21st January 2011, Volume 124 Number 1328

Erin Holmes, Anna Davies, Craig Wright, Neil Pearce, Barry Borman

Differences in mortality rates among different occupation groups have been well documented throughout the past century. The determinants of these disparities include not only the hazards inherent to the workplace, but also external factors such as diet, age, ethnicity and lifestyle which can also vary by occupation.1

In the past, coded occupation in routinely collected health information in New Zealand was used to investigate, and contribute to the evolution of knowledge on, the occupational health and safety risk factors in the New Zealand workforce.2 However, the coding of this field on many of the key datasets has since been discontinued, and this is one of the key issues limiting the effectiveness of New Zealand’s occupational disease and injury surveillance system.2,3 In particular, around 1997/1998 Statistics New Zealand (SNZ) stopped routinely coding the free text occupation field recorded on the Notification of Death Registration Form (BDM 28),2 and admission clerks stopped coding the free text occupation field on hospital discharges.

Following the discontinuation of coding, the high cost of ad hoc coding of routine data confronting researchers has discouraged many occupational health studies; the last published study investigating occupational mortality in New Zealand was for the period 1974–1978, and is now over 20 years old.1

Currently no regular monitoring of occupation-related mortality is occurring in New Zealand. This has resulted in major gaps in evidence and subsequent neglect by policy makers. Historical efforts to address this lack of evidence and reignite interest in this area have encountered various hurdles, delaying subsequent action.3

We have therefore conducted our own coding of the occupation free text field in the New Zealand death registration data for the period 2001-2005 and calculated age- and deprivation- standardised mortality rates and ratios per 100,000 person-years-at-risk for each disease and occupational group, in order to provide more current data on occupational differences in mortality in New Zealand.


We obtained the denominator data from the New Zealand Census 2006 (Statistics New Zealand (SNZ)) and the numerator data from the mortality collection held by the Ministry of Health. All deaths registered in New Zealand during 2001-2005 are included in the analyses. In each instance, we classified people into occupations, at their time of death, using the codes of the New Zealand Standard Classification of Occupations 1999 (NZSCO99)4 which is based on the International Standard Classification of Occupations (ISCO-88),5 a publication of the International Labour Office (ILO).
The Statistics New Zealand (SNZ) classification coder for occupation data was run over the data first, coding 60% of the data. The rest of the data was manually coded as per the coding manual. All coders had previous experience in occupational coding and were blinded to the diagnoses. The majority of the data was independently coded twice (two people).
Death data was further restricted to those records able to be mapped to valid deprivation quintiles based on the NZDep2006 Index of Deprivation.6 This index combines nine variables from the New Zealand 2006 Census which reflect eight dimensions of deprivation, including income, education and housing.
In this study it has been used as a method of controlling differences in mortality that might be attributed to socio-economic deprivation (based on area of residence), rather than those that may be associated with occupational group. We only included male deaths, as more than half of the female deaths were housewives, students, invalids and retired women who we could not assign to an occupational group using the limited death certificate information.
We restricted the occupation data to those aged 15–64 years for two reasons. Firstly, older age groups contained a high proportion of retired people and invalids which could not be assigned to an occupation, and secondly, to enable comparisons to be made with previous research.
The cause of death was determined from the 3-digit disease codes of the International Classification of Diseases (ICD-10-AM)and the data were grouped in to the major disease categories previously used in studies of social class and mortality in New Zealand.1 These were the same groupings as used in the ICD-10-AM manual, except that the grouping of diseases of the circulatory system was split into three categories: ischemic heart disease; cerebrovascular disease; and other diseases of the circulatory system.
Deaths from the following ICD-10-AM codes were excluded O00-O99, P00-P96, R00-R99, Z00-Z99 and U00-U99 (36 deaths were excluded in total). These codes cover pregnancy, childbirth, and the puerperium; certain conditions originating in the perinatal period; symptoms signs and abnormal clinical and laboratory findings; factors influencing health status and contact with health services; and codes for special purposes. This process of restriction and the final number of death records included in the analysis is illustrated in Figure 1. The calculations described below are based on a final numerator of 12,713 deaths and a denominator of 4,009,455 people.
Figure 1. Restriction of numerator and denominator data prior to analysis


