingly high. The high mortality in this group is associated with substance abuse and psychiatric disorders, but not with emotional disorders with an onset specific to childhood and adolescence.     

Statistical analysis of cost outcomes in a randomized controlled clinical trial

This paper suggests an approach to deal with an estimation problem which is often encountered in analyzing the longitudinal cost data gathered in a clinical trial. The source of that estimation problem is twofold: 1) a considerable number of missing data due to treatment-related withdrawal of severely affected patients with high health care costs in only one the treatment groups and 2) a heavily skewed cost distribution due to rare high-cost events. The approach is illustrated using data from a trial comparing 3 different drug regimes. In order to calculate costs per patient-year in case of selectively missing data we extrapolated the costs of patients with incomplete follow-up. Due to the skewness and the associated large variance in costs per patient-year, these costs cannot be analyzed using common parametric statistical methods relying on underlying normal distributions. A logarithmic transformation was performed to approximate a normal distribution, reduce the impact of extreme values and create similar size variances in the treatment groups. An ordinary least squares regression analysis of transformed data then standardized for differences in patient characteristics between the groups. For the retransformation, the so-called smearing estimate was used. This ‘transformation-standardization-retransformation’ approach enabled us to provide more consistent and efficient estimates of cost differences that were shown to be statistically significant and judged to be important.     

Cancer identification using a tumor registry versus death certificates in occupational cohort studies in the United States.

Studies of cancer incidences among occupational cohorts are rarely performed in the United States because of incomplete registration and a limited time period available for follow-up. This study used data from concurrent studies of cancer mortality and incidence among a cohort of 4,528 fire fighters and police officers employed by the cities of Seattle and Tacoma, Washington, between 1944 and 1979 to examine the relative advantages of tumor registry and death certificate information. As expected, an increased ability to study relatively common cancers with low fatality rates was demonstrated using incidence data. The most dramatic example was seen for bladder cancer. Twenty-four bladder cancers had been diagnosed among the study cohort between 1974 and 1989, whereas only two deaths were attributed to this malignancy. The standardized incidence ratio for bladder cancer was 1.05 (95% confidence interval 0.67-1.55), whereas the corresponding standardized mortality ratio was 0.46 (95% confidence interval 0.05-1.65). The observed relative risk estimates for rapidly fatal cancers were similar using the two sources of information, and no increase in precision was observed. Of 142 persons in the registry area who died of cancer during the study period, 20 (14%) had a different site listed on the death certificate than was identified by the registry. Approximately 7% of the potential person-years of follow-up were lost due to migration out of the registry area; loss to follow-up was greater among older and short-term workers, but did not exceed 13% of the person-years. Population-based tumor registries can be a useful resource in the investigation of occupational cancer in the United States, especially for the study of cancers with high survival rates.     

Obesity in adolescence and adulthood and the risk of adult mortality

BACKGROUND There are few long-term follow-up data on the relation between body mass index (BMI) in adolescence and in adulthood, and between adolescent BMI and adult mortality. The present study explores these relations. METHODS: In Norwegian health surveys during 1963-1999, height and weight were measured for 128,121 persons in a standardized way both in adolescence (age 14-19 years) and 10 or more years later. Persons were followed for an average of 9.7 years after the adult measurement. Cox proportional hazard regression models were used to study the association between adolescent and adult BMI and mortality. RESULTS: The odds ratio of obesity (BMI >/=30) in adulthood increased steadily with BMI in adolescence, from 0.2 for low BMI up to 16 for very high BMI. Very high adolescent BMI was associated with 30-40% higher adult mortality compared with medium BMI. Adjusting for adult BMI explained most of the association of adolescent obesity and mortality, especially among men. Adjustment for smoking did not change the results. CONCLUSIONS: Obesity in adolescence tends to persist into adulthood. Adolescent obesity is also connected to excess mortality, but this excess seems to be explained mostly by obesity in adulthood. High BMI in adolescence seems to be predictive of both adult obesity and mortality.     

Mortality by employment status in the National Longitudinal Mortality Study

A mortality follow-up of 452, 192 persons aged 25 years or more who were characterized with respect to employment status was conducted using the National Death Index for the years 1979 through 1983. The cohort, part of the National Longitudinal Mortality Study, was drawn from Current Population Survey samples representative of the US population using selected months during the years 1979-1983. Employed persons aged 25-64 years were found to have standardized mortality ratios from 61% to 74% of the average, depending upon their sex and race. Unemployed men had standardized mortality ratios slightly above 100, but these values were 1.6 and 2.2 times higher than those for employed white men and black men, respectively. Those classified as unable to work had very high mortality ratios, from two to seven times the average. In the older age groups, 65 years or more, very low mortality ratios were found for those who were still employed. These relations were maintained after adjustment for family income and educational level. These results 1) describe the magnitude of mortality risk for clearly defined employment categories, 2) identify segments of the population with especially high mortality requiring greater public health recognition, and 3) suggest further research into the health consequences of the various employment/nonemployment conditions.     

Long-term survival and cause-specific mortality in patients with cirrhosis of the liver: a nationwide cohort study in Denmark

Mortality from cirrhosis of the liver has been examined in few long-term follow-up studies. In the Danish National Registry of Patients, 1982-1989, we identified a cohort of 10,154 patients with liver cirrhosis and divided them according to the etiology of their liver disease. Causes of death were identified in the Danish Death Registry, 1982-1993. We estimated relative survival and standardized mortality ratios by comparing with the mortality in the general population. The 10-year relative survival was worse in patients with alcoholic cirrhosis (34%) or nonspecified cirrhosis (32%) than in patients with primary biliary cirrhosis (58%) or chronic hepatitis (66%). The standardized mortality ratio for all causes of death combined was 12-fold increased, 5-fold excluding cirrhosis-related causes. Mortality in all disease categories was increased, even in those not traditionally related to cirrhosis. In conclusion, patients with cirrhosis of the liver face reduced life expectancy due to several causes of death.     

Psychosocial factors as risk for coronary heart disease--status with special reference to the KORA platform

Psychosocial factors derived from concepts in health psychology and psychopathology are subject of extensive research to assess their power to predict a future coronary artery disease event in apparently healthy subjects. However, bio-behavioural factors have not been implemented in current guidelines of scoring schemes for calculating the risk of coronary events. The presented data were derived from the population-based MONICA Augsburg studies (S1-S3) conducted between 1984 and 1995. The psychosocial data set was available in approximately 13,000 subjects. The KORA follow-up study assessed the vital status for all participants (except for 56 persons) in 1998. Until then, 772 participants (531 men, 241 women) had died. The depressive symptomatology was derived from the von Zerrssen affective symptom check list combining 24 single symptom items with scores ranging from 0 to 3. Risks of total mortality and myocardial infarction were estimated from Cox proportional hazard ratio (HR) models adjusted for age and survey and multiple risk factors. Male participants with high scores in depression exhibited a significantly increased risk in total mortality (adjusted HR: 1.55; 95 % CI: 1.28 - 1.83, p < 0.0001) and for fatal and non-fatal coronary events (adjusted HR: 1.36; 95 % CI: 1.02 - 1.81, p < 0.035). Female participants reported higher values in depression scores; however, depression was not predictive for subsequent total mortality and fatal and non-fatal events in females. Depression in men yielded a significant interaction with obesity and increased levels of C-reactive protein (CRP). The inclusion of psychosocial factors, as demonstrated for depression, is likely to improve prediction of future adverse cardiovascular and total mortality. These factors may also play a crucial role in genotype-phenotype interaction.