* This also excludes 36 deaths from the following ICD-10-AM codes: O00-O99, P00-P96, R00-R99, Z00-Z99 and U00-U99 (see text for further detail).
Directly age-standardised mortality rates per 100,000 person-years at risk and 95% confidence intervals (based on Ulm’s method)8 were calculated for each occupational order using Statistical Analysis Software (SAS).9 Five-year age-strata were weighted using the World Health Organization (WHO) World Standard Population. Indirectly age-standardised ratios (also known as standardised mortality ratios) were calculated for the occupational group analyses due to the small denominator numbers for many groups, along with 95% confidence intervals.
Two sets of expected values were calculated. The first was based on mortality rates for all employed persons, whereas the second was standardised for socioeconomic deprivation using the 2006 New Zealand Index of Deprivation.6


Table 1A shows mortality by major occupational order. The lowest overall mortality rate was for legislators/administrators/managers (1, includes corporate managers). The highest overall rate was for plant and machine operators and assemblers (8, includes industrial plant operators; stationary machine operators and assemblers; drivers and mobile machinery operators; and building and related workers), followed by agriculture and fishery workers (6, includes crop growers, animal producers, forestry workers, hunters and trappers).

((view Table 1A and Table 1B here))

Both these occupation categories continued to have the highest overall mortality rates after standardising for socioeconomic deprivation, with the rate for agriculture and fishery workers becoming higher than that of plant and machine operators and assemblers. Clerks (4, includes office and customer service workers) and service and sales workers (5, includes personal and protective services workers, and sales persons and demonstrators) were the only group whose rate was close to that for all employed persons.

Plant and machine operators and assemblers had the highest rates of mortality for coronary heart/ischemic heart disease, other diseases of the circulatory system, diseases of the respiratory system, and endocrine, nutritional and metabolic diseases. Agriculture and fishery workers had the highest mortality rate for external causes, while Trades workers (7, includes printing, tailors, electricians, metal and machinery, and crafts workers) had the highest mortality rates for cancer and diseases of the digestive system.

Elementary occupations (9, including labourers, caretakers, cleaners and refuse collectors) had the highest mortality rate for diseases of the nervous and genitourinary systems, and mental and behavioural disorders, although only the rate for genitourinary disease was significantly different to that experienced by ‘all employed persons’.

Clerks had the highest mortality rate for cerebrovascular diseases, certain infectious and parasitic diseases and congenital malformations, deformations and chromosomal abnormalities, although only the rate for cerebrovascular disease was significantly different to ‘all employed persons’.

Service and sales workers had the highest mortality rate for ‘other’ diseases (including diseases of the eye, skin, ears, blood and musculoskeletal system and connective tissue), but this was not significantly different to the rate for ‘all employed persons’.

These trends remained following standardisation for socioeconomic deprivation, with the exception of ‘other diseases of the circulatory system’, for which the highest rate was observed for agriculture and fishery workers rather than plant and machine operators, and ‘mental and behavioural disorders’ which the rate was highest for trades workers rather than elementary workers (Table 1B).

Table 2 examines overall mortality for occupational groups by sub-major occupation (23 groups). There were seven groups with significantly low mortality and 10 with significantly high mortality, whereas only one group would be expected by chance alone.

Life science and health associate professionals (i.e. technicians and assistants); personal and protective service workers; market orientated agriculture and fishery workers; all trades workers; all plant and machine operators and assemblers; and labourers and related elementary service workers, had significantly higher mortality rates than expected. Standardising for socioeconomic deprivation only affected the significance of the result for building trades workers; drivers and mobile machinery operators; and labourers and related elementary service workers. This means that the elevated mortality experienced by these occupational groups, compared with all employed people, may be attributed to socioeconomic factors rather than occupational factors.

In contrast, after standardising for socioeconomic deprivation, life science and health professionals—includes life science professionals (i.e. biological scientists) and health professionals (i.e. doctors, nurses, vets, dentists and pharmacists)—experienced significantly higher mortality than expected.

Table 2. Observed mortality in New Zealand males aged 15–64 during 2001–2005 compared to that expected on the basis of all employed males and males in the same deprivation quintile, by occupational group
Occupational group (submajor)
Observed deaths
Relative risk vs all employed*
Relative risk vs deprivation quintile*
1 Legislators, administrators and managers
11 Legislators and Administrators
12 Corporate Managers


0.67 (0.56–0.79)
0.57 (0.54–0.60)

0.89 (0.75–1.04)
0.62 (0.59–0.65)
2 Professionals
21 Physical, Mathematical and Engineering Science Professionals
22 Life Science and Health Professionals
23 Teaching Professionals
24 Other Professionals


0.61 (0.55–0.68)
0.99 (0.86–1.14)
0.76 (0.67–0.87)
0.69 (0.62–0.76)

0.70 (0.62–0.78)
1.23 (1.07–1.42)¥
0.83 (0.73–0.94)
0.82 (0.74–0.90)
3 Technicians and Associate Professionals
31 Physical Science and Engineering Associate Professionals
32 Life Science and Health Associate Professionals
33 Other Associate Professionals


0.96 (0.86–1.06)
2.04 (1.41–2.85)¥
0.79 (0.74–0.85)

1.02 (0.92–1.13)
2.50 (1.73–3.49)¥
0.86 (0.79–0.92)
4 Clerks
41 Office Clerks
42 Customer Services Clerks


1.01 (0.92–1.11)
0.88 (0.67–1.15)

0.95 (0.86–1.04)
0.95 (0.71–1.24)
5 Service and Sales Workers
51 Personal and Protective Services Workers
52 Salespersons, Demonstrators and Models


1.33 (1.23–1.44)¥
0.49 (0.43–0.56)

1.26 (1.17–1.37)¥
0.48 (0.42–0.54)
6 Agriculture and Fishery Workers
61 Market Oriented Agricultural and Fishery Workers


1.26 (1.20–1.32)¥

1.37 (1.31–1.44)¥
7 Trades Workers
71 Building Trades Workers
72 Metal and Machinery Trades Workers
73 Precision Trades Workers
74 Other Craft and Related Trades Workers


1.10 (1.04–1.17)¥
1.25 (1.16–1.35)¥
1.72 (1.44–2.03)¥
2.86 (2.55–3.21)¥

1.06 (1.00–1.12)
1.15 (1.06–1.23)¥
1.67 (1.40–1.97)¥
2.69 (2.39–3.01)¥
8 Plant and Machine Operators and Assemblers
81 Industrial Plant Operators
82 Stationary Machine Operators and Assemblers
83 Drivers and Mobile Machinery Operators
84 Building and Related Workers


4.08 (3.61–4.59)¥
1.95 (1.82–2.08)¥
1.13 (1.06–1.20)¥
1.94 (1.61–2.30)¥

2.91 (2.58–3.28)¥
1.58 (1.47–1.69)¥
0.95 (0.89–1.01)
1.76 (1.47–2.10)¥
9 Elementary Occupations (excluding residuals)
91 Labourers and Related Elementary Service Workers


1.23 (1.16–1.30)¥

1.03 (0.98–1.09)
Total Employed
* Relative risk= Observed deaths/Expected deaths (see text); ¥ Significantly greater than expected.


This analysis has highlighted potential associations between different occupations and cause of death in males aged 15–64 years through the analysis of New Zealand mortality data for 2001–2005. Many of these findings are consistent with those observed in most developed countries, with lower mortality rates apparent in professional and non-manual occupations, and significantly elevated mortality rates in manual occupations.11–16

In particular, the finding that agriculture and fishery workers (including forestry, hunters and trappers), and plant and machine operators and assemblers (including mining, power generation, metal processing, glass, wood and chemical processing plant operators) experience significantly higher mortality ratios than expected, is also evident in other New Zealand1 studies and studies conducted in the United States.17–19

In most cases, differences in overall mortality by occupational group remained or were enhanced following adjustment for socioeconomic deprivation. There was a similar finding in previous research conducted in New Zealand1 and Britain.11,15 This provides further evidence that differences in mortality for selected occupations may be attributed to factors other than social status, income and education.

Many of the results for major disease groupings were also comparable with existing research, with significantly elevated mortality observed for the following disease groupings and occupational groups: Cancer in industrial plant operators20–21 and in other craft and related trades workers.22-24 Elevated risk for cancer has also been observed among meat workers in Australia and New Zealand25,26Ischemic heart disease in industrial plant operators27; and Other diseases of the circulatory system, particularly among industrial factory workers.28,29 This is consistent with the findings of Tamosiūnas et al (2005) that the risk of death from cardiovascular diseases is greater among manual than non-manual workers.12

The elevated risk of death from respiratory diseases among industrial plant operators has also been noted elsewhere, particularly asthma, emphysema and chronic bronchitis among aluminium plant workers30 and silicon carbide smelter workers.31

Higher mortality from external causes among market orientated agriculture and fishery workers24 and industrial plant operators, is evident from other studies, particularly from motor vehicle crashes,18 falling objects,32 machinery, falls,33,34 suicide,16,35 and drowning (among maritime workers).36

Limitations—The limitations of this type of study have been discussed in depth by numerous authors.1,37–39

Firstly, there are problems associated with selection into and survival in particular occupations. The ‘healthy worker effect’ means that anyone who is unemployed due to illness or disability at the time of their death may not be allocated an ‘occupation’.

A further limitation of using occupation at time of death is that the long incubation periods for many conditions mean that the cause(s) of death could be associated with exposure in a previous occupation, rather than that at the time of death.40 Actual exposures and measures of exposure— such as duration and intensity have also not been considered in this study.

Secondly, the occupation data reported on the death registration could be biased (e.g. surviving relatives reporting more prestigious occupations) and/or incomplete, resulting in misclassification. Therefore, some of the findings of this study may underestimate the true relative risks for the most ‘at risk’ populations. Najman et al show through imputation that estimates of inequalities in mortality can change when missing data are accounted for.37

Thirdly, death registrations have not been directly linked with census data which means that there is no guarantee that the individuals enumerated in each occupational group on the census are the same individuals identified with that occupation on their death certificates.41Biddle et al found that numerator-denominator bias can affect the accuracy of traumatic occupational fatality incidence.38

Furthermore, the use of 2006 denominator data for the analysis of deaths occurring between 2001 and 2005 also has implications. Between 2001 and 2006 the population of males aged 15–64 years, in the labour force, increased by approximately 11.1% (Statistics New Zealand). This means that it is likely that the denominator used (Census 2006) will have overestimated the population from which the deaths were drawn (2001–2005). The implications of this limitation are that the mortality rates and relative risks reported in this paper are likely to be much higher in reality.

While these are currently unavoidable limitations of death registration-based studies in New Zealand, in the future, this could be remedied through the linking of individual mortality records (numerator) to the National Health Index (NHI) population (denominator). The NHI is an administrative dataset comprising all individuals that have accessed health services in New Zealand.

While we have adjusted the analyses for socioeconomic deprivation, confounding by extrinsic factors such as smoking, diet and general lifestyle was not directly considered (although some of these factors may be partially controlled for because of their association with deprivation).

Finally, the categories of occupation and cause of death used were broad and may have masked important increases in risk in specific subgroups of occupation and disease and/or diseases. Similarly, while the broad occupational groupings in NZSCO provide a framework for discussing occupational statistics, our findings cannot be generalised to infer causation, particularly given the heterogeneous exposures that occur within these broad groups.

In spite of these limitations, the value of register-based studies in revealing new occupational risks and monitoring older ones is well-established. This approach has recently been used in a comparison of occupational mortality between the Nordic Countries and Japan,41 and remains the most feasible method for monitoring occupational mortality at a national level in New Zealand.

Conclusion—While register-based studies have many limitations if used as the sole basis for decision making and the formulation of intervention policies, they can nevertheless provide useful information on occupational differences in mortality rates, and can form an important component of occupational health.3,24

This paper shows that there continues to be marked differences in mortality between occupations in New Zealand and that many of these differences persist following adjustment for socioeconomic deprivation.

These trends have persisted in New Zealand for over two decades, a testament to the importance of continuing to monitor the situation through the routine coding of occupation on administrative datasets such as mortality, hospitalisations and cancer registrations. To routinely code this free-text field in a similar way to the routine coding of disease, at a centralised point, will ensure a consistent and comprehensive dataset.

Furthermore, the centralised coding of this field will enable the automation of this process, resulting in improvements in accuracy and efficiency over time. Such a resource would allow continued monitoring and encourage exposure studies of occupations with significantly elevated relative risks.


The paper uses records of New Zealand males that died between 2001 and 2005 to investigate variations in overall and disease-specific mortality for different occupation groups. The occupation categories are based on the New Zealand Standard Classification of Occupation 1999 (NZSCO 1999), while the disease categories follow the International Classification of Disease (ICD10-AM). The paper shows that there continues to be marked differences in mortality between occupations in New Zealand and that many of these differences remain once the effect of socioeconomic deprivation has been taken into account.



To estimate occupational mortality rates in New Zealand males for the period 2001–2005.


Occupation information noted in the free text of death records of males aged 15–64 years during 2001–2005 was classified to the New Zealand Standard Classification of Occupation 1999 and age and deprivation standardised mortality rates and ratios were calculated for the major ICD10-AM categories of disease and different occupational groups.


A total of 12,713 male deaths were included in the study. There were marked differences in mortality between occupations. Plant and machine operators and assemblers and agriculture and fishery workers had the overall highest rates. The former had the highest rates for ischemic heart disease, other diseases of the circulatory system, diseases of the respiratory system, and endocrine, nutritional and metabolic diseases, while the latter had the highest rate for external causes.


The last published study investigating occupational disease mortality in New Zealand is now over 20 years old and currently no regular monitoring of occupation-related mortality is occurring in New Zealand. This paper shows that there continues to be marked differences in mortality between occupations in New Zealand and that many of these differences persist following adjustment for socioeconomic deprivation.

Author Information

Erin Holmes, Advisor Epidemiology, Health & Disability Intelligence, Ministry of Health, Wellington; Anna Davies, Senior Advisor Epidemiology, Health & Disability Intelligence, Ministry of Health, Wellington; Craig Wright, Senior Advisor Statistics, Health & Disability Intelligence, Ministry of Health, Wellington; Neil Pearce, Professor and Director, Centre for Public Health Research, Massey University, Wellington; Barry Borman, Associate Professor, Centre for Public Health Research, Massey University, Wellington


This study was funded in part by a grant from the Health Research Council of New Zealand. The Centre for Public Health Research is supported by a Programme Grant from the Health Research Council. We also thank Anna Shum-Pearce, Miria Hudson, Amy Sutherland, and others employed by the Centre for Public Health Research (Massey University) and Health & Disability Intelligence (Ministry of Health) who undertook the time-consuming and tedious task of manually coding the free-text occupation field of the mortality dataset. In addition we thank Yvonne Galloway for her comments and suggestions.


Erin Holmes, Health & Disability Intelligence, Ministry of Health, PO Box 5013, Wellington 6154, New Zealand. Fax: +64 (04) 4962340

Correspondence Email



  1. Pearce N, Howard JK. Occupational Mortality in New Zealand Males 1974–78. Community Health Studies 1985;IX(3):212–8.
  2. Mannetje A, Pearce N. Occupation in routinely collected national health data. Wellington: Centre for Public Health Research, Massey University, 2004.
  3. Pearce N, Dryson E, Feyer A-M, et al. Surveillance of Occupational Disease and Injury in New Zealand: Report to the Minister of Labour. Wellington: NOHSAC, 2005.
  4. Statistics New Zealand. New Zealand Standard Classification of Occupations 1999. Wellington: New Zealand Government, 2001.
  5. International Labour Office. International Standard Classification of Occupations. Geneva: ILO, 1988.
  6. Salmond C, Crampton P, Atkinson J. NZDep2006 Index of Deprivation Wellington: Department of Public Health, University of Otago, 2007.
  7. World Health Organization. International Statistical Classification of Diseases and Related Health Problems. 10th Revision Geneva: WHO, 2007.
  8. Ulm K. Simple Method to Calculate the Confidence Interval of a Standardized Mortality Ratio (SMR). Am J Epidemiol. 1990;131(2):373-5.
  9. SAS 9.1 TS Level 1 M3 [program]. Cary, NC, USA: SAS Institute Inc 2002-2003.
  10. Aronson K, Howe G, Carpenter M, Fair M. Surveillance of potential associations between occupations and causes of death in Canada, 1965-91. Occup Environ Med. 1999 Apr;56(4):265-9.
  11. Wannamethee S, Shaper A. Socioeconomic status within social class and mortality: a prospective study in middle-aged British men. Int J Epidemiol. 1997 Jun;26(3):532-41.
  12. Tamosiūnas A, Reklaitiene R, Domarkiene S, et al. Prevalence of risk factors and risk of mortality in relation to occupational group. Medicina (Kaunas). 2005;41(8):705-12.
  13. Shkolnikov V, Scholz R, Jdanov D, et al. Length of life and the pensions of five million retired German men. Eur J Public Health. 2008 Jun;18(3):264-9.
  14. Rosvall M, Chaix B, Lynch J, et al. Contribution of main causes of death to social inequalities in mortality in the whole population of Scania, Sweden. BMC Public Health. 2006 Mar 28;6:79.
  15. Johnson N, Sorlie P, Backlund E. The impact of specific occupation on mortality in the U.S. National Longitudinal Mortality Study. Demography. 1999 Aug;36(3):355-67.
  16. Australian Institute of Health and Welfare. Australian Health Inequalities 2: Trends in male mortality by broad occupational group, 2005.
  17. Ahn Y, Bena J, Bailer A. Comparison of unintentional fatal occupational injuries in the Republic of Korea and the United States. Inj Prev. 2004 Aug;10(4):199-205.
  18. Cohen M, Clark R, Silverstein B, et al. Work-related deaths in Washington State, 1998-2002. J Safety Res. 2006;37(3):307-19.
  19. Helmkamp J, Lundstrom W. Work-related deaths in West Virginia from July 1996 through June 1999: surveillance, investigation, and prevention. J Occup Environ Med. 2000 Feb;42(2):156-62.
  20. Beall C, Bender T, Cheng H, et al. Mortality among semiconductor and storage device manufacturing workers. J Occup Environ Med. 2005 Oct;47(10):996-1014.
  21. Sathiakumar N, Delzell E. An updated mortality study of workers at a dye and resin manufacturing plant. J Occup Environ Med. 2000 Jul;42(7):762-71.
  22. Besson H, Banks R, Boffetta P. Cancer mortality among butchers: A 24-state death certificate study. J Occup Environ Med. 2006 Mar;48(3):289-93.
  23. Mikoczy Z, Hagmar L. Cancer incidence in the Swedish leather tanning industry: updated findings 1958-99. Occup Environ Med. 2005 Jul;62(7):461-4.
  24. Ugarte M, Artieda L, Ibáñez B, et al. A cohort study to estimate occupational mortality risks in Navarra. Eur J Public Health. 2005 Jun;15(3):305-12.
  25. Fritschi L, Fenwick S, Bulsara M. Mortality and cancer incidence in a cohort of meatworkers. Occup Environ Med. 2003 Sep;60(9):E4.
  26. McLean D, Cheng S, Mannetje A, et al. Mortality and cancer incidence in New Zealand meat workers. Occup Environ Medicine. 2004 Jun;61(6):541-7.
  27. Calvert G, Merling J, Burnett C. Ischemic heart disease mortality and occupation among 16- to 60-year-old males. J Occup Environ Med. 1999 Nov;41(11):960-6.
  28. Diderichsen F, Hallqvist J. Trends in occupational mortality among middle-aged men in Sweden 1961-1990. Int J Epidemiol. 1997 Aug;26(4):782-7.
  29. Pepllonska B, Sobala W, Szeszenia Dabrowska N. Mortality pattern in the cohort of workers exposed to carbon disulfide. Int J Occup Med Environ Health. 2001;14(3):267-74.
  30. Romundstad P, Andersen A, Haldorsen T. Nonmalignant mortality among workers in six Norwegian aluminum plants. Scand J Work Environ Health. 2000 Dec;26(6):470-5.
  31. Romundstad P, Andersen A, Haldorsen T. Non-malignant mortality among Norwegian silicon carbide smelter workers. Occup Environ Med. 2002 May;59(5):345-7.
  32. Fosbroke D, Kisner S, Myers J. Working lifetime risk of occupational fatal injury. Am J Ind Med. 1997 Apr;31(4):459-67.
  33. Loomis D, Richardson D, Wolf S, et al. Fatal occupational injuries in a southern state. Am J Epidemiol. 1997 Jun 15;145(12):1089-99.
  34. Barreto S, Swerdlow A, Smith P, et al. Mortality from injuries and other causes in a cohort of 21,800 Brazilian steel workers. Occup Environ Med. 1996 May;53(5):343-50.
  35. Maki N, Martikainen P. Socioeconomic differences in suicide mortality by sex in Finland in 1971-2000: a register-based study of trends, levels, and life expectancy differences. Scand J Public Health. 2007;35(4):387-95.
  36. Jaremin B. Work-site casualties and environmental risk assessment on Polish vessels in the years 1960-1999. Intl Marit Health 2005;56(1-4):17-27.
  37. Najman J, Williams G, Clavarino A. Correcting for numerator/denominator bias when assessing changing inequalities in occupational class mortality, Australia 1981-2002. Bull World Health Organ. 2006 Mar;84(3):198-203.
  38. Biddle E, Kisner S. Denominator effects on traumatic occupational fatality incidence rates. Stat Bull Metro Insur Co. 1998 Jan-Mar;79(1):28-36.
  39. Blakely T, Fawcett J. Bias measuring mortality gradients by occupational class in New Zealand. N Z Med J. 2005 Jan 28;118(1208):U1253.
  40. Collins J, Riordan S. Mortality surveillance and occupational hazards: the Solutia mortality experience 1980-94. Occup Environ Med. 2000 Oct;57(10):710-7.
  41. Fujioka M, Mori H, Yoshinaga K, Kaneko J. Comparison of occupational mortality between the Nordic countries and Japan, with analysis by age group in Japan, using micro-data and the Statistical Pattern Analysis (SPA) method. In: International Labour Organisation, editor. Japan, 2009.